[Fix] (MG_Impl): report a buffer texture as the wrong object, not the wrong token

glGetTextureParameter* resolve the texture by name and then hand the work to the
target-based getter, which validates the target it was given. For a buffer texture that
is GL_TEXTURE_BUFFER, and the target form correctly calls that an unaccepted token -
INVALID_ENUM.

By name there is no token to blame. The application named an object that carries none of
the sampler or level state the query reports, which is INVALID_OPERATION (GL 4.6 core
8.11). The four by-name getters check the resolved object before delegating, so the error
describes what the caller actually got wrong.

Fixes direct_state_access.textures_parameter_errors on both backends, taking the group to
74.93% on Espryt and 73.32% on Magma.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DnW6wQYv2Pvrfisqak8UJu

.gitignore

@@ -25,3 +25,5 @@ MobileGL/MG*/cmake-build*
 /android-plugin/app/src/trace/jniLibs
 /android-plugin/local.properties
 tools/trace_replay/work/
+__pycache__/
+*.py[cod]

CMakeLists.txt

@@ -191,12 +191,10 @@ set(SOURCE_FILES
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/DecoratePositionInvariantPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/LowerDrawParametersPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/RebaseInstanceIndexPass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/NormalizeRectCoordinatesPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/StripUboMemberRelaxedPrecisionPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/StripNoPerspectivePass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/EmulateNoPerspectivePass.cpp
-    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/FoldConstOffsetFor1DFetchPass.cpp
-    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/LowerClipDistanceForEsslPass.cpp
-    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/DefeatConstStructArrayLutPass.cpp
 
     MobileGL/MG_Util/BackendLoaders/OpenGL/Loader.cpp
     MobileGL/MG_Util/BackendLoaders/Vulkan/Loader.cpp
@@ -458,8 +456,21 @@ if (ANDROID)
 endif()
 
 if (APPLE AND NOT MOBILEGL_IOS)
+    # MobileGL statically embeds glslang, SPIRV-Tools, and SPIRV-Cross. When
+    # this dylib is injected with DYLD_INSERT_LIBRARIES, exporting those C++
+    # symbols interposes incompatible copies embedded by host libraries such
+    # as shaderc. Keep only the public GL/EGL/CGL loader surface globally
+    # visible; GetProcAddress can still return pointers to hidden internals.
+    set(MOBILEGL_MACOS_EXPORTED_SYMBOLS
+        "${CMAKE_CURRENT_SOURCE_DIR}/MobileGL/MG_Impl/DyldInterpose/ExportedSymbols.txt")
+    target_link_options(${CMAKE_PROJECT_NAME} PRIVATE
+        "LINKER:-exported_symbols_list,${MOBILEGL_MACOS_EXPORTED_SYMBOLS}")
+    set_property(TARGET ${CMAKE_PROJECT_NAME} APPEND PROPERTY
+        LINK_DEPENDS "${MOBILEGL_MACOS_EXPORTED_SYMBOLS}")
+
     target_link_libraries(${CMAKE_PROJECT_NAME} PUBLIC
         "-framework Cocoa"
+        "-framework CoreVideo"
         "-framework QuartzCore"
         "-framework Foundation"
         "-framework OpenGL"
@@ -467,6 +478,7 @@ if (APPLE AND NOT MOBILEGL_IOS)
     if(TARGET ${CMAKE_PROJECT_NAME}_s)
         target_link_libraries(${CMAKE_PROJECT_NAME}_s PUBLIC
             "-framework Cocoa"
+            "-framework CoreVideo"
             "-framework QuartzCore"
             "-framework Foundation"
             "-framework OpenGL"

MobileGL/Config.h

@@ -14,7 +14,7 @@ namespace MobileGL::MG_Config {
     inline const String ProjectName = "MobileGL";
     inline const String CoreName = "MobileGL Core";
     inline const String CoreVendor = "MobileGL-Dev (BZLZHH, Swung0x48, Tungsten)";
-    inline const Version CoreVersion = {26, 7, 0, "-dev", VersionType::Development};
+    inline const Version CoreVersion = {26, 8, 0, "-dev", VersionType::Development};
     inline const VersionStringFormatAttrib DefaultVersionStringFormatAttrib = {2, 2, 0, true, true};
     inline const Uint64 CacheVersion = 0;
 
@@ -80,11 +80,6 @@ namespace MobileGL::MG_Config {
         // rewrites the recognized workgroup prefix-scan template on Qualcomm devices with
         // subgroups wider than 32 lanes (see ShaderSourceProcessor's quirk registry).
         QuirkOverride SubgroupPrefixScanQuirk = QuirkOverride::Auto;
-        // MOBILEGL_QUIRK_CLIP_DISTANCE: overrides the DirectGLES quirk that lowers
-        // gl_ClipDistance for Adreno's ESSL compiler (shadow Private arrays with
-        // constant-index builtin flushes, dynamic-index gl_in copy loop, redeclaration
-        // strip, and const struct-array LUT splitting). Auto detects Qualcomm.
-        QuirkOverride ClipDistanceQuirk = QuirkOverride::Auto;
         // MOBILEGL_MAGMA_DISABLE_BLENDED_DEPTH_WRITE: overrides the DirectVulkan quirk that
         // strips depth writes from accumulation-blended pipelines (MIN/MAX or additive
         // ONE+ONE - the multi-pass depth-equality signature) on drivers without

MobileGL/ConfigLoader.cpp

@@ -135,7 +135,6 @@ namespace MobileGL::MG_ConfigLoader {
         features.DisableUboRing = QueryEnvFlag("MOBILEGL_DISABLE_UBO_RING");
         features.RelaxedSemantics = QueryEnvFlag("MOBILEGL_RELAXED_SEMANTICS");
         features.SubgroupPrefixScanQuirk = QueryEnvQuirkOverride("MOBILEGL_QUIRK_SUBGROUP_PREFIX_SCAN");
-        features.ClipDistanceQuirk = QueryEnvQuirkOverride("MOBILEGL_QUIRK_CLIP_DISTANCE");
         features.MagmaDisableBlendedDepthWriteQuirk =
             QueryEnvQuirkOverride("MOBILEGL_MAGMA_DISABLE_BLENDED_DEPTH_WRITE");
         features.DisableRobustBufferAccess = QueryEnvFlag("MOBILEGL_DISABLE_ROBUST_BUFFER_ACCESS");

MobileGL/Init.cpp

@@ -14,6 +14,7 @@
 #include <MG_State/EGLState/Core.h>
 #include <MG_Impl/GLImpl/Texture/ProxyTexture.h>
 #include <MG_Impl/GLImpl/Framebuffer/GL_Framebuffer.h>
+#include <MG_Impl/GLImpl/Sync/GL_Sync.h>
 
 #include <atomic>
 #include <mutex>
@@ -37,6 +38,12 @@ namespace MobileGL {
                 MGLOG_I("MobileGL closing...");
             }
             glslang::FinalizeProcess();
+            // GL syncs die with their contexts, and every context is gone by the
+            // time full teardown runs: drain the live-sync registry while the
+            // backend function table can still release the backend handles (and
+            // before a re-initialized library could pair them with the wrong
+            // backend's DeleteSync).
+            MG_Impl::GLImpl::DestroyAllSyncObjects();
             MG_Backend::pActiveBackendObject.reset();
             MG_State::pGLContext.reset();
             MG_State::pEGLContext.reset();
@@ -100,9 +107,11 @@ namespace MobileGL {
     // (EGL/WGL/CGL): initialization happens lazily on the first entry point
     // via EnsureInitialized(), and full teardown happens deterministically
     // when the last EGL display is terminated with nothing current (EGLImpl
-    // calls Destroy()). There is intentionally no static constructor, no
-    // static destructor, and no DllMain: the global singletons use
-    // leak-at-exit storage (see GlobalObjects.cpp), so a process that exits
+    // calls Destroy()). There is intentionally no backend-initializing static
+    // constructor, no static destructor, and no DllMain: the global singletons
+    // use leak-at-exit storage (see GlobalObjects.cpp), so a process that exits
     // without eglTerminate simply leaks them to the OS instead of running
-    // backend destructors during static teardown.
+    // backend destructors during static teardown. macOS has a lightweight
+    // dyld constructor that installs NSOpenGL dispatch hooks only; full backend
+    // initialization still enters here from the first hooked CGL context.
 } // namespace MobileGL

MobileGL/Init.h

@@ -13,9 +13,10 @@ namespace MobileGL {
     void Initialize();
     // Thread-safe, idempotent, and re-entrant wrapper around Initialize().
     // Host layers (EGL/WGL/CGL entry points) call this lazily on first use so
-    // MobileGL's lifecycle never depends on ELF/DLL static constructors, and
-    // so a fresh init can follow a full Destroy() (e.g. after the last
-    // eglTerminate).
+    // full backend initialization never depends on ELF/DLL static constructors,
+    // and so a fresh init can follow a full Destroy() (e.g. after the last
+    // eglTerminate). The macOS dyld bootstrap installs only lightweight
+    // NSOpenGL method hooks.
     void EnsureInitialized();
     void Destroy();
 

MobileGL/MG_Backend/BackendObject.h

@@ -220,6 +220,31 @@ namespace MobileGL {
             // and leave the query readable later.
             Bool (*GetQueryResult64)(BackendQueryHandle query, Bool wait, Uint64* outNanoseconds);
             void (*DeleteBackendQuery)(BackendQueryHandle query);
+            // GL_SAMPLES_PASSED occlusion queries (optional; null = unsupported,
+            // the frontend then rejects the target). Results/deletion flow through
+            // GetQueryResult64 / DeleteBackendQuery like timer queries.
+            BackendQueryHandle (*BeginOcclusionQuery)();
+            void (*EndOcclusionQuery)(BackendQueryHandle query);
+            // Transform feedback primitive queries backed by real GPU query pools
+            // (optional; null = frontend falls back to CPU accounting).
+            BackendQueryHandle (*BeginXfbPrimitivesQuery)(Bool generated);
+            void (*EndXfbPrimitivesQuery)(BackendQueryHandle query);
+            // Transform feedback capture spans, for backends whose own GL/ES driver
+            // performs the capture (DirectGLES). Both optional; null means the backend
+            // drives capture from its draw recording instead (DirectVulkan). End is
+            // called while the frontend capture state is still active, so the backend
+            // can still see the capture program and buffer bindings.
+            // GL_PATCH_VERTICES; ES 3.2 spells it the same way.
+            void (*PatchParameteri)(GLenum pname, GLint value);
+            void (*BeginTransformFeedback)(GLenum primitiveMode);
+            void (*EndTransformFeedback)();
+            // ARB_transform_feedback2. A backend that leaves these null keeps the single
+            // implicit capture span the frontend has always modelled; the frontend state
+            // (paused flag, per-object bindings) is tracked either way.
+            void (*PauseTransformFeedback)();
+            void (*ResumeTransformFeedback)();
+            void (*BindTransformFeedback)(GLuint name);
+            void (*DeleteTransformFeedback)(GLuint name);
             Int64 (*GetGpuTimestampNs)(); // glGetInteger64v(GL_TIMESTAMP); 0 if unsupported
         };
         struct GlobalBackendFunctionsTable {
@@ -272,6 +297,13 @@ namespace MobileGL {
             Int MaxIntegerSamples = 1;
             Int MaxSamples = 1;
             Int MaxSampleMaskWords = 1;
+            // Tessellation limits; defaults are the GL 4.0 core minimums.
+            Int MaxPatchVertices = 32;
+            Int MaxTessGenLevel = 64;
+            // GL_MIN/MAX_PROGRAM_TEXTURE_GATHER_OFFSET. Defaults are the GL 4.0 core
+            // minimums, which every ES 3.1 driver also guarantees.
+            Int MinProgramTextureGatherOffset = -8;
+            Int MaxProgramTextureGatherOffset = 7;
             Int MaxTextureImageUnits = 32;
             Int MaxVertexTextureImageUnits = 32;
             Int MaxComputeTextureImageUnits = 32;
@@ -283,6 +315,8 @@ namespace MobileGL {
             Int MaxComputeWorkGroupInvocations = 128;
             Int MaxShaderStorageBufferBindings = 8;
             Int MaxTextureBufferSize = 65536;
+            // GL_TEXTURE_BUFFER_OFFSET_ALIGNMENT; 1 means the offset is unconstrained.
+            Int TextureBufferOffsetAlignment = 1;
             Int MaxUniformBufferBindings = 24;
             Int MaxUniformBlockSize = 16384;
             Int MaxImageUnits = 8;
@@ -300,7 +334,21 @@ namespace MobileGL {
             Float ViewportBoundsRangeMin = 0.0f;
             Float ViewportBoundsRangeMax = 0.0f;
             Int ViewportSubpixelBits = 0;
+            // GL 4.x fragment-interpolation offset limits. These defaults are the
+            // core minimums and are replaced by live GLES/Vulkan device limits.
+            Float MinFragmentInterpolationOffset = -0.5f;
+            // For four fractional bits the greatest required legal offset is
+            // 0.5 - 2^-4 = 0.4375 (GL 4.6 table 23.70).
+            Float MaxFragmentInterpolationOffset = 0.4375f;
+            Int FragmentInterpolationOffsetBits = 4;
             Bool SupportsWideLines = false;
+            // Whether a framebuffer whose depth and stencil attachments are distinct
+            // images can be rendered to. GL only requires support when both refer to the
+            // same image and lets an implementation answer GL_FRAMEBUFFER_UNSUPPORTED
+            // otherwise, which is what DirectVulkan (one combined attachment) and the
+            // real ES drivers behind DirectGLES both do. Defaults to true so a backend
+            // that never sets it keeps the permissive behaviour.
+            Bool SupportsDistinctDepthStencilAttachments = true;
             SizeT MaxShaderStorageBlockSize = 128 * 1024 * 1024;
             Uint32 SubgroupSize = 0;
             Uint32 SubgroupSupportedStages = 0;

MobileGL/MG_Backend/DirectGLES/BackendObject_DirectGLES.cpp

@@ -20,6 +20,7 @@
 #include <MG_Util/Texture/TextureFormatProcessor.h>
 #include <Config.h>
 #include <algorithm>
+#include <cmath>
 #include <format>
 
 namespace MobileGL::MG_Backend::DirectGLES {
@@ -209,6 +210,12 @@ namespace MobileGL::MG_Backend::DirectGLES {
             if (options & PixelFormatNormalizeOptionBit::NoDepthComponent32) {
                 reasons.push_back("GL_DEPTH_COMPONENT32 native probe failed on OpenGL ES");
             }
+            if (options & PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget) {
+                reasons.push_back("no three-channel multisample storage format on OpenGL ES");
+            }
+            if (options & PixelFormatNormalizeOptionBit::NoSnorm16RenderTarget) {
+                reasons.push_back("EXT_render_snorm not supported");
+            }
 
             String reason;
             for (SizeT i = 0; i < reasons.size(); ++i) {
@@ -356,6 +363,42 @@ namespace MobileGL::MG_Backend::DirectGLES {
             return complete;
         }
 
+        // Whether the driver renders to a framebuffer whose depth and stencil come from
+        // two different renderbuffers. GL only requires support when both attachments are
+        // the same image, and ES drivers commonly answer GL_FRAMEBUFFER_UNSUPPORTED here;
+        // reporting COMPLETE from the frontend and then rendering into a framebuffer the
+        // driver refuses leaves the results silently empty.
+        Bool ProbeDistinctDepthStencilAttachments(const MG_External::GLESFunctionsTable& gl) {
+            if (!gl.glGenFramebuffers || !gl.glBindFramebuffer || !gl.glFramebufferRenderbuffer ||
+                !gl.glCheckFramebufferStatus || !gl.glDeleteFramebuffers || !gl.glGenRenderbuffers ||
+                !gl.glBindRenderbuffer || !gl.glRenderbufferStorage || !gl.glDeleteRenderbuffers) {
+                return true;
+            }
+
+            GLint prevFramebuffer = 0, prevRenderbuffer = 0;
+            gl.glGetIntegerv(GL_FRAMEBUFFER_BINDING, &prevFramebuffer);
+            gl.glGetIntegerv(GL_RENDERBUFFER_BINDING, &prevRenderbuffer);
+
+            GLuint framebuffer = 0;
+            GLuint renderbuffers[2] = {0, 0};
+            gl.glGenFramebuffers(1, &framebuffer);
+            gl.glGenRenderbuffers(2, renderbuffers);
+            gl.glBindRenderbuffer(GL_RENDERBUFFER, renderbuffers[0]);
+            gl.glRenderbufferStorage(GL_RENDERBUFFER, GL_DEPTH_COMPONENT16, 4, 4);
+            gl.glBindRenderbuffer(GL_RENDERBUFFER, renderbuffers[1]);
+            gl.glRenderbufferStorage(GL_RENDERBUFFER, GL_STENCIL_INDEX8, 4, 4);
+            gl.glBindFramebuffer(GL_FRAMEBUFFER, framebuffer);
+            gl.glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_RENDERBUFFER, renderbuffers[0]);
+            gl.glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_RENDERBUFFER, renderbuffers[1]);
+            const Bool supported = gl.glCheckFramebufferStatus(GL_FRAMEBUFFER) == GL_FRAMEBUFFER_COMPLETE;
+
+            gl.glBindFramebuffer(GL_FRAMEBUFFER, static_cast<GLuint>(prevFramebuffer));
+            gl.glBindRenderbuffer(GL_RENDERBUFFER, static_cast<GLuint>(prevRenderbuffer));
+            gl.glDeleteFramebuffers(1, &framebuffer);
+            gl.glDeleteRenderbuffers(2, renderbuffers);
+            return supported;
+        }
+
         Bool ProbeFramebufferCompletenessForRenderbuffer(const MG_External::GLESFunctionsTable& gl,
                                                         GLuint renderbuffer,
                                                         TextureInternalFormat format) {
@@ -519,20 +562,50 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 }
 
                 const GLESProbeFormatInfo nativeInfo = BuildNativeProbeFormatInfo(requestedInternalFormat);
-                GLESProbeFormatInfo fallbackInfo;
-                const Bool hasForcedFallback =
-                    BuildFallbackProbeFormatInfo(requestedInternalFormat, forcedOptions, true, fallbackInfo);
-                if (!hasForcedFallback) {
-                    BuildFallbackProbeFormatInfo(requestedInternalFormat, driverOptions, false, fallbackInfo);
+                GLESProbeFormatInfo outerFallbackInfo;
+                const Bool outerHasForcedFallback =
+                    BuildFallbackProbeFormatInfo(requestedInternalFormat, forcedOptions, true, outerFallbackInfo);
+                if (!outerHasForcedFallback) {
+                    BuildFallbackProbeFormatInfo(requestedInternalFormat, driverOptions, false, outerFallbackInfo);
                 }
 
                 for (SizeT targetIndex = 0; targetIndex < kFormatCapabilityTextureTargetCount; ++targetIndex) {
                     const auto target = static_cast<TextureTarget>(targetIndex);
+                    // A multisample texture can only ever be rendered into, so its storage format
+                    // has to stay colour-renderable; the ordinary fallback for a three-channel
+                    // format is a three-channel one, which ES accepts as a texture but rejects as
+                    // multisample storage. Recompute the fallback per target so those formats get
+                    // widened here and nowhere else.
+                    Flags<PixelFormatNormalizeOptionBit> targetOptions;
+                    if (IsGLESProbeMultisampleTarget(target)) {
+                        targetOptions |= PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget;
+                        if (!capabilities.SupportsRenderSnorm || !capabilities.SupportsNorm16Texture) {
+                            targetOptions |= PixelFormatNormalizeOptionBit::NoSnorm16RenderTarget;
+                        }
+                    }
+                    GLESProbeFormatInfo fallbackInfo = outerFallbackInfo;
+                    Bool hasForcedFallback = outerHasForcedFallback;
+                    if (targetOptions) {
+                        hasForcedFallback = BuildFallbackProbeFormatInfo(
+                            requestedInternalFormat, forcedOptions | targetOptions, true, fallbackInfo);
+                        if (!hasForcedFallback) {
+                            BuildFallbackProbeFormatInfo(requestedInternalFormat, driverOptions | targetOptions,
+                                                         false, fallbackInfo);
+                        }
+                    }
+
+                    // 1D, 1D-array and rectangle textures live on an ES target (see
+                    // TextureImpl::MapToBackendTextureTarget), so they have to be probed there too -
+                    // probing the desktop-only target itself always failed, which left those slots
+                    // of the cache empty and stopped any fallback format from being selected for
+                    // them (a GL_DEPTH_COMPONENT32 1D texture then got no storage at all).
+                    const TextureTarget probeTarget = TextureImpl::MapToBackendTextureTarget(target);
+
                     Bool shouldProbeFallback = hasForcedFallback;
                     if (!hasForcedFallback) {
                         Bool nativeRenderable = false;
                         const Bool nativeCreated =
-                            ProbeTexture(gl, target, nativeInfo.InternalFormat, nativeInfo.ImageFormat,
+                            ProbeTexture(gl, probeTarget, nativeInfo.InternalFormat, nativeInfo.ImageFormat,
                                          nativeInfo.ImageType, logicalFormat, &nativeRenderable);
                         if (nativeCreated) {
                             AddFullFormatCaps(cache, targetIndex, formatIndex,
@@ -547,7 +620,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
                     if (shouldProbeFallback && fallbackInfo.InternalFormat != GL_UNKNOWN_MGL) {
                         Bool fallbackRenderable = false;
                         const Bool fallbackCreated =
-                            ProbeTexture(gl, target, fallbackInfo.InternalFormat, fallbackInfo.ImageFormat,
+                            ProbeTexture(gl, probeTarget, fallbackInfo.InternalFormat, fallbackInfo.ImageFormat,
                                          fallbackInfo.ImageType, logicalFormat, &fallbackRenderable);
                         if (fallbackCreated) {
                             if (AddCaveatFormatCaps(cache, targetIndex, formatIndex,
@@ -563,8 +636,8 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 }
 
                 const SizeT renderbufferTargetIndex = GetRenderbufferFormatCapabilityTargetIndex();
-                Bool shouldProbeFallbackRenderbuffer = hasForcedFallback;
-                if (!hasForcedFallback) {
+                Bool shouldProbeFallbackRenderbuffer = outerHasForcedFallback;
+                if (!outerHasForcedFallback) {
                     const Bool nativeRenderbufferComplete =
                         ProbeRenderbuffer(gl, nativeInfo.InternalFormat, logicalFormat, false, 1);
                     if (nativeRenderbufferComplete) {
@@ -578,16 +651,16 @@ namespace MobileGL::MG_Backend::DirectGLES {
                         shouldProbeFallbackRenderbuffer = true;
                     }
                 }
-                if (shouldProbeFallbackRenderbuffer && fallbackInfo.InternalFormat != GL_UNKNOWN_MGL &&
-                    ProbeRenderbuffer(gl, fallbackInfo.InternalFormat, logicalFormat, false, 1)) {
+                if (shouldProbeFallbackRenderbuffer && outerFallbackInfo.InternalFormat != GL_UNKNOWN_MGL &&
+                    ProbeRenderbuffer(gl, outerFallbackInfo.InternalFormat, logicalFormat, false, 1)) {
                     if (AddCaveatFormatCaps(cache, renderbufferTargetIndex, formatIndex,
                                             GetRenderbufferFeatureCaps(logicalFormat))) {
-                        LogGLESFormatCaveat(logicalFormat, renderbufferTargetIndex, fallbackInfo);
+                        LogGLESFormatCaveat(logicalFormat, renderbufferTargetIndex, outerFallbackInfo);
                     }
                     const Int maxSamples =
-                        GetGLESFormatMaxSamples(capabilities, logicalFormat, fallbackInfo.ImageFormat);
+                        GetGLESFormatMaxSamples(capabilities, logicalFormat, outerFallbackInfo.ImageFormat);
                     cache.SampleCounts[renderbufferTargetIndex][formatIndex] =
-                        ProbeRenderbufferSampleCounts(gl, fallbackInfo.InternalFormat, logicalFormat, maxSamples);
+                        ProbeRenderbufferSampleCounts(gl, outerFallbackInfo.InternalFormat, logicalFormat, maxSamples);
                 }
             }
         }
@@ -603,7 +676,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 .ExtraVendor = Nullopt,              // Extra vendor
                 .RendererGLInfo =
                     {
-                        .TargetGLVersion = {3, 3, 0},   // Target OpenGL Version
+                        .TargetGLVersion = {3, 3, 0},   // GL target version
                         .TargetGLSLVersion = {4, 6, 0}, // Target Shading Language Version
                         // Baseline advertisement (no timer queries / anisotropy yet); reconciled
                         // once the ES capabilities exist, see UpdateAdvertisedCapabilityExtensions.
@@ -831,7 +904,17 @@ namespace MobileGL::MG_Backend::DirectGLES {
                                           E_GL_ARB_multi_draw_indirect, E_GL_ARB_indirect_parameters,
                                           E_GL_ARB_shader_draw_parameters, E_GL_ARB_gpu_shader5, E_GL_ARB_multi_bind,
                                           E_GL_ARB_shading_language_420pack, E_GL_ARB_vertex_attrib_binding,
-                                          E_GL_ARB_shader_image_size};
+                                          // Both are core from GL 3.2/3.3 on and implemented here for
+                                          // every advertised version, but an app targeting 3.0/3.1
+                                          // only reaches them through the extension string - the CTS
+                                          // picks a whole different shader for draw_buffers without
+                                          // explicit_attrib_location. DirectVulkan advertises both.
+                                          E_GL_ARB_explicit_attrib_location, E_GL_ARB_texture_multisample,
+                                          E_GL_ARB_shader_image_size,
+                                          // Advertised with GL_NUM_PROGRAM_BINARY_FORMATS = 0, which the
+                                          // extension explicitly permits. It is also the only thing that
+                                          // exposes glProgramParameteri before GL 4.1.
+                                          E_GL_ARB_get_program_binary};
         // Only advertised when the device driver actually has usable timer queries
         // (GL_EXT_disjoint_timer_query plus its entry points) and the
         // MOBILEGL_DISABLE_TIMERQUERY escape hatch is off.
@@ -934,11 +1017,30 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 funcsTable.GL.BeginTimeElapsedQuery = BeginTimeElapsedQuery;
                 funcsTable.GL.EndTimeElapsedQuery = EndTimeElapsedQuery;
                 funcsTable.GL.QueryCounterTimestamp = QueryCounterTimestamp;
-                funcsTable.GL.IsQueryResultAvailable = IsQueryResultAvailable;
-                funcsTable.GL.GetQueryResult64 = GetQueryResult64;
-                funcsTable.GL.DeleteBackendQuery = DeleteBackendQuery;
                 funcsTable.GL.GetGpuTimestampNs = GetGpuTimestampNs;
             }
+            // Occlusion queries are core ES3 (independent of MOBILEGL_DISABLE_TIMERQUERY)
+            // and share the handle-based result/delete entries, which must exist even
+            // when the timer-query group above is disabled.
+            funcsTable.GL.BeginOcclusionQuery = BeginOcclusionQuery;
+            funcsTable.GL.EndOcclusionQuery = EndOcclusionQuery;
+            // Real driver primitive counters: the frontend's CPU accounting cannot see a
+            // geometry shader's amplification.
+            funcsTable.GL.BeginXfbPrimitivesQuery = BeginXfbPrimitivesQuery;
+            funcsTable.GL.EndXfbPrimitivesQuery = EndXfbPrimitivesQuery;
+            funcsTable.GL.IsQueryResultAvailable = IsQueryResultAvailable;
+            funcsTable.GL.GetQueryResult64 = GetQueryResult64;
+            funcsTable.GL.DeleteBackendQuery = DeleteBackendQuery;
+            // Transform feedback is captured by the real ES driver rather than
+            // reconstructed from the draw recording, so the frontend has to hand the
+            // span boundaries over.
+            funcsTable.GL.PatchParameteri = DirectGLES::PatchParameteri;
+            funcsTable.GL.BeginTransformFeedback = XfbImpl::BeginTransformFeedback;
+            funcsTable.GL.EndTransformFeedback = XfbImpl::EndTransformFeedback;
+            funcsTable.GL.PauseTransformFeedback = XfbImpl::PauseTransformFeedback;
+            funcsTable.GL.ResumeTransformFeedback = XfbImpl::ResumeTransformFeedback;
+            funcsTable.GL.BindTransformFeedback = XfbImpl::BindTransformFeedback;
+            funcsTable.GL.DeleteTransformFeedback = XfbImpl::DeleteTransformFeedback;
             funcsTableInitialized = true;
         }
         return funcsTable;
@@ -979,6 +1081,10 @@ namespace MobileGL::MG_Backend::DirectGLES {
         m_dynamicParameters.MaxIntegerSamples = m_GLESCapabilities.MaxIntegerSamples;
         m_dynamicParameters.MaxSamples = m_GLESCapabilities.MaxSamples;
         m_dynamicParameters.MaxSampleMaskWords = m_GLESCapabilities.MaxSampleMaskWords;
+        m_dynamicParameters.MaxPatchVertices = m_GLESCapabilities.MaxPatchVertices;
+        m_dynamicParameters.MaxTessGenLevel = m_GLESCapabilities.MaxTessGenLevel;
+        m_dynamicParameters.MinProgramTextureGatherOffset = m_GLESCapabilities.MinProgramTextureGatherOffset;
+        m_dynamicParameters.MaxProgramTextureGatherOffset = m_GLESCapabilities.MaxProgramTextureGatherOffset;
         // Clamp the advertised sampler limits the same way the DirectVulkan backend does: per-stage
         // GL_MAX_TEXTURE_IMAGE_UNITS must never exceed host-side fixed arrays sized off it (e.g.
         // Minecraft's 128-entry Blaze3D GlStateManager.TEXTURES[], iterated by Iris), and the combined
@@ -1006,6 +1112,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
         m_dynamicParameters.MaxComputeWorkGroupInvocations = m_GLESCapabilities.MaxComputeWorkGroupInvocations;
         m_dynamicParameters.MaxShaderStorageBufferBindings = m_GLESCapabilities.MaxShaderStorageBufferBindings;
         m_dynamicParameters.MaxTextureBufferSize = m_GLESCapabilities.MaxTextureBufferSize;
+        m_dynamicParameters.TextureBufferOffsetAlignment = m_GLESCapabilities.TextureBufferOffsetAlignment;
         m_dynamicParameters.MaxUniformBufferBindings = m_GLESCapabilities.MaxUniformBufferBindings;
         m_dynamicParameters.MaxUniformBlockSize = m_GLESCapabilities.MaxUniformBlockSize;
         const Int maxSupportedTextureUnits =
@@ -1025,6 +1132,8 @@ namespace MobileGL::MG_Backend::DirectGLES {
             clampStageImageUniforms(m_GLESCapabilities.MaxFragmentImageUniforms);
         m_dynamicParameters.MaxComputeImageUniforms =
             clampStageImageUniforms(m_GLESCapabilities.MaxComputeImageUniforms);
+        m_dynamicParameters.SupportsDistinctDepthStencilAttachments =
+            ProbeDistinctDepthStencilAttachments(DirectGLES::g_GLESFuncs);
         m_dynamicParameters.MaxDrawBuffers = m_GLESCapabilities.MaxDrawBuffers;
         m_dynamicParameters.MaxColorAttachments = m_GLESCapabilities.MaxColorAttachments;
         m_dynamicParameters.MaxClipDistances = m_GLESCapabilities.MaxClipDistances;
@@ -1034,6 +1143,24 @@ namespace MobileGL::MG_Backend::DirectGLES {
         m_dynamicParameters.ViewportBoundsRangeMin = m_GLESCapabilities.ViewportBoundsRangeMin;
         m_dynamicParameters.ViewportBoundsRangeMax = m_GLESCapabilities.ViewportBoundsRangeMax;
         m_dynamicParameters.ViewportSubpixelBits = m_GLESCapabilities.ViewportSubpixelBits;
+        m_dynamicParameters.MinFragmentInterpolationOffset =
+            std::isfinite(m_GLESCapabilities.MinFragmentInterpolationOffset) &&
+                    m_GLESCapabilities.MinFragmentInterpolationOffset <= -0.5f
+                ? m_GLESCapabilities.MinFragmentInterpolationOffset
+                : -0.5f;
+        m_dynamicParameters.MaxFragmentInterpolationOffset = 0.4375f;
+        m_dynamicParameters.FragmentInterpolationOffsetBits = 4;
+        if (m_GLESCapabilities.FragmentInterpolationOffsetBits >= 4 &&
+            std::isfinite(m_GLESCapabilities.MaxFragmentInterpolationOffset)) {
+            const Float requiredMaxOffset =
+                0.5f - std::ldexp(1.0f, -m_GLESCapabilities.FragmentInterpolationOffsetBits);
+            if (m_GLESCapabilities.MaxFragmentInterpolationOffset >= requiredMaxOffset) {
+                m_dynamicParameters.MaxFragmentInterpolationOffset =
+                    m_GLESCapabilities.MaxFragmentInterpolationOffset;
+                m_dynamicParameters.FragmentInterpolationOffsetBits =
+                    m_GLESCapabilities.FragmentInterpolationOffsetBits;
+            }
+        }
         m_dynamicParameters.SupportsWideLines =
             m_GLESCapabilities.AliasedLineWidthRangeMax > 1.0f || m_GLESCapabilities.SmoothLineWidthRangeMax > 1.0f;
 

MobileGL/MG_Backend/DirectGLES/DirectGLES.h

@@ -130,6 +130,16 @@ namespace MobileGL::MG_Backend::DirectGLES {
     BackendQueryHandle BeginTimeElapsedQuery();
     void EndTimeElapsedQuery(BackendQueryHandle query);
     BackendQueryHandle QueryCounterTimestamp();
+    // GL_ANY_SAMPLES_PASSED(_CONSERVATIVE) occlusion queries: core ES3, independent of
+    // GL_EXT_disjoint_timer_query and of MOBILEGL_DISABLE_TIMERQUERY. Results/deletion
+    // flow through GetQueryResult64/DeleteBackendQuery like the timer queries above.
+    BackendQueryHandle BeginOcclusionQuery();
+    void EndOcclusionQuery(BackendQueryHandle query);
+    // GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN / GL_PRIMITIVES_GENERATED, also core ES
+    // (GL_PRIMITIVES_GENERATED from ES 3.2 on). Null when the target is unavailable, in
+    // which case the frontend falls back to counting primitives from the draw calls.
+    BackendQueryHandle BeginXfbPrimitivesQuery(Bool generated);
+    void EndXfbPrimitivesQuery(BackendQueryHandle query);
     Bool IsQueryResultAvailable(BackendQueryHandle query);
     // Returns true when a final value landed in *outNanoseconds (a zero for
     // null or stale-generation handles IS final: the frontend may cache it
@@ -154,6 +164,24 @@ namespace MobileGL::MG_Backend::DirectGLES {
     void SetGLESCapabilities(const MG_External::GLESCapabilities& capabilities);
     void DestroyEGLContext();
 
+    // Transform feedback capture spans, performed by the real ES driver. The
+    // capture set is declared on the backend program at link time; the driver-side
+    // begin is deferred to the first draw of the span (ES needs the capturing
+    // program current and the capture buffers bound), and the end also mirrors the
+    // captured bytes back into the frontend buffer shadows.
+    void PatchParameteri(GLenum pname, GLint value);
+
+    namespace XfbImpl {
+        Bool AreTransformFeedbacksSupported();
+        void BeginTransformFeedback(GLenum primitiveMode);
+        void EndTransformFeedback();
+        void PauseTransformFeedback();
+        void ResumeTransformFeedback();
+        void BindTransformFeedback(GLuint name);
+        void DeleteTransformFeedback(GLuint name);
+        void OnBackendContextDestroyed();
+    } // namespace XfbImpl
+
     extern MG_External::EGLFunctionsTable g_EGLFuncs;
     extern MG_External::GLESFunctionsTable g_GLESFuncs;
     extern MG_External::GLESCapabilities g_GLESCapabilities;

MobileGL/MG_Backend/DirectGLES/Managers.cpp

@@ -630,6 +630,32 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 resource->syncedChangeSerial = bufferObject.GetChangeSerial();
             }
 
+            // A shader wrote this buffer through a storage/atomic-counter binding, so the ES
+            // driver's copy is ahead of the frontend shadow. Pull the whole thing back so
+            // MapBuffer/GetBufferSubData/CopyBufferSubData see the real results.
+            void Ops_ReadbackFromGpu(BufferObject& bufferObject) {
+                auto* resource = ResourceOf(bufferObject);
+                if (!resource || resource->id == 0 || !resource->storageInitialized) return;
+                if (resource->persistentMapped) return; // shadow already IS the GPU storage
+                if (!CanTouchGLNow() || resource->contextGeneration != g_bufferContextGeneration) return;
+                if (!g_GLESFuncs.glMapBufferRange || !g_GLESFuncs.glUnmapBuffer) return;
+                const SizeT size = std::min<SizeT>(bufferObject.GetSize(), resource->storageSize);
+                if (size == 0) return;
+
+                BindBufferId(TempBufferTarget, resource->id);
+                void* mapped = g_GLESFuncs.glMapBufferRange(TempBufferTarget, 0, static_cast<GLsizeiptr>(size),
+                                                            GL_MAP_READ_BIT);
+                if (mapped == nullptr) {
+                    MGLOG_E("Ops_ReadbackFromGpu: glMapBufferRange(read) failed for buffer %u", resource->id);
+                    return;
+                }
+                bufferObject.WritebackFromBackend({mapped, size}, 0);
+                g_GLESFuncs.glUnmapBuffer(TempBufferTarget);
+                // The shadow now matches the backend byte for byte; without this the next
+                // draw would see a newer change serial and re-upload the readback over it.
+                resource->syncedChangeSerial = bufferObject.GetChangeSerial();
+            }
+
             void Ops_OnDestroy(SharedPtr<BackendBufferResource>&& resource) {
                 if (!resource) return;
                 auto* glesResource = static_cast<GLESBufferResource*>(resource.get());
@@ -662,6 +688,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 .FlushMappedRange = Ops_FlushMappedRange,
                 .OnDestroy = Ops_OnDestroy,
                 .AcquirePersistentMap = Ops_AcquirePersistentMap,
+                .ReadbackFromGpu = Ops_ReadbackFromGpu,
             };
         } // namespace
 
@@ -1262,31 +1289,15 @@ namespace MobileGL::MG_Backend::DirectGLES {
             const auto& allAttributes = stateVAOObject->GetAllAttributes();
             for (Uint attribIndex = 0; attribIndex < allAttributes.size(); ++attribIndex) {
                 const auto& attrib = allAttributes[attribIndex];
-                const Uint32 attribBit = 1u << attribIndex;
-
-                // An enabled attrib with neither a buffer object nor a client pointer has no
-                // source; GL tolerates the state (only draws consuming it are undefined), but
-                // Adreno's ES driver memcpys the "client array" from address 0 at draw time
-                // (SIGSEGV). Keep such attribs disabled on the backend VAO and re-enable them
-                // the moment they gain a source - the mask-vs-current compare below triggers
-                // the enable even when only the Buffer/Format versions changed.
-                const Bool unsourceable = attrib.Enabled && !attrib.Buffer && attrib.Offset == 0;
-                const Bool wasForceDisabled = (m_forceDisabledAttribsMask & attribBit) != 0;
-
                 Bool needsSyncSwitch = allAttributeVersions[attribIndex].SwitchVersion !=
                                        m_syncedAttributeVersions[attribIndex].SwitchVersion;
-                if (needsSyncSwitch || unsourceable != wasForceDisabled) {
-                    if (attrib.Enabled && !unsourceable) {
+                if (needsSyncSwitch) {
+                    if (attrib.Enabled) {
                         g_GLESFuncs.glEnableVertexAttribArray(attribIndex);
                     } else {
                         g_GLESFuncs.glDisableVertexAttribArray(attribIndex);
                     }
                 }
-                if (unsourceable) {
-                    m_forceDisabledAttribsMask |= attribBit;
-                } else {
-                    m_forceDisabledAttribsMask &= ~attribBit;
-                }
 
                 Bool needsSyncFormat = allAttributeVersions[attribIndex].FormatVersion !=
                                        m_syncedAttributeVersions[attribIndex].FormatVersion;
@@ -1294,17 +1305,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
                                        m_syncedAttributeVersions[attribIndex].BufferVersion;
                 if (!needsSyncFormat && !needsSyncBuffer) continue;
 
-                if (unsourceable) continue;
-
-                // Client-side array with a non-null pointer: the pointer is uploaded and applied
-                // per draw by SyncClientSideAttributesForDrawArrays.
-                if (!attrib.Buffer) continue;
-
                 if (!BindAttributeBuffer(attrib)) {
-                    if (attrib.Enabled) {
-                        g_GLESFuncs.glDisableVertexAttribArray(attribIndex);
-                        m_forceDisabledAttribsMask |= attribBit;
-                    }
                     continue;
                 }
 
@@ -2231,7 +2232,24 @@ namespace MobileGL::MG_Backend::DirectGLES {
                             "ID: %u, buffer ID: %u, buffer size: %zu, format: %s",
                             m_backendTextureId, backendId, buffer->GetSize(),
                             MG_Util::ConvertGLEnumToString(glInternalFormat).c_str());
-                    g_GLESFuncs.glTexBuffer(GL_TEXTURE_BUFFER, glInternalFormat, backendId);
+                    // A texture that names a window of the buffer needs the range form; the
+                    // whole-buffer forms report offset 0 and the buffer's current size, which
+                    // glTexBuffer expresses more directly (and works where the range entry point
+                    // is absent).
+                    const SizeT rangeOffset = textureBufferObject->GetBufferRangeOffset();
+                    const SizeT rangeSize = textureBufferObject->GetBufferRangeSizeInBytes();
+                    if (rangeOffset == 0 && rangeSize == buffer->GetSize()) {
+                        g_GLESFuncs.glTexBuffer(GL_TEXTURE_BUFFER, glInternalFormat, backendId);
+                    } else if (g_GLESFuncs.glTexBufferRange != nullptr) {
+                        g_GLESFuncs.glTexBufferRange(GL_TEXTURE_BUFFER, glInternalFormat, backendId,
+                                                     static_cast<GLintptr>(rangeOffset),
+                                                     static_cast<GLsizeiptr>(rangeSize));
+                    } else {
+                        MGLOG_E("Texture buffer %u names a sub-range but the driver has no "
+                                "glTexBufferRange; binding the whole buffer instead",
+                                stateTextureObject->GetExternalIndex());
+                        g_GLESFuncs.glTexBuffer(GL_TEXTURE_BUFFER, glInternalFormat, backendId);
+                    }
                     DebugImpl::ErrorLopper::Loop(
                         [file = __FILE__, line = __LINE__, func = __func__, glInternalFormat, backendId](GLenum err) {
                             MGLOG_D("%s(%s:%d) glTexBuffer(format=%s, buffer=%u) ES error: %s",
@@ -2431,7 +2449,19 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 MGLOG_D("%s(%s:%d) ES error %s", func, file, line, MG_Util::ConvertGLEnumToString(err).c_str());
             });
 
-            const auto& swizzleParams = stateTextureObject->GetAllSwizzleParams();
+            // A three-channel format widened to four for a multisample target (see
+            // NormalizePixelFormat) gains an alpha channel the frontend format does not have, and
+            // whatever the draw that filled it wrote there is not what GL would report: a format
+            // without alpha reads back as 1.0. Answer the ALPHA swizzle source with ONE so the
+            // promotion stays invisible, composed with the swizzle the application asked for.
+            Vec4<TextureSwizzleParam> swizzleParams = stateTextureObject->GetAllSwizzleParams();
+            if (TextureImpl::BackendTextureFormatAddsAlpha(stateTextureObject->GetFormat(), targetInternal)) {
+                for (SizeT channel = 0; channel < 4; ++channel) {
+                    if (swizzleParams[channel] == TextureSwizzleParam::Alpha) {
+                        swizzleParams[channel] = TextureSwizzleParam::One;
+                    }
+                }
+            }
             if (swizzleParams != m_cacheSwizzleParams) {
 #define SYNC_TEX_SWIZZLE_PARAM_IF_CHANGED(func, glEnum)                                                                \
     if (m_cacheSwizzleParams.func != swizzleParams.func) {                                                             \
@@ -2705,6 +2735,31 @@ namespace MobileGL::MG_Backend::DirectGLES {
             return false;
         }
 
+        Bool IsFixedPointFallbackReadAttachment() {
+            const auto& readFBO =
+                MG_State::pGLContext->GetFramebufferBindingSlot(FramebufferTarget::Read).GetBoundObject();
+            if (!readFBO) {
+                return false;
+            }
+            const auto readBuffer = readFBO->GetReadBuffer();
+            if (readBuffer < FramebufferAttachmentType::Color0 || readBuffer > FramebufferAttachmentType::Color31) {
+                return false;
+            }
+            // Any signed-normalized attachment, not just the ones currently substituted:
+            // ES has no GL_CLAMP_READ_COLOR at all, so even a natively stored SNORM buffer
+            // hands back the negative half that desktop GL clamps away.
+            const auto& attachmentObject = readFBO->GetAttachment(readBuffer);
+            if (attachmentObject.IsTexture()) {
+                const auto& textureObject = attachmentObject.GetTexture();
+                return textureObject && IsSnormFormat(textureObject->GetFormat());
+            }
+            if (attachmentObject.IsRenderbuffer()) {
+                const auto& renderbufferObject = attachmentObject.GetRenderbuffer();
+                return renderbufferObject && IsSnormFormat(renderbufferObject->GetInternalFormat());
+            }
+            return false;
+        }
+
         void BackendFramebufferObject::SyncReadBufferToBackend(
             const SharedPtr<MG_State::GLState::FramebufferObject>& stateFBOObject) {
             if (!stateFBOObject) {
@@ -3207,6 +3262,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
 
     namespace PrgramImpl {
         Uint32 g_snormFallbackClampOutputMask = 0;
+        Uint g_fragColorBroadcastCount = 1;
         Uint32 g_unormFallbackClampOutputMask = 0;
         Uint g_lastUsedBackendProgramId = 0;
         StateBackendObjectRegistry<MG_State::GLState::ProgramObject, BackendProgramObjectImpl> g_backendProgramObjects;
@@ -3258,8 +3314,10 @@ namespace MobileGL::MG_Backend::DirectGLES {
 
             MGLOG_D("Syncing program to backend. State program ID: %u, Backend ID: %u",
                     stateProgramObject->GetExternalIndex(), m_backendProgramId);
+            m_backendProgramUsable = true;
             m_snormFallbackClampOutputMask = g_snormFallbackClampOutputMask;
             m_unormFallbackClampOutputMask = g_unormFallbackClampOutputMask;
+            m_fragColorBroadcastCount = g_fragColorBroadcastCount;
 
             // Detach all existing shaders
             GLint attachedCount = 0;
@@ -3291,19 +3349,6 @@ namespace MobileGL::MG_Backend::DirectGLES {
             }
             auto& shaderSpirvs = stateProgramObject->GetGeneratedSpirv();
 
-            // Adreno's ESSL compiler mishandles gl_ClipDistance (rejects redeclarations,
-            // miscompiles non-constant-index writes and constant-index gl_in element reads,
-            // crashes on whole-array gl_in reads) and cannot dynamically index the global
-            // const struct[] LUTs SPIRV-Cross likes to emit. Gate the workarounds to
-            // Qualcomm; MOBILEGL_QUIRK_CLIP_DISTANCE overrides the device detection.
-            const MG_Config::QuirkOverride clipDistanceQuirkOverride =
-                MG_Config::Features.ClipDistanceQuirk;
-            const Bool applyClipDistanceQuirk =
-                clipDistanceQuirkOverride == MG_Config::QuirkOverride::ForceOn ||
-                (clipDistanceQuirkOverride == MG_Config::QuirkOverride::Auto &&
-                 pActiveBackendObject &&
-                 pActiveBackendObject->GetDynamicParameters().GpuVendor == GpuVendorKind::Qualcomm);
-
             for (int index = 0; index < attachedShaders.size(); ++index) {
                 auto& shader = attachedShaders[index];
                 GLenum glShaderType = MG_Util::ConvertShaderStageToGLEnum(shader->GetShaderStage());
@@ -3326,63 +3371,6 @@ namespace MobileGL::MG_Backend::DirectGLES {
                     effectiveSpirv = &loweredSpirv;
                 }
 
-                // GL 3.3 only promises undefined *values* for out-of-bounds array indexing, but
-                // Adreno's ESSL compiler constant-folds a provably out-of-bounds local-array
-                // index into poison that corrupts the whole shader's output. Clamp every
-                // access-chain index to its declared bounds before transpiling.
-                Vector<unsigned int> clampedSpirv;
-                if (MG_Util::ShaderTranspiler::ShaderCompiler::ClampAccessChainIndicesForEssl(*effectiveSpirv,
-                                                                                              clampedSpirv) &&
-                    !clampedSpirv.empty()) {
-                    effectiveSpirv = &clampedSpirv;
-                } else {
-                    MGLOG_W("ClampAccessChainIndicesForEssl failed, continuing with unclamped SPIR-V.");
-                }
-
-                // SPIRV-Cross emulates 1D samplers as 2D for ES: it widens texelFetch coordinates
-                // to ivec2 but keeps the ConstOffset operand scalar, which is not a valid ESSL
-                // texelFetchOffset overload (Adreno rejects it). Fold the constant offset into the
-                // coordinate instead (texelFetchOffset(t,P,l,o) == texelFetch(t,P+o,l)).
-                Vector<unsigned int> foldedOffsetSpirv;
-                if (MG_Util::ShaderTranspiler::ShaderCompiler::FoldConstOffsetFor1DFetchForEssl(
-                        *effectiveSpirv, foldedOffsetSpirv) &&
-                    !foldedOffsetSpirv.empty()) {
-                    effectiveSpirv = &foldedOffsetSpirv;
-                } else {
-                    MGLOG_W("FoldConstOffsetFor1DFetchForEssl failed, continuing with unfolded SPIR-V.");
-                }
-
-                // Adreno quirk: shadow gl_ClipDistance in Private arrays so the transpiled
-                // ESSL only writes the builtin with literal constant indices (flushed before
-                // EmitVertex/return) and only reads gl_in clip distances through dynamic loop
-                // indices - the shapes this driver compiles correctly. Must run after the
-                // access-chain clamp above so the flush indices stay literal constants.
-                Vector<unsigned int> clipDistanceSpirv;
-                if (applyClipDistanceQuirk &&
-                    (glShaderType == GL_VERTEX_SHADER || glShaderType == GL_GEOMETRY_SHADER)) {
-                    if (MG_Util::ShaderTranspiler::ShaderCompiler::LowerClipDistanceForEssl(
-                            *effectiveSpirv, clipDistanceSpirv) &&
-                        !clipDistanceSpirv.empty()) {
-                        effectiveSpirv = &clipDistanceSpirv;
-                    } else {
-                        MGLOG_W("LowerClipDistanceForEssl failed, continuing with unlowered SPIR-V.");
-                    }
-                }
-
-                // Adreno quirk: split single constant-composite stores of struct arrays so
-                // SPIRV-Cross does not promote them to global const struct[] LUTs, which this
-                // driver cannot dynamically index ("Cannot offset into the structure").
-                Vector<unsigned int> structLutSpirv;
-                if (applyClipDistanceQuirk) {
-                    if (MG_Util::ShaderTranspiler::ShaderCompiler::DefeatConstStructArrayLutForEssl(
-                            *effectiveSpirv, structLutSpirv) &&
-                        !structLutSpirv.empty()) {
-                        effectiveSpirv = &structLutSpirv;
-                    } else {
-                        MGLOG_W("DefeatConstStructArrayLutForEssl failed, continuing with unsplit SPIR-V.");
-                    }
-                }
-
                 // ESSL stage-matches uniform blocks by member precision, but SPIRV-Cross prints
                 // a RelaxedPrecision member as explicit "mediump" in the vertex stage and as
                 // UNQUALIFIED (mediump-by-default) in the fragment stage; after
@@ -3412,6 +3400,16 @@ namespace MobileGL::MG_Backend::DirectGLES {
                     effectiveSpirv = &noperspectiveSpirv;
                 }
 
+                // ES has no rectangle sampler, and SPIRV-Cross refuses the whole module rather
+                // than approximating one. The shared pass turns the type into the 2D one and
+                // divides the coordinate of every normalized-coordinate lookup by the texture
+                // size, which is the whole of the difference between the two.
+                Vector<unsigned int> rectLoweredSpirv;
+                if (MG_Util::ShaderTranspiler::ShaderCompiler::LowerRectImages(*effectiveSpirv, rectLoweredSpirv) &&
+                    !rectLoweredSpirv.empty()) {
+                    effectiveSpirv = &rectLoweredSpirv;
+                }
+
                 MG_Util::ShaderTranspiler::SpvcSession spvcSession(*effectiveSpirv,
                     MG_Util::ShaderTranspiler::SessionUsageBit::Transpile);
 
@@ -3434,6 +3432,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
                     r.log += spvcSession.GetLastErrorString();
                     r.errc = -5;
                     MGLOG_E("%s", r.log.c_str());
+                    m_backendProgramUsable = false;
                     continue;
                 }
 
@@ -3441,14 +3440,10 @@ namespace MobileGL::MG_Backend::DirectGLES {
 
                 source = RebindImageUniformsToFrontendUnits(std::move(source), stateProgramObject);
                 source = RemoveLayoutBinding(source);
-                if (applyClipDistanceQuirk) {
-                    // Adreno rejects the gl_ClipDistance redeclaration SPIRV-Cross still emits
-                    // ("reserved built-in name") but accepts plain usage with
-                    // GL_EXT_clip_cull_distance required; drop the line, keep the #extension.
-                    source = RemoveClipDistanceRedeclaration(source);
-                }
                 source = ProcessOutColorLocations(source);
                 source = ForceFlatIntegerVaryings(source, glShaderType);
+                source = BroadcastLegacyFragColor(std::move(source), glShaderType, m_fragColorBroadcastCount);
+                source = EmulateTextureLodBias(source);
                 source = EmulateBaseInstanceInVertexShader(std::move(source), glShaderType);
                 source = PromoteDrawParameterGlobalsToUniforms(std::move(source), glShaderType);
                 source = ForceSupporterOutput(source);
@@ -3479,6 +3474,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
                     Vector<GLchar> log(logLength);
                     g_GLESFuncs.glGetShaderInfoLog(backendShaderId, logLength, nullptr, log.data());
                     MGLOG_E("Shader compilation failed for backend ID %u: %s", backendShaderId, log.data());
+                    m_backendProgramUsable = false;
                     continue;
                 }
 
@@ -3488,12 +3484,33 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 MGLOG_D("Processed shader source length: %zu", source.length());
             }
 
+            // Transform feedback capture runs on the real driver (see XfbImpl in
+            // DirectGLES.cpp), so the capture set has to be declared on the backend
+            // program before it links. SPIRV-Cross keeps user output names verbatim in
+            // the transpiled ESSL (`out vec4 result_0;` stays `result_0`), so the
+            // frontend's requested names carry over unchanged.
+            if (stateProgramObject->GetTransformFeedbackVaryingCount() > 0 &&
+                g_GLESFuncs.glTransformFeedbackVaryings != nullptr) {
+                const auto& xfbVaryings = stateProgramObject->GetTransformFeedbackVaryings();
+                Vector<const GLchar*> xfbNames;
+                xfbNames.reserve(xfbVaryings.size());
+                for (const auto& xfbVarying : xfbVaryings) {
+                    xfbNames.push_back(xfbVarying.name.c_str());
+                }
+                MGLOG_D("Declaring %zu transform feedback varyings on program %u", xfbNames.size(),
+                        m_backendProgramId);
+                g_GLESFuncs.glTransformFeedbackVaryings(m_backendProgramId, static_cast<GLsizei>(xfbNames.size()),
+                                                        xfbNames.data(),
+                                                        stateProgramObject->GetTransformFeedbackBufferMode());
+            }
+
             // Link program
             MGLOG_D("Linking program %u", m_backendProgramId);
             g_GLESFuncs.glLinkProgram(m_backendProgramId);
 
             GLint linkStatus;
             g_GLESFuncs.glGetProgramiv(m_backendProgramId, GL_LINK_STATUS, &linkStatus);
+            m_backendProgramUsable = m_backendProgramUsable && linkStatus == GL_TRUE;
             if (linkStatus != GL_TRUE) {
                 GLint logLength;
                 g_GLESFuncs.glGetProgramiv(m_backendProgramId, GL_INFO_LOG_LENGTH, &logLength);
@@ -3606,6 +3623,11 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 binding.backendLocation = backendLoc;
                 binding.uniformType = uniformType;
                 binding.lastAssignedUnit = -1;
+                // Present only for the samplers EmulateTextureLodBias actually rewrote; the
+                // pass names it after the sampler, which SPIRV-Cross preserves verbatim.
+                binding.lodBiasLocation =
+                    g_GLESFuncs.glGetUniformLocation(m_backendProgramId, (String(LOD_BIAS_UNIFORM_PREFIX) + name).c_str());
+                binding.lastAssignedLodBias = 0.0f;
                 m_samplerUniformBindings.push_back(binding);
             }
         }
@@ -3614,12 +3636,18 @@ namespace MobileGL::MG_Backend::DirectGLES {
 #ifdef TRACY_ENABLE
             ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
-            if (g_lastUsedBackendProgramId == m_backendProgramId) {
+            // glUseProgram on a program that did not link is an INVALID_OPERATION and
+            // leaves the *previous* program current, so the draw would silently render
+            // with an unrelated shader (KHR-GL3x.texture_size_promotion read another
+            // test case's alpha that way once a sampler2DRect stage failed to
+            // transpile). Bind nothing instead: the draw is then a visible no-op.
+            const Uint programToBind = m_backendProgramUsable ? m_backendProgramId : 0;
+            if (g_lastUsedBackendProgramId == programToBind) {
                 return;
             }
-            MGLOG_D("Using program %u", m_backendProgramId);
-            g_GLESFuncs.glUseProgram(m_backendProgramId);
-            g_lastUsedBackendProgramId = m_backendProgramId;
+            MGLOG_D("Using program %u", programToBind);
+            g_GLESFuncs.glUseProgram(programToBind);
+            g_lastUsedBackendProgramId = programToBind;
         }
 
         void BackendProgramObjectImpl::SetBaseInstance(Uint32 baseInstance) const {

MobileGL/MG_Backend/DirectGLES/Managers.h

@@ -258,11 +258,6 @@ namespace MobileGL::MG_Backend::DirectGLES {
         private:
             Uint m_backendVAOId = 0;
             Array<Uint, MG_State::GLState::VertexArrayObject::MAX_VERTEX_ATTRIBS> m_clientAttributeBufferIds;
-            // Attribs the frontend has Enabled but that have no source at all (no buffer object
-            // and NULL client pointer). GL keeps such attribs latently enabled, but Adreno's ES
-            // driver treats them as client arrays and memcpys from address 0 at draw time
-            // (SIGSEGV), so they are kept disabled on the backend VAO until they gain a source.
-            Uint32 m_forceDisabledAttribsMask = 0;
             Bool m_isInitialized = false;
             Uint16 m_syncedIndexBufferVersion = 0;
             Array<MG_State::GLState::VertexAttributeVersion, MG_State::GLState::VertexArrayObject::MAX_VERTEX_ATTRIBS>
@@ -275,18 +270,21 @@ namespace MobileGL::MG_Backend::DirectGLES {
 
     namespace TextureImpl {
         inline Bool IsSupportedTextureTarget(TextureTarget target) {
-            // Rectangle textures need non-normalized sampling ES cannot express; everything else is
-            // either native or emulated (1D -> 2D with height 1, 1D array -> 2D array, see
-            // MapToBackendTextureTarget). SPIRV-Cross already emits the matching ESSL samplers and
-            // coordinate padding for 1D/1D-array shaders.
-            return target != TextureTarget::TextureRectangle;
+            // Every desktop-only target is stored on an ES one; see MapToBackendTextureTarget.
+            (void)target;
+            return true;
         }
 
-        // ES has no 1D targets: 1D textures are stored as 2D (height 1) and 1D arrays as 2D arrays
-        // (height 1, layers in depth). Must match SPIRV-Cross's ES 1D-as-2D shader emulation.
+        // ES has none of the desktop-only targets: 1D textures are stored as 2D (height 1), 1D
+        // arrays as 2D arrays (height 1, layers in depth), and rectangle textures as plain 2D -
+        // they are single-level and already clamp, so only the non-normalized coordinates differ.
+        // Must match the shader-side emulation: SPIRV-Cross handles 1D/1D-array itself, and
+        // ShaderCompiler::LowerRectImages rewrites rectangle images (declining any module
+        // whose lookups are not integer-coordinate, which SPIRV-Cross then still rejects).
         inline TextureTarget MapToBackendTextureTarget(TextureTarget target) {
             switch (target) {
             case TextureTarget::Texture1D:
+            case TextureTarget::TextureRectangle:
                 return TextureTarget::Texture2D;
             case TextureTarget::Texture1DArray:
                 return TextureTarget::Texture2DArray;
@@ -302,6 +300,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
         inline GLenum ConvertTextureUploadTargetToBackendGLEnum(TextureUploadTarget uploadTarget) {
             switch (uploadTarget) {
             case TextureUploadTarget::Texture1D:
+            case TextureUploadTarget::TextureRectangle:
                 return GL_TEXTURE_2D;
             case TextureUploadTarget::Texture1DArray:
                 return GL_TEXTURE_2D_ARRAY;
@@ -443,6 +442,13 @@ namespace MobileGL::MG_Backend::DirectGLES {
 
         extern StateBackendObjectRegistry<MG_State::GLState::FramebufferObject, BackendFramebufferObject>
             g_backendFramebufferObjects;
+        // True when the read buffer names a fixed-point (norm/snorm) attachment that the
+        // backend actually stores in a floating-point format. GL clamps a read from a
+        // fixed-point colour buffer to [0,1] (GL_CLAMP_READ_COLOR defaults to
+        // GL_FIXED_ONLY); the substituted float storage would not, so the readback path
+        // has to apply the clamp itself.
+        Bool IsFixedPointFallbackReadAttachment();
+
         extern Array<Uint16, SizeT(FramebufferTarget::FramebufferTargetCount)> g_fboBindVersions;
         // Tracks the bound FBO's object version (bumped on any attachment/drawbuffer change)
         // per target: re-attaching textures or changing draw buffers on an already-bound FBO
@@ -581,6 +587,12 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 Int backendLocation = -1;
                 GLenum uniformType = 0;
                 Int lastAssignedUnit = -1;
+                // Location of this sampler's emulated GL_TEXTURE_LOD_BIAS uniform
+                // (PrgramImpl::EmulateTextureLodBias), -1 when the shader has none.
+                // lastAssignedLodBias mirrors the value the program currently holds,
+                // so an unbiased shader issues no per-draw glUniform1f at all.
+                Int lodBiasLocation = -1;
+                Float lastAssignedLodBias = 0.0f;
             };
 
             BackendProgramObjectImpl();
@@ -592,9 +604,14 @@ namespace MobileGL::MG_Backend::DirectGLES {
             void SetDrawID(Uint32 drawId) const;
             Int GetIndirectParamsBinding() const { return m_indirectParamsBinding; }
             Uint GetBackendProgramId() const { return m_backendProgramId; }
+            // False when the last SyncToBackend could not produce a usable program (a
+            // shader failed to transpile or compile, or the link itself failed). Use()
+            // must not leave the previously bound program current in that case.
+            Bool IsBackendProgramUsable() const { return m_backendProgramUsable; }
             Uint GetBackendGlobalUBOId() const { return m_backendGlobalUBOId; }
             Uint32 GetSnormFallbackClampOutputMask() const { return m_snormFallbackClampOutputMask; }
             Uint32 GetUnormFallbackClampOutputMask() const { return m_unormFallbackClampOutputMask; }
+            Uint GetFragColorBroadcastCount() const { return m_fragColorBroadcastCount; }
 
             Bool HasGlobalUboBlock() const { return m_globalUboBackendBlockIndex >= 0; }
             const Vector<Int>& GetUniformBlockBackendIndices() const { return m_uniformBlockBackendIndices; }
@@ -621,7 +638,11 @@ namespace MobileGL::MG_Backend::DirectGLES {
             Int m_indirectParamsBinding = -1;
             Uint32 m_snormFallbackClampOutputMask = 0;
             Uint32 m_unormFallbackClampOutputMask = 0;
+            // Draw buffers a legacy gl_FragColor write has to reach (see
+            // PrgramImpl::BroadcastLegacyFragColor); 1 keeps the plain single-output shader.
+            Uint m_fragColorBroadcastCount = 1;
             Bool m_isInitialized = false;
+            Bool m_backendProgramUsable = false;
 
             Int m_globalUboBackendBlockIndex = -1;
             Int m_globalUboBackendBlockSize = 0;
@@ -634,6 +655,10 @@ namespace MobileGL::MG_Backend::DirectGLES {
 
         extern Uint32 g_snormFallbackClampOutputMask;
         extern Uint32 g_unormFallbackClampOutputMask;
+        // Draw buffers the current draw framebuffer enables. Like the clamp masks above it
+        // is framebuffer state that the shader has to be compiled against, so a program
+        // whose snapshot no longer matches is relinked.
+        extern Uint g_fragColorBroadcastCount;
         // Backend id of the last glUseProgram issued through this backend; lets Use()
         // skip redundant rebinds. Reset to 0 wherever glUseProgram(0) is issued or the
         // ES context is recreated.

MobileGL/MG_Backend/DirectGLES/Utils.cpp

@@ -22,6 +22,9 @@
 #include <MG_Util/Math/SmallFloat.h>
 
 #include <cmath>
+#include <cctype>
+#include <cstring>
+#include <regex>
 
 namespace MobileGL::MG_Backend::DirectGLES {
     namespace {
@@ -45,15 +48,40 @@ namespace MobileGL::MG_Backend::DirectGLES {
             return options;
         }
 
-        Flags<PixelFormatNormalizeOptionBit> GetRuntimeFallbackNormalizeOptions(GLenum requestedInternalFormat) {
+        Flags<PixelFormatNormalizeOptionBit>
+        GetRuntimeFallbackNormalizeOptions(GLenum requestedInternalFormat,
+                                           Flags<PixelFormatNormalizeOptionBit> extraOptions) {
             using namespace MG_Util::TextureFormatProcessor;
-            const Flags<PixelFormatNormalizeOptionBit> forcedOptions =
-                GetApplicablePixelFormatNormalizeOptions(requestedInternalFormat, GetForcedPixelFormatNormalizeOptions());
+            const Flags<PixelFormatNormalizeOptionBit> forcedOptions = GetApplicablePixelFormatNormalizeOptions(
+                requestedInternalFormat, GetForcedPixelFormatNormalizeOptions() | extraOptions);
             if (forcedOptions) {
                 return forcedOptions;
             }
-            return GetApplicablePixelFormatNormalizeOptions(requestedInternalFormat,
-                                                            GetDriverPixelFormatNormalizeOptions());
+            return GetApplicablePixelFormatNormalizeOptions(
+                requestedInternalFormat, GetDriverPixelFormatNormalizeOptions() | extraOptions);
+        }
+
+        // Multisample textures can only ever be rendered into, never uploaded to, so a fallback
+        // format for them has to stay colour-renderable - a three-channel float fallback is a legal
+        // ES texture format but not a legal multisample storage format. Widening to four channels
+        // is safe here precisely because there is no transfer path that would have to expand
+        // three-channel client data, and the alpha the draw writes for a three-channel source is
+        // already the 1.0 the frontend format implies.
+        Bool TargetRequiresRenderableFormat(SizeT targetIndex) {
+            return targetIndex == static_cast<SizeT>(TextureTarget::Texture2DMultisample) ||
+                   targetIndex == static_cast<SizeT>(TextureTarget::Texture2DMultisampleArray);
+        }
+
+        Flags<PixelFormatNormalizeOptionBit> GetRenderTargetNormalizeOptions(SizeT targetIndex) {
+            Flags<PixelFormatNormalizeOptionBit> options;
+            if (!TargetRequiresRenderableFormat(targetIndex)) {
+                return options;
+            }
+            options |= PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget;
+            if (!g_GLESCapabilities.SupportsRenderSnorm || !g_GLESCapabilities.SupportsNorm16Texture) {
+                options |= PixelFormatNormalizeOptionBit::NoSnorm16RenderTarget;
+            }
+            return options;
         }
 
         Bool HasCachedFormatCapability(TextureInternalFormat internalFormat,
@@ -113,7 +141,8 @@ namespace MobileGL::MG_Backend::DirectGLES {
             const GLenum requestedInternalFormat = MG_Util::ConvertTextureInternalFormatToGLEnum(internalFormat);
             Flags<PixelFormatNormalizeOptionBit> options;
             if (!pActiveBackendObject || ShouldUseCaveatFormat(internalFormat, targetIndex)) {
-                options = GetRuntimeFallbackNormalizeOptions(requestedInternalFormat);
+                options = GetRuntimeFallbackNormalizeOptions(requestedInternalFormat,
+                                                             GetRenderTargetNormalizeOptions(targetIndex));
             }
             NormalizePixelFormat(requestedInternalFormat, options, outInternalFormat, outFormat, outType);
         }
@@ -148,6 +177,22 @@ namespace MobileGL::MG_Backend::DirectGLES {
         Bool ShouldUseCaveatRenderbufferFormat(TextureInternalFormat internalFormat) {
             return ShouldUseCaveatFormat(internalFormat, GetRenderbufferFormatCapabilityTargetIndex());
         }
+
+        Bool BackendTextureFormatAddsAlpha(TextureInternalFormat internalFormat, TextureTarget target) {
+            const SizeT targetIndex =
+                target == TextureTarget::Unknown ? kFormatCapabilityTargetCount : GetFormatCapabilityTargetIndex(target);
+            if (!TargetRequiresRenderableFormat(targetIndex)) {
+                return false;
+            }
+            if (pActiveBackendObject && !ShouldUseCaveatFormat(internalFormat, targetIndex)) {
+                return false;
+            }
+            const GLenum requestedInternalFormat = MG_Util::ConvertTextureInternalFormatToGLEnum(internalFormat);
+            const Flags<PixelFormatNormalizeOptionBit> options =
+                GetRuntimeFallbackNormalizeOptions(requestedInternalFormat,
+                                                   GetRenderTargetNormalizeOptions(targetIndex));
+            return static_cast<Bool>(options & PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget);
+        }
     } // namespace TextureImpl
     namespace PrgramImpl {
         String ProcessOutColorLocations(const String& glslCode) {
@@ -276,6 +321,55 @@ namespace MobileGL::MG_Backend::DirectGLES {
             return glslCode;
         }
 
+        String BroadcastLegacyFragColor(String glslCode, GLenum shaderType, Uint drawBufferCount) {
+#ifdef TRACY_ENABLE
+            ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
+#endif
+            // The name is the marker: ShaderSourceProcessor only emits it when the source
+            // wrote gl_FragColor, and such a shader can have no other output.
+            static const char* const kLoweredName = "mg_FragColor";
+            if (shaderType != GL_FRAGMENT_SHADER || drawBufferCount <= 1) {
+                return glslCode;
+            }
+            static const std::regex declRegex(
+                R"(layout\s*\(\s*location\s*=\s*0\s*\)\s*out\s+((?:lowp|mediump|highp)\s+)?vec4\s+mg_FragColor\s*;)");
+            std::smatch declMatch;
+            if (!std::regex_search(glslCode, declMatch, declRegex)) {
+                return glslCode;
+            }
+            const String precision = declMatch[1].matched ? declMatch[1].str() : String();
+
+            String replicaDecls;
+            String replicaCopies;
+            for (Uint location = 1; location < drawBufferCount; ++location) {
+                const String name = String(kLoweredName) + "_" + std::to_string(location);
+                replicaDecls += "\nlayout(location = " + std::to_string(location) + ") out " + precision + "vec4 " +
+                                name + ";";
+                replicaCopies += "\n    " + name + " = " + kLoweredName + ";";
+            }
+
+            static const std::regex mainRegex(R"(void\s+main\s*\([^)]*\)\s*\{)");
+            std::smatch mainMatch;
+            if (!std::regex_search(glslCode, mainMatch, mainRegex)) {
+                return glslCode;
+            }
+            SizeT bracePos = static_cast<SizeT>(mainMatch.position(0) + mainMatch.length(0) - 1);
+            Int depth = 0;
+            for (SizeT pos = bracePos; pos < glslCode.size(); ++pos) {
+                if (glslCode[pos] == '{') {
+                    ++depth;
+                } else if (glslCode[pos] == '}') {
+                    --depth;
+                    if (depth == 0) {
+                        glslCode.insert(pos, replicaCopies + "\n");
+                        break;
+                    }
+                }
+            }
+            glslCode.insert(static_cast<SizeT>(declMatch.position(0)) + declMatch[0].str().size(), replicaDecls);
+            return glslCode;
+        }
+
         String ForceFlatIntegerVaryings(const String& glslCode, GLenum shaderType) {
 #ifdef TRACY_ENABLE
             ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
@@ -343,40 +437,166 @@ namespace MobileGL::MG_Backend::DirectGLES {
             return result;
         }
 
-        String RemoveClipDistanceRedeclaration(const String& glslCode) {
+        namespace {
+            // How a lookup carries its level of detail, and how many arguments it takes
+            // before the optional bias.
+            struct LodLookupForm {
+                const char* name;
+                Int requiredArgs; // arguments before the optional bias (implicit form)
+                Int explicitLodArg; // index of the explicit LOD argument, -1 for implicit
+            };
+
+            // texelFetch* is deliberately absent: an integer fetch names its level directly
+            // and takes no LOD bias. textureGather has no bias either. textureGrad* derives
+            // the LOD from gradients and offers no argument to fold a bias into, so it is
+            // left alone rather than rewritten incorrectly.
+            constexpr LodLookupForm LOD_LOOKUP_FORMS[] = {
+                {"textureProjLodOffset", 0, 2}, {"textureProjOffset", 4, -1}, {"textureProjLod", 0, 2},
+                {"textureLodOffset", 0, 2},     {"textureOffset", 3, -1},     {"textureProj", 2, -1},
+                {"textureLod", 0, 2},           {"texture", 2, -1},
+            };
+
+            // Sampler types with no mip chain, or whose GLSL lookups have no bias overload
+            // at all (the array-shadow forms), so nothing can or should be folded in.
+            Bool IsBiasableSamplerType(const String& samplerType) {
+                if (samplerType.find("MS") != String::npos) return false;      // multisample
+                if (samplerType.find("Buffer") != String::npos) return false;  // texture buffer
+                if (samplerType.find("Rect") != String::npos) return false;    // rectangle: no mips
+                if (samplerType == "sampler2DArrayShadow") return false;
+                if (samplerType == "samplerCubeArrayShadow") return false;
+                return true;
+            }
+
+            Bool IsIdentifierChar(char c) { return std::isalnum(static_cast<unsigned char>(c)) || c == '_'; }
+
+            // Byte offsets of the top-level argument separators and of the closing paren,
+            // starting from the '(' at openParen. Empty when the parentheses do not balance.
+            Vector<SizeT> SplitCallArguments(const String& code, SizeT openParen) {
+                Vector<SizeT> marks;
+                Int depth = 0;
+                for (SizeT i = openParen; i < code.size(); ++i) {
+                    const char c = code[i];
+                    if (c == '(' || c == '[') {
+                        ++depth;
+                    } else if (c == ']') {
+                        --depth;
+                    } else if (c == ')') {
+                        --depth;
+                        if (depth == 0) {
+                            marks.push_back(i);
+                            return marks;
+                        }
+                    } else if (c == ',' && depth == 1) {
+                        marks.push_back(i);
+                    }
+                }
+                return {};
+            }
+        } // namespace
+
+        String EmulateTextureLodBias(const String& glslCode) {
 #ifdef TRACY_ENABLE
             ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
-            // Adreno rejects any redeclaration of gl_ClipDistance/gl_CullDistance ("reserved
-            // built-in name") even with GL_EXT_clip_cull_distance required, but accepts plain
-            // usage of the builtin. Drop the desktop-style redeclaration line SPIRV-Cross
-            // prints; the "#extension GL_EXT_clip_cull_distance : require" line stays.
-            static const std::regex redeclarationRegex(
-                R"(^\s*(?:out|in)\s+(?:(?:high|medium|low)p\s+)?float\s+gl_(?:Clip|Cull)Distance\[[0-9]+\];\s*$)");
+            if (glslCode.find("sampler") == String::npos || glslCode.find("texture") == String::npos) {
+                return glslCode;
+            }
 
-            String result;
-            result.reserve(glslCode.size());
-            SizeT lineStart = 0;
-            Bool firstLine = true;
-            while (lineStart <= glslCode.size()) {
-                SizeT lineEnd = glslCode.find('\n', lineStart);
-                const Bool lastLine = lineEnd == String::npos;
-                String line = glslCode.substr(lineStart, lastLine ? String::npos : lineEnd - lineStart);
+            // Collect the mip-capable sampler uniforms this shader declares.
+            static const std::regex samplerDeclRegex(
+                R"(uniform\s+(?:(?:highp|mediump|lowp)\s+)?([iu]?sampler[A-Za-z0-9]*)\s+([A-Za-z_][A-Za-z0-9_]*)\s*;)");
+            UnorderedMap<String, String> samplerNames; // name -> bias uniform name
+            for (std::sregex_iterator it(glslCode.begin(), glslCode.end(), samplerDeclRegex), end; it != end; ++it) {
+                const String samplerType = (*it)[1].str();
+                if (!IsBiasableSamplerType(samplerType)) continue;
+                const String name = (*it)[2].str();
+                samplerNames.emplace(name, String(LOD_BIAS_UNIFORM_PREFIX) + name);
+            }
+            if (samplerNames.empty()) {
+                return glslCode;
+            }
 
-                if (!std::regex_match(line, redeclarationRegex)) {
-                    if (!firstLine) {
-                        result += '\n';
-                    }
-                    result += line;
-                    firstLine = false;
-                }
-                if (lastLine) {
+            // Rewrite the lookups. Right-to-left so earlier offsets stay valid, and only for
+            // samplers named directly as the first argument (SPIRV-Cross never produces an
+            // expression there for ES output, which has no separate sampler objects).
+            String result = glslCode;
+            Vector<String> usedSamplers;
+            for (SizeT scan = result.size(); scan-- > 0;) {
+                if (result[scan] != 't') continue;
+                if (scan > 0 && IsIdentifierChar(result[scan - 1])) continue;
+
+                const LodLookupForm* form = nullptr;
+                SizeT openParen = 0;
+                for (const auto& candidate : LOD_LOOKUP_FORMS) {
+                    const SizeT nameLength = std::strlen(candidate.name);
+                    if (result.compare(scan, nameLength, candidate.name) != 0) continue;
+                    SizeT after = result.find_first_not_of(" \t", scan + nameLength);
+                    if (after == String::npos || result[after] != '(') continue;
+                    form = &candidate;
+                    openParen = after;
                     break;
                 }
-                lineStart = lineEnd + 1;
+                if (form == nullptr) continue;
+
+                const Vector<SizeT> marks = SplitCallArguments(result, openParen);
+                if (marks.empty()) continue;
+                const SizeT argCount = marks.size();
+                const SizeT closeParen = marks.back();
+
+                // First argument must be one of our samplers.
+                const SizeT firstArgStart = result.find_first_not_of(" \t", openParen + 1);
+                SizeT firstArgEnd = marks.front();
+                while (firstArgEnd > firstArgStart && (result[firstArgEnd - 1] == ' ' || result[firstArgEnd - 1] == '\t')) {
+                    --firstArgEnd;
+                }
+                if (firstArgStart == String::npos || firstArgEnd <= firstArgStart) continue;
+                const String samplerName = result.substr(firstArgStart, firstArgEnd - firstArgStart);
+                const auto samplerIt = samplerNames.find(samplerName);
+                if (samplerIt == samplerNames.end()) continue;
+
+                const String& biasName = samplerIt->second;
+                if (form->explicitLodArg >= 0) {
+                    // Explicit LOD: the bias adds to it, as Vulkan does for
+                    // OpImageSampleExplicitLod and as the CTS reference expects.
+                    const SizeT lodIndex = static_cast<SizeT>(form->explicitLodArg);
+                    if (argCount <= lodIndex) continue;
+                    const SizeT lodStart = marks[lodIndex - 1] + 1;
+                    const SizeT lodEnd = marks[lodIndex];
+                    result.insert(lodEnd, String(") + ") + biasName + ")");
+                    result.insert(lodStart, "((");
+                } else {
+                    const SizeT required = static_cast<SizeT>(form->requiredArgs);
+                    if (argCount == required) {
+                        result.insert(closeParen, String(", ") + biasName);
+                    } else if (argCount == required + 1) {
+                        const SizeT biasStart = marks[argCount - 2] + 1;
+                        result.insert(closeParen, String(") + ") + biasName + ")");
+                        result.insert(biasStart, "((");
+                    } else {
+                        continue;
+                    }
+                }
+                usedSamplers.push_back(samplerName);
+            }
+            if (usedSamplers.empty()) {
+                return glslCode;
+            }
+
+            // Declare the bias uniforms that were actually referenced, right after the
+            // sampler declaration line they belong to.
+            for (const auto& samplerName : usedSamplers) {
+                const String& biasName = samplerNames[samplerName];
+                if (result.find(String("float ") + biasName + ";") != String::npos) continue;
+                const std::regex declRegex(
+                    R"(uniform\s+(?:(?:highp|mediump|lowp)\s+)?[iu]?sampler[A-Za-z0-9]*\s+)" + samplerName + R"(\s*;)");
+                std::smatch match;
+                if (!std::regex_search(result, match, declRegex)) continue;
+                const SizeT declEnd = static_cast<SizeT>(match.position(0)) + match[0].str().size();
+                result.insert(declEnd, String("\nuniform highp float ") + biasName + ";");
             }
             return result;
         }
+
     } // namespace PrgramImpl
 
     namespace Utils {

MobileGL/MG_Backend/DirectGLES/Utils.h

@@ -40,6 +40,11 @@ namespace MobileGL::MG_Backend::DirectGLES {
         void GenerateRenderbufferFormatInfo(TextureInternalFormat internalFormat, GLenum* outInternalFormat,
                                             GLenum* outFormat, GLenum* outType);
         Bool ShouldUseCaveatTextureFormat(TextureInternalFormat internalFormat, TextureTarget target);
+
+        // True when the format the texture is actually created with has an alpha channel the
+        // frontend format does not (the three-channel multisample widening). GL reads such a
+        // channel back as 1.0, so any swizzle source of ALPHA has to be answered with ONE.
+        Bool BackendTextureFormatAddsAlpha(TextureInternalFormat internalFormat, TextureTarget target);
         Bool ShouldUseCaveatRenderbufferFormat(TextureInternalFormat internalFormat);
     } // namespace TextureImpl
 
@@ -104,8 +109,26 @@ namespace MobileGL::MG_Backend::DirectGLES {
         String ClampNormFallbackOutputs(String glslCode, GLenum shaderType, Uint32 snormOutputMask,
                                         Uint32 unormOutputMask);
         String ForceFlatIntegerVaryings(const String& glslCode, GLenum shaderType);
+        // Legacy GLSL's gl_FragColor is broadcast to every enabled draw buffer (GL 4.6
+        // 15.2.3), but ShaderSourceProcessor lowers it to the single output mg_FragColor,
+        // which only ever reaches draw buffer 0. Replicates it across `drawBufferCount`
+        // outputs and copies the value into them at the end of main. A no-op for
+        // drawBufferCount <= 1, i.e. for everything but a framebuffer that actually
+        // enables several draw buffers, so the ordinary single-target shader is untouched.
+        String BroadcastLegacyFragColor(String glslCode, GLenum shaderType, Uint drawBufferCount);
         String RemoveLayoutBinding(const String& glslCode);
-        String RemoveClipDistanceRedeclaration(const String& glslCode);
+        // Prefix of the per-sampler float uniform that carries GL_TEXTURE_LOD_BIAS into
+        // the shader (see EmulateTextureLodBias); the suffix is the sampler's own name.
+        constexpr const char* LOD_BIAS_UNIFORM_PREFIX = "mg_lodBias_";
+        // ES has no per-texture/sampler LOD bias at all (GL_TEXTURE_LOD_BIAS is desktop
+        // only; Vulkan spells it VkSamplerCreateInfo::mipLodBias), so it has to reach the
+        // shader as a uniform and be folded into every lookup's level of detail. Declares
+        // one `uniform highp float mg_lodBias_<sampler>;` per mip-capable sampler and adds
+        // it to the bias / explicit-LOD argument of every lookup that takes one. Draws push
+        // the bound texture's (or sampler object's) value into it; a shader whose samplers
+        // all have a zero bias is therefore unaffected. Returns the source unchanged when
+        // there is nothing to rewrite.
+        String EmulateTextureLodBias(const String& glslCode);
     } // namespace PrgramImpl
 
     namespace Utils {

MobileGL/MG_Backend/DirectVulkan/BackendObject_DirectVulkan.cpp

@@ -18,6 +18,7 @@
 #include "MG_Util/Texture/TextureFormatProcessor.h"
 
 #include <Config.h>
+#include <cmath>
 #include <cstdlib>
 #include <cstring>
 
@@ -469,6 +470,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // treat them as signaled/available with zero results from here on.
         BumpRendererGeneration();
         pVulkanRenderer.reset();
+        // The reflection cache is file-scope, not renderer-owned; without this the
+        // deleted programs' reflection strings survive full context teardown.
+        ClearProgramResourceCaches();
         BackendObject::ReleaseEGLResources();
     }
 
@@ -478,6 +482,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // treat them as signaled/available with zero results from here on.
         BumpRendererGeneration();
         pVulkanRenderer.reset();
+        // The reflection cache is file-scope, not renderer-owned; without this the
+        // deleted programs' reflection strings survive full context teardown.
+        ClearProgramResourceCaches();
     }
 
     const RendererInfo& BackendObject_DirectVulkan::GetRendererInfo() const {
@@ -519,10 +526,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                           E_GL_ARB_multi_draw_indirect, E_GL_ARB_indirect_parameters,
                                           E_GL_EXT_framebuffer_object, E_GL_ARB_depth_texture, E_GL_ARB_buffer_storage,
                                           E_GL_ARB_texture_storage, E_GL_ARB_texture_storage_multisample,
-                                          E_GL_ARB_clear_texture, E_GL_ARB_direct_state_access,
+                                          E_GL_ARB_texture_multisample, E_GL_ARB_clear_texture, E_GL_ARB_direct_state_access,
                                           E_GL_ARB_shader_draw_parameters, E_GL_ARB_gpu_shader_int64, E_GL_KHR_debug,
                                           E_GL_ARB_gpu_shader5, E_GL_ARB_multi_bind, E_GL_ARB_shading_language_420pack,
-                                          E_GL_ARB_vertex_attrib_binding, E_GL_ARB_shader_image_size};
+                                          E_GL_ARB_vertex_attrib_binding, E_GL_ARB_shader_image_size,
+                                          E_GL_ARB_explicit_attrib_location,
+                                          // Advertised with GL_NUM_PROGRAM_BINARY_FORMATS = 0, which the
+                                          // extension explicitly permits. It is also the only thing that
+                                          // exposes glProgramParameteri before GL 4.1.
+                                          E_GL_ARB_get_program_binary};
         if (shaderSubgroupSupported && !MG_Config::Features.DisableSubgroup) {
             extensions.push_back(E_GL_KHR_shader_subgroup);
         }
@@ -628,6 +640,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 funcsTable.GL.DeleteBackendQuery = DeleteBackendQuery;
                 funcsTable.GL.GetGpuTimestampNs = GetGpuTimestampNs;
             }
+            // Occlusion queries share the handle-based result/delete entries, which must
+            // exist even when timer queries are disabled.
+            funcsTable.GL.BeginOcclusionQuery = BeginOcclusionQuery;
+            funcsTable.GL.EndOcclusionQuery = EndOcclusionQuery;
+            funcsTable.GL.BeginXfbPrimitivesQuery = BeginXfbPrimitivesQuery;
+            funcsTable.GL.EndXfbPrimitivesQuery = EndXfbPrimitivesQuery;
+            funcsTable.GL.IsQueryResultAvailable = IsQueryResultAvailable;
+            funcsTable.GL.GetQueryResult64 = GetQueryResult64;
+            funcsTable.GL.DeleteBackendQuery = DeleteBackendQuery;
             funcsTableInitialized = true;
         }
         return funcsTable;
@@ -759,6 +780,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_dynamicParameters.MaxComputeWorkGroupInvocations = m_vulkanCaps.MaxComputeWorkGroupInvocations;
         m_dynamicParameters.MaxShaderStorageBufferBindings = m_vulkanCaps.MaxShaderStorageBufferBindings;
         m_dynamicParameters.MaxTextureBufferSize = m_vulkanCaps.MaxTextureBufferSize;
+        m_dynamicParameters.TextureBufferOffsetAlignment = m_vulkanCaps.TextureBufferOffsetAlignment;
         m_dynamicParameters.MaxUniformBufferBindings = m_vulkanCaps.MaxUniformBufferBindings;
         m_dynamicParameters.MaxUniformBlockSize = m_vulkanCaps.MaxUniformBlockSize;
         m_dynamicParameters.MaxImageUnits =
@@ -790,6 +812,23 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_dynamicParameters.ViewportBoundsRangeMin = m_vulkanCaps.ViewportBoundsRangeMin;
         m_dynamicParameters.ViewportBoundsRangeMax = m_vulkanCaps.ViewportBoundsRangeMax;
         m_dynamicParameters.ViewportSubpixelBits = m_vulkanCaps.ViewportSubpixelBits;
+        m_dynamicParameters.MinFragmentInterpolationOffset =
+            std::isfinite(m_vulkanCaps.MinFragmentInterpolationOffset) &&
+                    m_vulkanCaps.MinFragmentInterpolationOffset <= -0.5f
+                ? m_vulkanCaps.MinFragmentInterpolationOffset
+                : -0.5f;
+        m_dynamicParameters.MaxFragmentInterpolationOffset = 0.4375f;
+        m_dynamicParameters.FragmentInterpolationOffsetBits = 4;
+        if (m_vulkanCaps.FragmentInterpolationOffsetBits >= 4 &&
+            std::isfinite(m_vulkanCaps.MaxFragmentInterpolationOffset)) {
+            const Float requiredMaxOffset =
+                0.5f - std::ldexp(1.0f, -m_vulkanCaps.FragmentInterpolationOffsetBits);
+            if (m_vulkanCaps.MaxFragmentInterpolationOffset >= requiredMaxOffset) {
+                m_dynamicParameters.MaxFragmentInterpolationOffset = m_vulkanCaps.MaxFragmentInterpolationOffset;
+                m_dynamicParameters.FragmentInterpolationOffsetBits =
+                    m_vulkanCaps.FragmentInterpolationOffsetBits;
+            }
+        }
         m_dynamicParameters.SupportsWideLines = m_vulkanCaps.SupportsWideLines;
         m_dynamicParameters.MaxShaderStorageBlockSize =
             std::min(m_vulkanCaps.MaxShaderStorageBlockSize, kMaxAdvertisedShaderStorageBlockSize);

MobileGL/MG_Backend/DirectVulkan/DirectVulkan.cpp

@@ -61,6 +61,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         };
 
         struct ProgramResourceCache {
+            // Lifetime id of the program the cached reflection belongs to. GL names are
+            // recycled (IndexGenerator hands freed indices straight back), and a
+            // recreated program's backendStateVersion restarts at the same small values,
+            // so the version alone can collide; the never-reused lifetime id makes the
+            // slot's ownership unambiguous.
+            Uint64 programLifetimeId = 0;
             Uint32 backendStateVersion = 0;
             Vector<StorageBlockResource> storageBlocks;
             Vector<BufferVariableResource> bufferVariables;
@@ -82,6 +88,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             Uint32 baseInstance = 0;
         };
 
+        // Keyed by GL program name so the freed-name reuse in IndexGenerator bounds the
+        // map at the peak-simultaneous-program high-water mark; each slot's ownership is
+        // checked against the program's lifetime id before it is served (see
+        // GetProgramResourceCache). Cleared wholesale at EGL teardown via
+        // ClearProgramResourceCaches.
         UnorderedMap<GLuint, ProgramResourceCache> g_programResourceCaches;
 
         void ClearReadPixelsOutput(GLsizei width, GLsizei height, GLenum format, GLenum type, void* pixels) {
@@ -142,13 +153,19 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         ProgramResourceCache& GetProgramResourceCache(const MG_State::GLState::ProgramObject& program) {
             auto& cache = g_programResourceCaches[program.GetExternalIndex()];
+            const Uint64 programLifetimeId = program.GetLifetimeId();
             const Uint32 backendStateVersion = program.GetBackendStateVersion();
-            if (cache.backendStateVersion == backendStateVersion &&
+            // The lifetime id must match too: a new program that reuses a deleted
+            // program's name and happens to land on the same backendStateVersion (both
+            // count from zero) would otherwise be served the dead program's reflection.
+            if (cache.programLifetimeId == programLifetimeId &&
+                cache.backendStateVersion == backendStateVersion &&
                 (!cache.storageBlocks.empty() || !cache.bufferVariables.empty())) {
                 return cache;
             }
 
             cache = {};
+            cache.programLifetimeId = programLifetimeId;
             cache.backendStateVersion = backendStateVersion;
 
             Vector<SpvReflectShaderModule> modules;
@@ -366,6 +383,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
     } // namespace
 
+    void ClearProgramResourceCaches() {
+        // Called from EGL teardown while the backend's m_eglStateMutex is held; GL
+        // calls are serialized in this codebase (contexts migrate threads but never
+        // run concurrently), so no other thread can be inside the unsynchronized map.
+        // Live programs in another context self-heal: their entry rebuilds from the
+        // retained generated SPIR-V on the next resource query.
+        g_programResourceCaches.clear();
+    }
+
     GLuint GetShaderStorageBlockIndex(const MG_State::GLState::ProgramObject& program, const String& name) {
         auto& cache = GetProgramResourceCache(program);
         const auto it = std::find_if(cache.storageBlocks.begin(), cache.storageBlocks.end(),
@@ -1224,10 +1250,76 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         pVulkanRenderer->Clear(mask);
     }
 
+    // Vulkan has no LINE_LOOP topology; rewrite the draw as an indexed LINE_STRIP
+    // whose synthesized index list revisits the first vertex at the end.
+    static void DrawLineLoopAsIndexedStrip(const Vector<Uint32>& closedIndices, GLint basevertex) {
+        DrawIndexedCmd payload{};
+        payload.mode = GL_LINE_STRIP;
+        payload.indexBufferView.indexType = GL_UNSIGNED_INT;
+        payload.indexBufferView.indexByteOffset = reinterpret_cast<SizeT>(closedIndices.data());
+        payload.indexBufferView.indexByteSize = closedIndices.size() * sizeof(Uint32);
+        payload.indexBufferView.forceClientMemory = true;
+        payload.params.indexCount = static_cast<Uint32>(closedIndices.size());
+        payload.params.instanceCount = 1;
+        payload.params.vertexOffset = basevertex;
+        pVulkanRenderer->DrawElements(payload);
+    }
+
+    // Resolve a DrawElements index list (bound element-array buffer or client
+    // memory) into uint32 values with the loop-closing first index appended.
+    static Bool BuildClosedLineLoopIndices(GLsizei count, GLenum type, const void* indices,
+                                           Vector<Uint32>& outIndices) {
+        const SizeT indexSize = MG_Util::GetGLTypeSize(type);
+        if (indexSize == 0 || count < 2) {
+            return false;
+        }
+        const Uint8* indexBytes = nullptr;
+        const auto& vao = *MG_State::pGLContext->GetBoundVertexArray();
+        const auto& indexBufferShared = vao.GetIndexBufferBindingSlot().GetBoundObject();
+        if (indexBufferShared != nullptr) {
+            const SizeT offset = reinterpret_cast<SizeT>(indices);
+            const SizeT bufferSize = indexBufferShared->GetSize();
+            if (indexBufferShared->MappedData() == nullptr || offset > bufferSize ||
+                static_cast<SizeT>(count) * indexSize > bufferSize - offset) {
+                return false;
+            }
+            indexBufferShared->SyncPersistentMappedRange();
+            indexBytes = indexBufferShared->MappedData() + offset;
+        } else {
+            indexBytes = static_cast<const Uint8*>(indices);
+            if (indexBytes == nullptr) {
+                return false;
+            }
+        }
+        outIndices.resize(static_cast<SizeT>(count) + 1);
+        for (GLsizei i = 0; i < count; ++i) {
+            switch (indexSize) {
+            case 1: outIndices[i] = indexBytes[i]; break;
+            case 2: outIndices[i] = reinterpret_cast<const Uint16*>(indexBytes)[i]; break;
+            default: outIndices[i] = reinterpret_cast<const Uint32*>(indexBytes)[i]; break;
+            }
+        }
+        outIndices[count] = outIndices[0];
+        return true;
+    }
+
     void DrawArrays(GLenum mode, GLint first, GLsizei count) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::DrawArrays called with null VulkanRenderer");
         MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::DrawArrays called with null GL context");
 
+        if (mode == GL_LINE_LOOP) {
+            if (count < 2) {
+                return;
+            }
+            Vector<Uint32> closedIndices(static_cast<SizeT>(count) + 1);
+            for (GLsizei i = 0; i < count; ++i) {
+                closedIndices[i] = static_cast<Uint32>(first + i);
+            }
+            closedIndices[count] = static_cast<Uint32>(first);
+            DrawLineLoopAsIndexedStrip(closedIndices, 0);
+            return;
+        }
+
         DrawCmd payload{};
         payload.mode = mode;
         payload.params.firstVertex = first;
@@ -1240,6 +1332,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::DrawElements called with null VulkanRenderer");
         MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::DrawElements called with null GL context");
 
+        if (mode == GL_LINE_LOOP) {
+            Vector<Uint32> closedIndices;
+            if (BuildClosedLineLoopIndices(count, type, indices, closedIndices)) {
+                DrawLineLoopAsIndexedStrip(closedIndices, 0);
+            }
+            return;
+        }
+
         DrawIndexedCmd payload{};
         payload.mode = mode;
         payload.indexBufferView.indexType = type;
@@ -1308,6 +1408,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     void DrawElementsBaseVertex(GLenum mode, GLsizei count, GLenum type, const GLvoid* indices, GLint basevertex) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::DrawElementsBaseVertex called with null VulkanRenderer");
         MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::DrawElementsBaseVertex called with null GL context");
+        if (mode == GL_LINE_LOOP) {
+            Vector<Uint32> closedIndices;
+            if (BuildClosedLineLoopIndices(count, type, indices, closedIndices)) {
+                DrawLineLoopAsIndexedStrip(closedIndices, basevertex);
+            }
+            return;
+        }
         DrawIndexedCmd payload{};
         payload.mode = mode;
         payload.indexBufferView.indexType = type;
@@ -1457,8 +1564,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // records are shared (SharedPtr) with the owning pool's pending list,
         // so deleting the query while results are still in flight is safe.
         struct VulkanTimerQuery {
+            enum class Kind : Uint8 { Timer, Occlusion, XfbWritten, XfbGenerated };
+            Kind kind = Kind::Timer;
             SharedPtr<VkTimerQueryManager::TimestampRecord> begin;
             SharedPtr<VkTimerQueryManager::TimestampRecord> end;
+            // Kind::Occlusion - pool slots recorded between Begin/End; summed at result time.
+            Vector<Uint32> occlusionSlots;
             // Renderer generation the records were written under (see
             // g_rendererGeneration). A stale generation resolves as available
             // with a final zero result: the records' pool indices and frame
@@ -1467,6 +1578,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // (and, via the SharedPtrs, the records), never pool slots, so
             // stale queries are always safe to delete.
             Uint64 rendererGeneration = 0;
+            // Kind::XfbGenerated - the frontend's paused-draw primitive counter when the
+            // query began. VK_QUERY_TYPE_TRANSFORM_FEEDBACK_STREAM_EXT counts only what the
+            // capture saw, so a draw made while the span was paused is invisible to it -
+            // but GL_PRIMITIVES_GENERATED counts what the last vertex processing stage
+            // emitted regardless. The delta closes that gap at result time.
+            Uint64 pausedPrimitiveSnapshot = 0;
         };
     } // namespace
 
@@ -1548,6 +1665,30 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // ever be produced, so resolve with a final 0.
             return true;
         }
+        if (query->kind == VulkanTimerQuery::Kind::Occlusion) {
+            Uint64 samples = 0;
+            if (!pVulkanRenderer->ResolveOcclusionQueryResult(query->occlusionSlots, samples)) {
+                return false;
+            }
+            query->occlusionSlots.clear(); // slots are recycled by the resolve
+            *outNanoseconds = samples;
+            return true;
+        }
+        if (query->kind == VulkanTimerQuery::Kind::XfbWritten ||
+            query->kind == VulkanTimerQuery::Kind::XfbGenerated) {
+            Uint64 primitives = 0;
+            if (!pVulkanRenderer->ResolveXfbQueryResult(query->occlusionSlots,
+                                                        query->kind == VulkanTimerQuery::Kind::XfbGenerated,
+                                                        primitives)) {
+                return false;
+            }
+            if (query->kind == VulkanTimerQuery::Kind::XfbGenerated && MG_State::pGLContext != nullptr) {
+                primitives += MG_State::pGLContext->GetTransformFeedbackPausedPrimitiveCounter() -
+                              query->pausedPrimitiveSnapshot;
+            }
+            *outNanoseconds = primitives;
+            return true;
+        }
         // With wait, mirrors ClientWaitSync: a query ended this frame cannot
         // complete until Present submits the commands, so the wait refuses to
         // block on the current unsubmitted serial. Returning false keeps the
@@ -1580,6 +1721,49 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         delete static_cast<VulkanTimerQuery*>(handle);
     }
 
+    BackendQueryHandle BeginXfbPrimitivesQuery(Bool generated) {
+        MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::BeginXfbPrimitivesQuery called with null VulkanRenderer");
+        if (!pVulkanRenderer->StartXfbQueryCapture(generated ? 1u : 0u)) {
+            return nullptr;
+        }
+        auto* query = new VulkanTimerQuery{};
+        query->kind = generated ? VulkanTimerQuery::Kind::XfbGenerated : VulkanTimerQuery::Kind::XfbWritten;
+        query->rendererGeneration = GetRendererGeneration();
+        query->pausedPrimitiveSnapshot =
+            MG_State::pGLContext ? MG_State::pGLContext->GetTransformFeedbackPausedPrimitiveCounter() : 0;
+        return query;
+    }
+
+    void EndXfbPrimitivesQuery(BackendQueryHandle handle) {
+        auto* query = static_cast<VulkanTimerQuery*>(handle);
+        MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::EndXfbPrimitivesQuery called with null VulkanRenderer");
+        if (query == nullptr || query->rendererGeneration != GetRendererGeneration()) {
+            return;
+        }
+        pVulkanRenderer->StopXfbQueryCapture(
+            query->kind == VulkanTimerQuery::Kind::XfbGenerated ? 1u : 0u, query->occlusionSlots);
+    }
+
+    BackendQueryHandle BeginOcclusionQuery() {
+        MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::BeginOcclusionQuery called with null VulkanRenderer");
+        if (!pVulkanRenderer->StartOcclusionQueryCapture()) {
+            return nullptr;
+        }
+        auto* query = new VulkanTimerQuery{};
+        query->kind = VulkanTimerQuery::Kind::Occlusion;
+        query->rendererGeneration = GetRendererGeneration();
+        return query;
+    }
+
+    void EndOcclusionQuery(BackendQueryHandle handle) {
+        auto* query = static_cast<VulkanTimerQuery*>(handle);
+        MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::EndOcclusionQuery called with null VulkanRenderer");
+        if (query == nullptr || query->rendererGeneration != GetRendererGeneration()) {
+            return;
+        }
+        pVulkanRenderer->StopOcclusionQueryCapture(query->occlusionSlots);
+    }
+
     Int64 GetGpuTimestampNs() {
         // Vulkan cannot synchronously sample the GPU clock: timestamps only
         // exist as vkCmdWriteTimestamp results read back later, and

MobileGL/MG_Backend/DirectVulkan/DirectVulkan.h

@@ -23,6 +23,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     Uint64 GetRendererGeneration();
     void BumpRendererGeneration();
 
+    // Drops every cached program-resource reflection entry (CPU-side strings/vectors
+    // only, no Vulkan handles). Called at EGL teardown next to the renderer reset;
+    // safe because GL calls are serialized in this codebase, and any still-live
+    // program rebuilds its entry from the retained generated SPIR-V on demand.
+    void ClearProgramResourceCaches();
+
     void ClearBufferfi(GLenum buffer, GLint drawbuffer, GLfloat depth, GLint stencil);
     void ClearBufferfv(GLenum buffer, GLint drawbuffer, const GLfloat* value);
     void ClearBufferuiv(GLenum buffer, GLint drawbuffer, const GLuint* value);
@@ -117,6 +123,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     // only while a live renderer exists whose device can actually time.
     Bool IsTimerQuerySupported();
     BackendQueryHandle BeginTimeElapsedQuery();
+    BackendQueryHandle BeginXfbPrimitivesQuery(Bool generated);
+    void EndXfbPrimitivesQuery(BackendQueryHandle query);
+    BackendQueryHandle BeginOcclusionQuery();
+    void EndOcclusionQuery(BackendQueryHandle query);
     void EndTimeElapsedQuery(BackendQueryHandle query);
     BackendQueryHandle QueryCounterTimestamp();
     Bool IsQueryResultAvailable(BackendQueryHandle query);

MobileGL/MG_Backend/DirectVulkan/Renderer/FrameContext.cpp

@@ -16,18 +16,19 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_device = device;
         m_commandPool = commandPool;
 
-        Vector<VkCommandBuffer> commandBuffers(frameCount, VK_NULL_HANDLE);
+        Vector<VkCommandBuffer> commandBuffers(frameCount * 2, VK_NULL_HANDLE);
         VkCommandBufferAllocateInfo allocInfo{};
         allocInfo.sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_ALLOCATE_INFO;
         allocInfo.commandPool = commandPool;
         allocInfo.level = VK_COMMAND_BUFFER_LEVEL_PRIMARY;
-        allocInfo.commandBufferCount = frameCount;
+        allocInfo.commandBufferCount = frameCount * 2;
         VkResult result = vkAllocateCommandBuffers(device, &allocInfo, commandBuffers.data());
         if (result != VK_SUCCESS) {
             return result;
         }
         for (Uint32 i = 0; i < frameCount; ++i) {
             m_frames[i].commandBuffer = commandBuffers[i];
+            m_frames[i].preCommandBuffer = commandBuffers[frameCount + i];
         }
 
         VkSemaphoreCreateInfo semaphoreInfo{VK_STRUCTURE_TYPE_SEMAPHORE_CREATE_INFO};
@@ -47,9 +48,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
     void FrameContext::Destroy(VkDevice device, VkCommandPool commandPool) {
         const Uint32 frameCount = static_cast<Uint32>(m_frames.size());
-        Vector<VkCommandBuffer> commandBuffers(frameCount, VK_NULL_HANDLE);
+        Vector<VkCommandBuffer> commandBuffers(frameCount * 2, VK_NULL_HANDLE);
         for (Uint32 i = 0; i < frameCount; ++i) {
             commandBuffers[i] = m_frames[i].commandBuffer;
+            commandBuffers[frameCount + i] = m_frames[i].preCommandBuffer;
         }
 
         for (Uint32 i = 0; i < frameCount; ++i) {
@@ -60,7 +62,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             for (auto& frame : m_frames) {
                 FreeRetiredCommandBuffers(frame);
             }
-            vkFreeCommandBuffers(device, commandPool, frameCount, commandBuffers.data());
+            vkFreeCommandBuffers(device, commandPool, frameCount * 2, commandBuffers.data());
         }
         m_frames.clear();
         currentFrameIndex = 0;
@@ -87,6 +89,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         currentFrameIndex = (currentFrameIndex + 1) % static_cast<Uint32>(m_frames.size());
         GetCurrent().isCommandRecording = false;
         GetCurrent().hasCommandBufferRecorded = false;
+        GetCurrent().isPreCommandRecording = false;
+        GetCurrent().hasPreCommandBufferRecorded = false;
     }
 
     VkCommandBuffer& FrameContext::BeginCommandRecording(VkCommandBufferUsageFlags flags,
@@ -118,6 +122,41 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         frame.hasCommandBufferRecorded = true;
     }
 
+    VkCommandBuffer FrameContext::BeginPreCommandRecording() {
+        auto& frame = GetCurrent();
+        if (frame.isPreCommandRecording) {
+            return frame.preCommandBuffer;
+        }
+        MOBILEGL_ASSERT(!frame.hasPreCommandBufferRecorded,
+                        "BeginPreCommandRecording: a recorded pre stream is still awaiting submission");
+        VK_VERIFY(vkResetCommandBuffer(frame.preCommandBuffer, 0), "BeginPreCommandRecording, vkResetCommandBuffer");
+        VkCommandBufferBeginInfo beginInfo{};
+        beginInfo.sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_BEGIN_INFO;
+        VK_VERIFY(vkBeginCommandBuffer(frame.preCommandBuffer, &beginInfo),
+                  "BeginPreCommandRecording, vkBeginCommandBuffer");
+        frame.isPreCommandRecording = true;
+        return frame.preCommandBuffer;
+    }
+
+    void FrameContext::EndPreCommandRecordingIfOpen() {
+        auto& frame = GetCurrent();
+        if (!frame.isPreCommandRecording) {
+            return;
+        }
+        VK_VERIFY(vkEndCommandBuffer(frame.preCommandBuffer), "EndPreCommandRecordingIfOpen, vkEndCommandBuffer");
+        frame.isPreCommandRecording = false;
+        frame.hasPreCommandBufferRecorded = true;
+    }
+
+    void FrameContext::AbandonPreCommandRecording() {
+        auto& frame = GetCurrent();
+        if (frame.isPreCommandRecording) {
+            VK_VERIFY(vkEndCommandBuffer(frame.preCommandBuffer), "AbandonPreCommandRecording, vkEndCommandBuffer");
+        }
+        frame.isPreCommandRecording = false;
+        frame.hasPreCommandBufferRecorded = false;
+    }
+
     VkResult FrameContext::InitializeSwapchainSemaphores(VkDevice device, Uint32 swapchainImageCount) {
         DestroySwapchainSemaphores(device);
         if (swapchainImageCount == 0) {
@@ -150,12 +189,30 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
     Bool FrameContext::TransitionToPresent(VkImage image, VkImageLayout oldLayout, VkImageLayout presentLayout) {
         auto& frame = GetCurrent();
-        if (frame.hasCommandBufferRecorded || frame.isCommandRecording || oldLayout == presentLayout ||
-            oldLayout == VK_IMAGE_LAYOUT_SHARED_PRESENT_KHR) {
+        if (oldLayout == presentLayout || oldLayout == VK_IMAGE_LAYOUT_SHARED_PRESENT_KHR) {
             return false;
         }
 
-        auto& commandBuffer = BeginCommandRecording();
+        // The barrier belongs in the frame's own recording. Bailing out because
+        // something was already recorded (the previous behaviour) dropped the
+        // transition entirely for every frame that never ran a default-framebuffer
+        // render pass - the only other thing that carries the image to
+        // PRESENT_SRC_KHR, via that pass's finalLayout - so the swapchain image was
+        // handed to the WSI still in the layout it was acquired in.
+        // A closed-but-unsubmitted buffer can only come from a submit that already
+        // failed (SubmitPendingCommandBuffer leaves the flag set on error), and
+        // appending to it is illegal while reopening would reset the frame's own
+        // commands away. The device is gone on that path anyway - stay silent-safe
+        // rather than trade a lost device for a barrier into a closed buffer.
+        if (frame.hasCommandBufferRecorded) {
+            MGLOG_E("TransitionToPresent: command buffer already closed; skipping the present barrier");
+            return false;
+        }
+
+        // Reopening a recording here would vkResetCommandBuffer this frame's own
+        // commands away, so append to the open one and let the caller close it.
+        const Bool openedRecording = !frame.isCommandRecording;
+        VkCommandBuffer commandBuffer = openedRecording ? BeginCommandRecording() : frame.commandBuffer;
 
         VkImageMemoryBarrier presentBarrier{};
         presentBarrier.sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER;
@@ -174,7 +231,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         vkCmdPipelineBarrier(commandBuffer, VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT, VK_PIPELINE_STAGE_BOTTOM_OF_PIPE_BIT, 0, 0,
                              nullptr, 0, nullptr, 1, &presentBarrier);
 
-        EndCommandRecording();
+        if (openedRecording) {
+            EndCommandRecording();
+        }
         return true;
     }
 
@@ -182,17 +241,27 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                                                Uint32 swapchainImageIndex) const {
         const auto& frame = GetCurrent();
         MOBILEGL_ASSERT(!frame.isCommandRecording, "GetSubmitInfo called while command buffer recording is still active");
+        MOBILEGL_ASSERT(!frame.isPreCommandRecording,
+                        "GetSubmitInfo called while the pre-pass stream is still recording");
         AssertValidSwapchainImageIndex(swapchainImageIndex);
         SubmitInfoPacket packet{};
         packet.waitSemaphore = frame.imageAvailableSemaphore;
         packet.signalSemaphore = m_swapchainImageRenderFinishedSemaphores[swapchainImageIndex];
-        packet.commandBuffer = frame.commandBuffer;
+
+        Uint32 commandBufferCount = 0;
+        // The pre-pass stream executes strictly before the frame's commands.
+        if (frame.hasPreCommandBufferRecorded) {
+            packet.commandBuffers[commandBufferCount++] = frame.preCommandBuffer;
+        }
+        if (shouldSubmitCommandBuffer) {
+            packet.commandBuffers[commandBufferCount++] = frame.commandBuffer;
+        }
 
         packet.submitInfo.waitSemaphoreCount = frame.imageAvailableSemaphoreConsumed ? 0U : 1U;
         packet.submitInfo.pWaitSemaphores = frame.imageAvailableSemaphoreConsumed ? nullptr : &packet.waitSemaphore;
         packet.submitInfo.pWaitDstStageMask = frame.imageAvailableSemaphoreConsumed ? nullptr : &packet.waitDstStageMask;
-        packet.submitInfo.commandBufferCount = shouldSubmitCommandBuffer ? 1U : 0U;
-        packet.submitInfo.pCommandBuffers = shouldSubmitCommandBuffer ? &packet.commandBuffer : nullptr;
+        packet.submitInfo.commandBufferCount = commandBufferCount;
+        packet.submitInfo.pCommandBuffers = commandBufferCount > 0 ? packet.commandBuffers : nullptr;
         packet.submitInfo.signalSemaphoreCount = 1;
         packet.submitInfo.pSignalSemaphores = &packet.signalSemaphore;
         return packet;
@@ -227,12 +296,21 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         result = vkAcquireNextImageKHR(device, swapchain, timeout, frame.imageAvailableSemaphore, acquireFence,
                                        &outImageIndex);
-        if (result != VK_SUCCESS) {
+        // VK_SUBOPTIMAL_KHR is a success code: an image *was* acquired and
+        // imageAvailableSemaphore *will* be signaled. Bailing out on it skipped both
+        // the consumed-flag reset (leaving a stale "already consumed", so the next
+        // submit never waited on the pending signal) and the fence reset (leaving
+        // the slot's fence signaled for the next submit to reuse). Only a genuine
+        // failure - VK_ERROR_OUT_OF_DATE_KHR and friends, where nothing is acquired
+        // and nothing is signaled - skips the bookkeeping.
+        if (result != VK_SUCCESS && result != VK_SUBOPTIMAL_KHR) {
             return result;
         }
 
         frame.imageAvailableSemaphoreConsumed = false;
-        return vkResetFences(device, 1, &frame.imageInFlightFence);
+        const VkResult resetResult = vkResetFences(device, 1, &frame.imageInFlightFence);
+        // Hand the acquire's own code back so the caller can schedule a rebuild.
+        return resetResult == VK_SUCCESS ? result : resetResult;
     }
 
     Uint32 FrameContext::GetCurrentFrameIndex() const {
@@ -247,12 +325,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_recordingObserver = observer;
     }
 
-    VkResult FrameContext::RetireCurrentCommandBuffer() {
+    VkResult FrameContext::RetireCurrentCommandBuffer(Bool retirePreCommandBuffer) {
         MOBILEGL_ASSERT(m_device != VK_NULL_HANDLE && m_commandPool != VK_NULL_HANDLE,
                         "RetireCurrentCommandBuffer requires an initialized FrameContext");
         auto& frame = GetCurrent();
         MOBILEGL_ASSERT(!frame.isCommandRecording,
                         "RetireCurrentCommandBuffer called while the command buffer is still recording");
+        MOBILEGL_ASSERT(!frame.isPreCommandRecording,
+                        "RetireCurrentCommandBuffer called while the pre-pass stream is still recording");
 
         VkCommandBufferAllocateInfo allocInfo{};
         allocInfo.sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_ALLOCATE_INFO;
@@ -260,11 +340,23 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         allocInfo.level = VK_COMMAND_BUFFER_LEVEL_PRIMARY;
         allocInfo.commandBufferCount = 1;
         VkCommandBuffer replacement = VK_NULL_HANDLE;
-        const VkResult result = vkAllocateCommandBuffers(m_device, &allocInfo, &replacement);
+        VkResult result = vkAllocateCommandBuffers(m_device, &allocInfo, &replacement);
         if (result != VK_SUCCESS) {
             return result;
         }
-        frame.retiredCommandBuffers.push_back(frame.commandBuffer);
+        if (retirePreCommandBuffer) {
+            VkCommandBuffer preReplacement = VK_NULL_HANDLE;
+            result = vkAllocateCommandBuffers(m_device, &allocInfo, &preReplacement);
+            if (result != VK_SUCCESS) {
+                vkFreeCommandBuffers(m_device, m_commandPool, 1, &replacement);
+                return result;
+            }
+            frame.retiredCommandBuffers.push_back({frame.preCommandBuffer, frame.lastSubmitIndex});
+            frame.preCommandBuffer = preReplacement;
+        }
+        // lastSubmitIndex was just written by the renderer for the submission
+        // that carried this command buffer.
+        frame.retiredCommandBuffers.push_back({frame.commandBuffer, frame.lastSubmitIndex});
         frame.commandBuffer = replacement;
         return VK_SUCCESS;
     }
@@ -274,12 +366,40 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return;
         }
         if (m_device != VK_NULL_HANDLE && m_commandPool != VK_NULL_HANDLE) {
-            vkFreeCommandBuffers(m_device, m_commandPool, static_cast<Uint32>(frame.retiredCommandBuffers.size()),
-                                 frame.retiredCommandBuffers.data());
+            for (const auto& retired : frame.retiredCommandBuffers) {
+                vkFreeCommandBuffers(m_device, m_commandPool, 1, &retired.commandBuffer);
+            }
         }
         frame.retiredCommandBuffers.clear();
     }
 
+    void FrameContext::FreeRetiredCommandBuffersCompletedUpTo(Uint64 completedSubmitIndex) {
+        if (m_device == VK_NULL_HANDLE || m_commandPool == VK_NULL_HANDLE) {
+            return;
+        }
+        for (auto& frame : m_frames) {
+            // Retired buffers are appended in submit order, so the completed
+            // ones form a prefix.
+            SizeT completedCount = 0;
+            while (completedCount < frame.retiredCommandBuffers.size() &&
+                   frame.retiredCommandBuffers[completedCount].submitIndex <= completedSubmitIndex) {
+                vkFreeCommandBuffers(m_device, m_commandPool, 1,
+                                     &frame.retiredCommandBuffers[completedCount].commandBuffer);
+                ++completedCount;
+            }
+            if (completedCount > 0) {
+                frame.retiredCommandBuffers.erase(frame.retiredCommandBuffers.begin(),
+                                                  frame.retiredCommandBuffers.begin() + completedCount);
+            }
+        }
+    }
+
+    void FrameContext::FreeAllRetiredCommandBuffers() {
+        for (auto& frame : m_frames) {
+            FreeRetiredCommandBuffers(frame);
+        }
+    }
+
     void FrameContext::AssertValidFrameIndex(Uint32 frameIndex) const {
         MOBILEGL_ASSERT(frameIndex < m_frames.size(), "FrameContext index out of range");
     }

MobileGL/MG_Backend/DirectVulkan/Renderer/FrameContext.h

@@ -29,7 +29,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             VkPipelineStageFlags waitDstStageMask = VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT;
             VkSemaphore waitSemaphore = VK_NULL_HANDLE;
             VkSemaphore signalSemaphore = VK_NULL_HANDLE;
-            VkCommandBuffer commandBuffer = VK_NULL_HANDLE;
+            // [0] = pre-pass command buffer (when recorded), then the frame
+            // command buffer; submitInfo.pCommandBuffers points here.
+            VkCommandBuffer commandBuffers[2] = {VK_NULL_HANDLE, VK_NULL_HANDLE};
             VkSubmitInfo submitInfo{VK_STRUCTURE_TYPE_SUBMIT_INFO};
         };
 
@@ -40,17 +42,35 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             VkPresentInfoKHR presentInfo{VK_STRUCTURE_TYPE_PRESENT_INFO_KHR};
         };
 
+        // A command buffer submitted mid-frame (FlushPendingCommands), tagged
+        // with the submit-tracker index it was submitted under so it can be
+        // freed as soon as that submission is observed complete - without
+        // waiting for the slot's fence to be waited again (present-less flush
+        // loops never wait it).
+        struct RetiredCommandBuffer {
+            VkCommandBuffer commandBuffer = VK_NULL_HANDLE;
+            Uint64 submitIndex = 0;
+        };
+
         struct FrameData {
             VkCommandBuffer commandBuffer = VK_NULL_HANDLE;
+            // Pre-pass work stream: out-of-pass commands (deferred clear
+            // materialization, sampled-layout transitions) for resources the
+            // frame's recording has not touched yet. Submitted immediately
+            // BEFORE commandBuffer in the same vkQueueSubmit, so recording
+            // into it never has to split the frame's active render pass.
+            VkCommandBuffer preCommandBuffer = VK_NULL_HANDLE;
             VkSemaphore imageAvailableSemaphore = VK_NULL_HANDLE;
             VkFence imageInFlightFence = VK_NULL_HANDLE;
             Bool isCommandRecording = false;
             Bool hasCommandBufferRecorded = false;
+            Bool isPreCommandRecording = false;
+            Bool hasPreCommandBufferRecorded = false;
             Bool imageAvailableSemaphoreConsumed = false;
-            // Command buffers submitted mid-frame (FlushPendingCommands) whose
-            // execution is only known complete once this slot's fence has been
-            // waited again; freed at that point.
-            Vector<VkCommandBuffer> retiredCommandBuffers;
+            // Command buffers submitted mid-frame (FlushPendingCommands),
+            // appended in submit order; freed once their submission is known
+            // complete (fence wait or completion poll).
+            Vector<RetiredCommandBuffer> retiredCommandBuffers;
             // Submit-tracker index of this slot's most recent queue submission
             // (written by the renderer at submit time).
             Uint64 lastSubmitIndex = 0;
@@ -67,6 +87,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         VkCommandBuffer& BeginCommandRecording(VkCommandBufferUsageFlags flags = 0,
                                                const VkCommandBufferInheritanceInfo* pInheritanceInfo = nullptr);
         void EndCommandRecording();
+        // Lazily opens the pre-pass work stream (see FrameData::preCommandBuffer).
+        VkCommandBuffer BeginPreCommandRecording();
+        // Closes the pre stream if open, marking it for submission ahead of the
+        // frame command buffer. Safe to call when it never opened.
+        void EndPreCommandRecordingIfOpen();
+        // Drops an in-progress or recorded-but-unsubmitted pre stream (dropped
+        // frame recordings, swapchain recreation).
+        void AbandonPreCommandRecording();
         VkResult InitializeSwapchainSemaphores(VkDevice device, Uint32 swapchainImageCount);
         void DestroySwapchainSemaphores(VkDevice device);
         Bool TransitionToPresent(VkImage image, VkImageLayout oldLayout,
@@ -79,8 +107,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // Parks the current (already ended and submitted) command buffer on the
         // slot's retired list and installs a freshly allocated one, so recording
         // can restart while the submitted buffer is still executing. Retired
-        // buffers are freed after the slot's fence is next waited.
-        VkResult RetireCurrentCommandBuffer();
+        // buffers are freed after the slot's fence is next waited, or as soon
+        // as their submission is observed complete.
+        VkResult RetireCurrentCommandBuffer(Bool retirePreCommandBuffer = false);
+
+        // Frees every retired command buffer whose tagged submission index is
+        // known complete. Driven by the renderer's submit tracker on completion
+        // events (fence waits and non-blocking polls), so present-less flush
+        // loops reclaim their buffers without any extra wait.
+        void FreeRetiredCommandBuffersCompletedUpTo(Uint64 completedSubmitIndex);
+        // Frees every slot's retired command buffers. Only valid when the
+        // caller has proven every queue submission complete.
+        void FreeAllRetiredCommandBuffers();
 
         Uint32 GetCurrentFrameIndex() const;
         Uint32 GetFrameCount() const;

MobileGL/MG_Backend/DirectVulkan/Renderer/PipelineFactory.cpp

@@ -8,6 +8,7 @@
 
 #include "PipelineFactory.h"
 
+#include <algorithm>
 
 namespace MobileGL::MG_Backend::DirectVulkan {
     static const char* PrimitiveTopologyToString(VkPrimitiveTopology topology) {
@@ -204,6 +205,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         XXHASH_VERIFY(XXH64_update(m_hashState, &payload.topology, sizeof(payload.topology)));
         XXHASH_VERIFY(
             XXH64_update(m_hashState, &payload.primitiveRestartEnable, sizeof(payload.primitiveRestartEnable)));
+        XXHASH_VERIFY(XXH64_update(m_hashState, &payload.patchControlPoints, sizeof(payload.patchControlPoints)));
         XXHASH_VERIFY(XXH64_update(m_hashState, &payload.polygonMode, sizeof(payload.polygonMode)));
         XXHASH_VERIFY(XXH64_update(m_hashState, &payload.cullMode, sizeof(payload.cullMode)));
         XXHASH_VERIFY(XXH64_update(m_hashState, &payload.frontFace, sizeof(payload.frontFace)));
@@ -243,23 +245,108 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         const HashType hash = ComputeHash(payload);
         auto it = m_cache.find(hash);
         if (it != m_cache.end()) {
-            return it->second;
+            it->second.lastUsedFrame = m_frameCounter;
+            return it->second.pipeline;
         }
 
         VkPipeline pipeline = CreatePipeline(payload);
-        m_cache.emplace(hash, pipeline);
+        m_cache.emplace(hash, PipelineCacheEntry{pipeline, payload.programHash, payload.renderPass,
+                                                 m_frameCounter});
         return pipeline;
     }
 
     void PipelineFactory::DestroyAll() {
         for (auto& pair : m_cache) {
-            if (pair.second != VK_NULL_HANDLE) {
-                vkDestroyPipeline(m_device, pair.second, nullptr);
+            if (pair.second.pipeline != VK_NULL_HANDLE) {
+                vkDestroyPipeline(m_device, pair.second.pipeline, nullptr);
             }
         }
         m_cache.clear();
     }
 
+    Uint32 PipelineFactory::OnFrameBoundary() {
+        ++m_frameCounter;
+
+        // Sweep cadence and retire age mirror VkRenderPassManager::OnPresent: an entry
+        // idle for more than kRetireAgeFrames frame boundaries cannot be referenced by
+        // any in-flight command buffer (frames-in-flight <= MOBILEGL_MAGMA_FRAMESINFLIGHT),
+        // so immediate vkDestroyPipeline is safe. The caller must drop its "last
+        // pipeline" memo when this returns non-zero: the memo can return a cached
+        // handle without touching this cache, so an evicted pipeline may still be
+        // memoized (present-less flush loops never reset the memo per frame).
+        constexpr Uint64 kSweepInterval = 256;
+        constexpr Uint64 kRetireAgeFrames = 1024;
+        if ((m_frameCounter % kSweepInterval) != 0) {
+            return 0;
+        }
+
+        Uint32 evicted = 0;
+        for (auto it = m_cache.begin(); it != m_cache.end();) {
+            if (m_frameCounter - it->second.lastUsedFrame > kRetireAgeFrames) {
+                if (it->second.pipeline != VK_NULL_HANDLE) {
+                    vkDestroyPipeline(m_device, it->second.pipeline, nullptr);
+                }
+                it = m_cache.erase(it);
+                ++evicted;
+            } else {
+                ++it;
+            }
+        }
+        if (evicted > 0) {
+            MGLOG_D("PipelineFactory::OnFrameBoundary: evicted %u idle pipelines (%zu remain)", evicted,
+                    m_cache.size());
+        }
+        return evicted;
+    }
+
+    Uint32 PipelineFactory::EvictByRenderPasses(const Vector<VkRenderPass>& renderPasses) {
+        if (renderPasses.empty() || m_cache.empty()) {
+            return 0;
+        }
+        // Sorted-batch membership test keeps a mass eviction (shader-pack switch,
+        // dimension exit) at one O(cache * log batch) scan instead of one full scan
+        // per dying pass.
+        Vector<VkRenderPass> sortedPasses = renderPasses;
+        std::sort(sortedPasses.begin(), sortedPasses.end());
+        Uint32 evicted = 0;
+        for (auto it = m_cache.begin(); it != m_cache.end();) {
+            if (std::binary_search(sortedPasses.begin(), sortedPasses.end(), it->second.renderPass)) {
+                if (it->second.pipeline != VK_NULL_HANDLE) {
+                    vkDestroyPipeline(m_device, it->second.pipeline, nullptr);
+                }
+                it = m_cache.erase(it);
+                ++evicted;
+            } else {
+                ++it;
+            }
+        }
+        if (evicted > 0) {
+            MGLOG_D("PipelineFactory::EvictByRenderPasses: evicted %u pipelines for %zu destroyed render passes",
+                    evicted, sortedPasses.size());
+        }
+        return evicted;
+    }
+
+    Uint32 PipelineFactory::EvictByProgramHash(HashType programHash) {
+        Uint32 evicted = 0;
+        for (auto it = m_cache.begin(); it != m_cache.end();) {
+            if (it->second.programHash == programHash) {
+                if (it->second.pipeline != VK_NULL_HANDLE) {
+                    vkDestroyPipeline(m_device, it->second.pipeline, nullptr);
+                }
+                it = m_cache.erase(it);
+                ++evicted;
+            } else {
+                ++it;
+            }
+        }
+        if (evicted > 0) {
+            MGLOG_D("PipelineFactory::EvictByProgramHash: evicted %u pipelines for program hash 0x%llx",
+                    evicted, static_cast<unsigned long long>(programHash));
+        }
+        return evicted;
+    }
+
     VkPipeline PipelineFactory::CreatePipeline(const PipelineCreatePayload& payload) const {
         MOBILEGL_ASSERT(payload.stages != nullptr && !payload.stages->empty(), "PipelineFactory: stages are empty");
         MOBILEGL_ASSERT(payload.vertexInputState != nullptr, "PipelineFactory: vertexInputState is null");
@@ -294,6 +381,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         ia.topology = payload.topology;
         ia.primitiveRestartEnable = payload.primitiveRestartEnable ? VK_TRUE : VK_FALSE;
 
+        // Only a patch topology has a tessellation stage to configure; leaving the pointer null
+        // otherwise is what the spec expects.
+        VkPipelineTessellationStateCreateInfo tessellation{VK_STRUCTURE_TYPE_PIPELINE_TESSELLATION_STATE_CREATE_INFO};
+        tessellation.patchControlPoints = payload.patchControlPoints;
+
         VkPipelineViewportStateCreateInfo vpci{VK_STRUCTURE_TYPE_PIPELINE_VIEWPORT_STATE_CREATE_INFO};
         vpci.viewportCount = 1;
         vpci.scissorCount = 1;
@@ -358,6 +450,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         gpi.pStages = payload.stages->data();
         gpi.pVertexInputState = payload.vertexInputState;
         gpi.pInputAssemblyState = &ia;
+        gpi.pTessellationState =
+            payload.topology == VK_PRIMITIVE_TOPOLOGY_PATCH_LIST ? &tessellation : nullptr;
         gpi.pViewportState = &vpci;
         gpi.pRasterizationState = &raster;
         gpi.pMultisampleState = &ms;

MobileGL/MG_Backend/DirectVulkan/Renderer/PipelineFactory.h

@@ -30,6 +30,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             Uint32 subpass = 0;
             VkPrimitiveTopology topology = VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST;
             Bool primitiveRestartEnable = false;
+            // GL_PATCH_VERTICES; only read for a PATCH_LIST topology.
+            Uint32 patchControlPoints = 3;
             VkPolygonMode polygonMode = VK_POLYGON_MODE_FILL;
             VkCullModeFlags cullMode = VK_CULL_MODE_BACK_BIT;
             VkFrontFace frontFace = VK_FRONT_FACE_CLOCKWISE;
@@ -65,6 +67,26 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         VkPipeline GetOrCreatePipeline(const PipelineCreatePayload& payload);
         void DestroyAll();
 
+        // Frame boundary hook: ages the pipeline cache and destroys long-unused entries
+        // (their command buffers retired many frames ago), mirroring
+        // VkRenderPassManager::OnPresent's sweep. Returns the number of pipelines
+        // destroyed so the caller can drop any memoized VkPipeline handle.
+        Uint32 OnFrameBoundary();
+        // Destroys every cached pipeline hashed on one of `renderPasses`. Only safe
+        // when the caller guarantees GPU idleness for them - the render-pass manager
+        // calls this (via the renderer) for passes its own >1024-boundary-idle sweep
+        // just evicted, and a pipeline hashed on those handles is only ever bound by
+        // draws that also hit the render-pass entries. Also closes the handle-recycling
+        // hazard: a recycled VkRenderPass value must never serve a stale pipeline.
+        // Batched: one cache scan regardless of how many passes died in the sweep.
+        // Returns the number destroyed (callers invalidate memos when non-zero).
+        Uint32 EvictByRenderPasses(const Vector<VkRenderPass>& renderPasses);
+        // Destroys every cached pipeline built from the program with content hash
+        // `programHash`. Called from the ProgramFactory eviction path, which proves the
+        // same >1024-boundary idleness (the program's pipelines are only bound by draws
+        // that stamp its factory entry). Returns the number destroyed.
+        Uint32 EvictByProgramHash(HashType programHash);
+
         // Driver quirk: suppress depth writes on accumulation-blended pipelines. Multi-pass
         // depth-equality rendering (a blended prepass writes depth that later passes re-test
         // with an equality-inclusive compare on the re-rasterized geometry) requires
@@ -87,12 +109,26 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         static Bool ShouldSuppressDepthWrite(const PipelineCreatePayload& payload);
 
     private:
+        struct PipelineCacheEntry {
+            VkPipeline pipeline = VK_NULL_HANDLE;
+            // The hashed inputs the eviction paths key on: programHash ties the entry to
+            // its ProgramFactory entry, renderPass records the exact handle the hash
+            // folded in (the hash is one-way, so targeted eviction needs them verbatim).
+            HashType programHash = 0;
+            VkRenderPass renderPass = VK_NULL_HANDLE;
+            // Frame-boundary counter value of the last GetOrCreatePipeline hit; drives
+            // cache eviction (see OnFrameBoundary).
+            Uint64 lastUsedFrame = 0;
+        };
+
         VkPipeline CreatePipeline(const PipelineCreatePayload& payload) const;
 
         VkDevice m_device = VK_NULL_HANDLE;
         const VulkanRendererConfig& m_config;
         VkPipelineCache m_pipelineCache = VK_NULL_HANDLE;
-        UnorderedMap<HashType, VkPipeline> m_cache;
+        UnorderedMap<HashType, PipelineCacheEntry> m_cache;
+        // Monotonic frame-boundary counter (bumped in OnFrameBoundary) for cache aging.
+        Uint64 m_frameCounter = 0;
         static inline XXH64_state_t* m_hashState = XXH64_createState();
         static inline Bool s_suppressBlendedDepthWrite = false;
     };

MobileGL/MG_Backend/DirectVulkan/Renderer/ProgramFactory.cpp

@@ -923,11 +923,403 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             ProgramFactory::CompileOptionFlags m_transformFlags;
         };
 
+        // Decorates the module's captured varyings for VK_EXT_transform_feedback:
+        // user outputs get XfbBuffer/XfbStride/Offset directly; a captured
+        // gl_Position (a gl_PerVertex member) is mirrored into a dedicated output
+        // variable copied before every OpReturn, BEFORE the position fixup runs,
+        // so the captured value is the shader's own (pre-remap) gl_Position.
+        class XfbCaptureDecoratePass final : public spvtools::opt::Pass {
+        public:
+            struct CapturedVarying {
+                std::string name;
+                Uint32 bufferIndex = 0;
+                Uint32 offsetBytes = 0;
+            };
+            const char* name() const override { return "mobilegl-xfb-capture-decorate"; }
+            XfbCaptureDecoratePass(Vector<CapturedVarying> varyings, Vector<Uint32> strides)
+                : m_varyings(Move(varyings)), m_strides(Move(strides)) {}
+
+            Status Process() override {
+                using namespace spvtools::opt;
+                if (m_varyings.empty()) return Status::SuccessWithoutChange;
+
+                auto entryPointIter = get_module()->entry_points().begin();
+                if (entryPointIter == get_module()->entry_points().end()) return Status::SuccessWithoutChange;
+                spvtools::opt::Instruction* entryPoint = &*entryPointIter;
+                const Uint32 entryFunctionId = entryPoint->GetSingleWordInOperand(1);
+
+                // Name -> result id map from the debug section.
+                std::unordered_map<std::string, Uint32> idsByName;
+                for (auto& debugInst : get_module()->debugs2()) {
+                    if (debugInst.opcode() != spv::Op::OpName) continue;
+                    idsByName[debugInst.GetInOperand(1).AsString()] = debugInst.GetSingleWordInOperand(0);
+                }
+
+                auto* decorationManager = context()->get_decoration_mgr();
+                const auto decorateForXfb = [&](Uint32 targetId, Uint32 bufferIndex, Uint32 offsetBytes) {
+                    const Uint32 stride = bufferIndex < m_strides.size() ? m_strides[bufferIndex] : 0;
+                    decorationManager->AddDecorationVal(targetId, static_cast<Uint32>(spv::Decoration::XfbBuffer),
+                                                        bufferIndex);
+                    decorationManager->AddDecorationVal(targetId, static_cast<Uint32>(spv::Decoration::XfbStride),
+                                                        stride);
+                    decorationManager->AddDecorationVal(targetId, static_cast<Uint32>(spv::Decoration::Offset),
+                                                        offsetBytes);
+                };
+
+                Bool modified = false;
+                Bool needsPositionMirror = false;
+                Uint32 positionBufferIndex = 0;
+                Uint32 positionOffset = 0;
+                for (const auto& varying : m_varyings) {
+                    if (varying.name == "gl_Position") {
+                        needsPositionMirror = true;
+                        positionBufferIndex = varying.bufferIndex;
+                        positionOffset = varying.offsetBytes;
+                        continue;
+                    }
+                    const auto idIt = idsByName.find(varying.name);
+                    if (idIt == idsByName.end()) {
+                        MGLOG_E("XfbCaptureDecoratePass: no SPIR-V variable named '%s'", varying.name.c_str());
+                        continue;
+                    }
+                    decorateForXfb(idIt->second, varying.bufferIndex, varying.offsetBytes);
+                    modified = true;
+                }
+
+                if (needsPositionMirror) {
+                    modified |= MirrorPositionForCapture(entryFunctionId, *entryPoint, positionBufferIndex,
+                                                         positionOffset, decorateForXfb);
+                }
+
+                if (!modified) return Status::SuccessWithoutChange;
+
+                context()->AddCapability(spv::Capability::TransformFeedback);
+                {
+                    auto executionMode = MakeUnique<spvtools::opt::Instruction>(
+                        context(), spv::Op::OpExecutionMode, 0, 0,
+                        std::initializer_list<spvtools::opt::Operand>{
+                            {SPV_OPERAND_TYPE_ID, {entryPoint->GetSingleWordInOperand(1)}},
+                            {SPV_OPERAND_TYPE_EXECUTION_MODE, {static_cast<Uint32>(spv::ExecutionMode::Xfb)}}});
+                    get_module()->AddExecutionMode(Move(executionMode));
+                }
+                context()->InvalidateAnalysesExceptFor(spvtools::opt::IRContext::kAnalysisNone);
+                return Status::SuccessWithChange;
+            }
+
+        private:
+            template <typename DecorateFn>
+            Bool MirrorPositionForCapture(Uint32 entryFunctionId, spvtools::opt::Instruction& entryPoint,
+                                          Uint32 bufferIndex, Uint32 offsetBytes, const DecorateFn& decorateForXfb) {
+                const Uint32 entryPointModel = entryPoint.GetSingleWordInOperand(0);
+                using namespace spvtools::opt;
+                PositionTargetInfo target{};
+                if (!FindPositionTarget(context(), &target)) {
+                    MGLOG_E("XfbCaptureDecoratePass: gl_Position capture requested but no position output found");
+                    return false;
+                }
+                if (!target.isMember) {
+                    // Standalone gl_Position variable: decorate it directly.
+                    decorateForXfb(target.variableId, bufferIndex, offsetBytes);
+                    return true;
+                }
+
+                auto* typeManager = context()->get_type_mgr();
+                const Uint32 mirrorPointerTypeId =
+                    typeManager->FindPointerToType(target.vectorTypeId, spv::StorageClass::Output);
+                if (mirrorPointerTypeId == 0) return false;
+
+                const Uint32 mirrorVariableId = context()->TakeNextId();
+                auto mirrorVariable = MakeUnique<Instruction>(
+                    context(), spv::Op::OpVariable, mirrorPointerTypeId, mirrorVariableId,
+                    std::initializer_list<Operand>{
+                        {SPV_OPERAND_TYPE_STORAGE_CLASS, {static_cast<Uint32>(spv::StorageClass::Output)}}});
+                get_module()->AddGlobalValue(Move(mirrorVariable));
+
+                // A free output location: past every explicitly decorated output.
+                Uint32 mirrorLocation = 0;
+                for (auto& annotation : get_module()->annotations()) {
+                    if (annotation.opcode() != spv::Op::OpDecorate ||
+                        annotation.GetSingleWordInOperand(1) != static_cast<Uint32>(spv::Decoration::Location)) {
+                        continue;
+                    }
+                    mirrorLocation = std::max(mirrorLocation, annotation.GetSingleWordInOperand(2) + 1);
+                }
+                auto* decorationManager = context()->get_decoration_mgr();
+                decorationManager->AddDecorationVal(mirrorVariableId,
+                                                    static_cast<Uint32>(spv::Decoration::Location), mirrorLocation);
+                decorateForXfb(mirrorVariableId, bufferIndex, offsetBytes);
+                entryPoint.AddOperand({SPV_OPERAND_TYPE_ID, {mirrorVariableId}});
+
+                auto* function = context()->GetFunction(entryFunctionId);
+                if (function == nullptr) return false;
+                const auto model = static_cast<spv::ExecutionModel>(entryPointModel);
+                Bool injected = false;
+                for (auto& block : *function) {
+                    for (auto instIter = block.begin(); instIter != block.end(); ++instIter) {
+                        // Geometry stages capture per emitted vertex; other stages at return.
+                        const Bool isInjectionSite =
+                            model == spv::ExecutionModel::Geometry
+                                ? instIter->opcode() == spv::Op::OpEmitVertex
+                                : instIter->opcode() == spv::Op::OpReturn;
+                        if (!isInjectionSite) continue;
+                        InstructionBuilder builder(context(), &*instIter, IRContext::kAnalysisNone);
+                        const Uint32 memberIndexId = builder.GetUintConstantId(target.memberIndex);
+                        auto* access =
+                            builder.AddAccessChain(target.vectorPtrTypeId, target.variableId, {memberIndexId});
+                        if (access == nullptr) return injected;
+                        auto* value = builder.AddLoad(target.vectorTypeId, access->result_id());
+                        if (value == nullptr) return injected;
+                        builder.AddStore(mirrorVariableId, value->result_id());
+                        injected = true;
+                    }
+                }
+                return injected;
+            }
+
+            Vector<CapturedVarying> m_varyings;
+            Vector<Uint32> m_strides;
+        };
+
+        // Adreno 650 (driver 512.502) faults the GPU on an implicit-LOD sample of a full-screen
+        // colour render target: the texture unit's derivative path reads outside the image's
+        // allocation even though the sampler clamps LOD to 0 and the mapping is 1:1. MobileGL's
+        // own default-framebuffer blit shader works around it with textureLod, but an
+        // application's shader (Minecraft's blit.fsh is `texture(InSampler, texCoord)`) cannot be
+        // edited - so rewrite the sample at the SPIR-V level instead.
+        //
+        // The rewrite is only requested for draws whose every sampler binding is clamped to one
+        // mip level, where explicit LOD 0 is exactly what the implicit form must already produce:
+        // lambda' = clamp(lambda + bias, minLod, maxLod) with minLod = maxLod = 0. Bias and MinLod
+        // operands are therefore dropped rather than translated.
+        class ForceExplicitLod0SamplePass final : public spvtools::opt::Pass {
+        public:
+            const char* name() const override { return "force-explicit-lod0-sample"; }
+
+            Status Process() override {
+                Bool isFragment = false;
+                for (auto& entryPoint : get_module()->entry_points()) {
+                    if (entryPoint.opcode() != spv::Op::OpEntryPoint) continue;
+                    if (static_cast<spv::ExecutionModel>(entryPoint.GetSingleWordInOperand(0)) ==
+                        spv::ExecutionModel::Fragment) {
+                        isFragment = true;
+                        break;
+                    }
+                }
+                if (!isFragment) return Status::SuccessWithoutChange;
+
+                // Plan first, mutate second. Materializing the LOD constant is itself a module
+                // change, so it must not happen unless at least one rewrite is going to follow -
+                // otherwise the pass would grow the binary while reporting SuccessWithoutChange.
+                Vector<RewritePlan> plans;
+                for (auto& function : *get_module()) {
+                    for (auto& block : function) {
+                        for (auto& inst : block) {
+                            RewritePlan plan{};
+                            if (PlanRewrite(&inst, plan)) plans.push_back(Move(plan));
+                        }
+                    }
+                }
+                if (plans.empty()) return Status::SuccessWithoutChange;
+
+                const Uint32 zeroId = GetFloatZeroId();
+                if (zeroId == 0) return Status::SuccessWithoutChange;
+
+                for (auto& plan : plans) {
+                    plan.operands.push_back({SPV_OPERAND_TYPE_ID, {zeroId}});
+                    for (auto& operand : plan.trailingOperands) {
+                        plan.operands.push_back(operand);
+                    }
+                    plan.instruction->SetOpcode(plan.opcode);
+                    plan.instruction->SetInOperands(Move(plan.operands));
+                }
+                // Opcodes and operand lists changed underneath every cached analysis.
+                context()->InvalidateAnalysesExceptFor(spvtools::opt::IRContext::kAnalysisNone);
+                return Status::SuccessWithChange;
+            }
+
+        private:
+            struct RewritePlan {
+                spvtools::opt::Instruction* instruction = nullptr;
+                spv::Op opcode = spv::Op::OpNop;
+                // Everything up to and including the Image Operands mask; the Lod id and the
+                // trailing operand values are appended once the constant exists.
+                Vector<spvtools::opt::Operand> operands;
+                Vector<spvtools::opt::Operand> trailingOperands;
+            };
+
+            // Image Operands bits that may accompany an implicit-LOD sample, in the canonical
+            // ascending order SPIR-V requires the operand values to appear in.
+            static constexpr Uint32 kBias = 0x1;
+            static constexpr Uint32 kLod = 0x2;
+            static constexpr Uint32 kGrad = 0x4;
+            static constexpr Uint32 kConstOffset = 0x8;
+            static constexpr Uint32 kOffset = 0x10;
+            static constexpr Uint32 kConstOffsets = 0x20;
+            static constexpr Uint32 kSample = 0x40;
+            static constexpr Uint32 kMinLod = 0x80;
+            static constexpr Uint32 kKnownMask = 0xFF;
+
+            Uint32 GetFloatZeroId() {
+                // Reuse a 32-bit float type already in the module; a shader that samples always has
+                // one, and looking it up avoids depending on type-creation API details.
+                Uint32 floatTypeId = 0;
+                for (auto& inst : get_module()->types_values()) {
+                    if (inst.opcode() == spv::Op::OpTypeFloat && inst.NumInOperands() >= 1 &&
+                        inst.GetSingleWordInOperand(0) == 32) {
+                        floatTypeId = inst.result_id();
+                        break;
+                    }
+                }
+                if (floatTypeId == 0) return 0;
+
+                const auto* floatType = context()->get_type_mgr()->GetType(floatTypeId);
+                if (floatType == nullptr) return 0;
+                const auto zeroBits = std::bit_cast<Uint32>(0.0f);
+                const auto* zeroConst = context()->get_constant_mgr()->GetConstant(floatType, {zeroBits});
+                if (zeroConst == nullptr) return 0;
+                auto* zeroInst = context()->get_constant_mgr()->GetDefiningInstruction(zeroConst);
+                return zeroInst != nullptr ? zeroInst->result_id() : 0;
+            }
+
+            static Bool MapOpcode(spv::Op op, spv::Op& outOpcode, Uint32& outFixedOperandCount) {
+                switch (op) {
+                case spv::Op::OpImageSampleImplicitLod:
+                    outOpcode = spv::Op::OpImageSampleExplicitLod;
+                    outFixedOperandCount = 2; // sampled image, coordinate
+                    return true;
+                case spv::Op::OpImageSampleProjImplicitLod:
+                    outOpcode = spv::Op::OpImageSampleProjExplicitLod;
+                    outFixedOperandCount = 2;
+                    return true;
+                case spv::Op::OpImageSampleDrefImplicitLod:
+                    outOpcode = spv::Op::OpImageSampleDrefExplicitLod;
+                    outFixedOperandCount = 3; // sampled image, coordinate, Dref
+                    return true;
+                case spv::Op::OpImageSampleProjDrefImplicitLod:
+                    outOpcode = spv::Op::OpImageSampleProjDrefExplicitLod;
+                    outFixedOperandCount = 3;
+                    return true;
+                default:
+                    return false;
+                }
+            }
+
+            static Bool PlanRewrite(spvtools::opt::Instruction* inst, RewritePlan& outPlan) {
+                spv::Op newOpcode = spv::Op::OpNop;
+                Uint32 fixedCount = 0;
+                if (!MapOpcode(inst->opcode(), newOpcode, fixedCount)) return false;
+                if (inst->NumInOperands() < fixedCount) return false;
+
+                Uint32 mask = 0;
+                Uint32 next = fixedCount;
+                if (inst->NumInOperands() > fixedCount) {
+                    mask = inst->GetSingleWordInOperand(fixedCount);
+                    next = fixedCount + 1;
+                }
+                // An operand this pass does not model would be silently reordered or dropped, and
+                // Grad cannot legally accompany an implicit-LOD sample: leave such an instruction be.
+                if ((mask & ~kKnownMask) != 0 || (mask & kGrad) != 0) return false;
+
+                Vector<spvtools::opt::Operand> fixedOperands;
+                fixedOperands.reserve(fixedCount + 1);
+                for (Uint32 i = 0; i < fixedCount; ++i) {
+                    fixedOperands.push_back(inst->GetInOperand(i));
+                }
+
+                // Collect the surviving operand values in the same ascending-bit order they were
+                // encoded in, so the rebuilt list stays canonical.
+                Uint32 keptMask = kLod;
+                Vector<spvtools::opt::Operand> keptOperands;
+                static constexpr Uint32 kOrderedBits[] = {kBias,   kLod,          kGrad,   kConstOffset,
+                                                          kOffset, kConstOffsets, kSample, kMinLod};
+                for (const Uint32 bit : kOrderedBits) {
+                    if ((mask & bit) == 0) continue;
+                    if (next >= inst->NumInOperands()) return false;
+                    const spvtools::opt::Operand value = inst->GetInOperand(next++);
+                    // Bias and MinLod only shift a lambda that is already clamped to 0, and any
+                    // original Lod is replaced by the constant the caller appends.
+                    if (bit == kBias || bit == kMinLod || bit == kLod) continue;
+                    keptMask |= bit;
+                    keptOperands.push_back(value);
+                }
+
+                fixedOperands.push_back({SPV_OPERAND_TYPE_IMAGE, {keptMask}});
+                outPlan.instruction = inst;
+                outPlan.opcode = newOpcode;
+                outPlan.operands = Move(fixedOperands);
+                outPlan.trailingOperands = Move(keptOperands);
+                return true;
+            }
+        };
+
+        spvtools::Optimizer::PassToken CreateForceExplicitLod0SamplePass() {
+            return spvtools::Optimizer::PassToken(MakeUnique<ForceExplicitLod0SamplePass>());
+        }
+
+        Bool TransformSpirvForExplicitLod0Sampling(const Vector<Uint>& input, Vector<Uint>& output) {
+            if (input.empty()) {
+                output.clear();
+                return true;
+            }
+            spvtools::Optimizer optimizer(SPV_ENV_VULKAN_1_3);
+            spvtools::OptimizerOptions options;
+            // Matches the position-fix pass: this build of spirv-tools asserts rather than
+            // reporting, so validation stays off in the shipping path.
+            options.set_run_validator(false);
+            optimizer.SetMessageConsumer([](spv_message_level_t, const char*, const spv_position_t&,
+                                            const char* message) {
+                MGLOG_E("Vulkan: explicit-LOD0 pass: %s", message != nullptr ? message : "");
+            });
+            optimizer.RegisterPass(CreateForceExplicitLod0SamplePass());
+
+            const Bool success = optimizer.Run(input.data(), input.size(), &output, options);
+            if (!success) {
+                MGLOG_E("Vulkan: explicit-LOD0 sampling pass failed; keeping the original module");
+                output = input;
+            }
+            return success;
+        }
+
         spvtools::Optimizer::PassToken CreateGlToVulkanPositionFixPass(
             ProgramFactory::CompileOptionFlags transformFlags) {
             return spvtools::Optimizer::PassToken(MakeUnique<GlToVulkanPositionFixPass>(transformFlags));
         }
 
+        Bool TransformSpirvForXfbCapture(const Vector<Uint>& input, Vector<Uint>& output,
+                                         const MG_State::GLState::ProgramObject& program) {
+            if (input.empty()) {
+                output.clear();
+                return true;
+            }
+            Vector<XfbCaptureDecoratePass::CapturedVarying> varyings;
+            varyings.reserve(program.GetTransformFeedbackVaryingCount());
+            for (const auto& varying : program.GetTransformFeedbackVaryings()) {
+                varyings.push_back({varying.name, varying.bufferIndex, varying.offsetBytes});
+            }
+            Vector<Uint32> strides;
+            strides.reserve(program.GetTransformFeedbackBufferCount());
+            for (SizeT i = 0; i < program.GetTransformFeedbackBufferCount(); ++i) {
+                strides.push_back(program.GetTransformFeedbackStride(static_cast<Uint32>(i)));
+            }
+
+            spvtools::Optimizer optimizer(SPV_ENV_VULKAN_1_3);
+            spvtools::OptimizerOptions options;
+            options.set_run_validator(false);
+            optimizer.SetMessageConsumer([](spv_message_level_t, const char*, const spv_position_t&,
+                                            const char* message) {
+                MGLOG_E("Vulkan: xfb capture pass: %s", message != nullptr ? message : "");
+            });
+            optimizer.RegisterPass(spvtools::Optimizer::PassToken(
+                MakeUnique<XfbCaptureDecoratePass>(Move(varyings), Move(strides))));
+
+            const Bool success = optimizer.Run(input.data(), input.size(), &output, options);
+            if (!success) {
+                MGLOG_E("Vulkan: xfb capture decoration pass failed; keeping the original module");
+                output = input;
+            }
+            return success;
+        }
+
         Bool TransformSpirvForVulkanPositionFix(const Vector<Uint>& input, Vector<Uint>& output,
                                                 ProgramFactory::CompileOptionFlags transformFlags) {
             if (input.empty()) {
@@ -1369,6 +1761,25 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             XXHASH_VERIFY(XXH64_update(m_hashState, &binding, sizeof(binding)));
         }
 
+        // The transform feedback capture layout is baked into the modules by
+        // XfbCaptureDecoratePass rather than coming from the SPIR-V, so it has to be part of
+        // the key: two programs can share every shader and still capture differently, which
+        // is exactly what changing the buffer mode does (glTransformFeedbackVaryings with the
+        // same varyings but GL_SEPARATE_ATTRIBS instead of GL_INTERLEAVED_ATTRIBS). Only
+        // hashed for a capturing compile, so nothing else changes key.
+        if (flags & CompileOptionBit::XfbCapture) {
+            for (const auto& varying : program.GetTransformFeedbackVaryings()) {
+                XXHASH_VERIFY(XXH64_update(m_hashState, varying.name.data(), varying.name.size()));
+                XXHASH_VERIFY(XXH64_update(m_hashState, &varying.bufferIndex, sizeof(varying.bufferIndex)));
+                XXHASH_VERIFY(XXH64_update(m_hashState, &varying.offsetBytes, sizeof(varying.offsetBytes)));
+            }
+            const SizeT bufferCount = program.GetTransformFeedbackBufferCount();
+            for (SizeT i = 0; i < bufferCount; ++i) {
+                const Uint32 stride = program.GetTransformFeedbackStride(static_cast<Uint32>(i));
+                XXHASH_VERIFY(XXH64_update(m_hashState, &stride, sizeof(stride)));
+            }
+        }
+
         HashType hash = XXH64_digest(m_hashState);
         return hash;
     }
@@ -1950,11 +2361,17 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
         auto it = m_cache.find(hash);
         if (it != m_cache.end()) {
+            // Every draw/dispatch funnels through this lookup (the renderer memos only
+            // skip re-hashing, never the factory lookup), so an actively-used entry is
+            // stamped at least once per frame boundary and can never be aged out while
+            // any in-flight command buffer still references it.
+            it->second.lastUsedFrame = m_frameCounter;
             return it->second;
         }
 
         auto& entry = m_cache[hash];
         entry.hash = hash;
+        entry.lastUsedFrame = m_frameCounter;
         auto& shaders = program.GetAttachedShaders();
         auto& spirv = program.GetGeneratedSpirv();
         Vector<Vector<Uint>> moduleSpirvs(spirv.size());
@@ -1967,11 +2384,40 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
             // Apply position fixup if needed
             if (fixupStage != ShaderStage::Unknown && shaders[i] && shaders[i]->GetShaderStage() == fixupStage) {
-                TransformSpirvForVulkanPositionFix(spv, moduleSpirvs[i], flags);
+                const Vector<Uint>* fixupInput = &spv;
+                Vector<Uint> xfbSpirv;
+                if ((flags & ProgramFactory::CompileOptionBit::XfbCapture) &&
+                    program.GetTransformFeedbackVaryingCount() > 0) {
+                    // Decorate BEFORE the position fixup so a captured gl_Position
+                    // mirror copies the shader's own (pre-remap) value.
+                    if (TransformSpirvForXfbCapture(spv, xfbSpirv, program)) {
+                        fixupInput = &xfbSpirv;
+                    }
+                }
+                TransformSpirvForVulkanPositionFix(*fixupInput, moduleSpirvs[i], flags);
             } else {
                 moduleSpirvs[i] = spv;
             }
 
+            if ((flags & ProgramFactory::CompileOptionBit::ExplicitLod0Sampling) && shaders[i] &&
+                shaders[i]->GetShaderStage() == ShaderStage::Fragment) {
+                Vector<Uint> explicitLodSpirv;
+                if (TransformSpirvForExplicitLod0Sampling(moduleSpirvs[i], explicitLodSpirv)) {
+                    moduleSpirvs[i] = Move(explicitLodSpirv);
+                }
+            }
+
+            // Vulkan's SPIR-V environment has no rectangle image dimension, so a
+            // GL_TEXTURE_RECTANGLE lookup has to become the 2D one the texture is really
+            // stored as - which addresses [0,1] where the application addressed texels.
+            {
+                Vector<Uint> rectLoweredSpirv;
+                if (MG_Util::ShaderTranspiler::ShaderCompiler::LowerRectImages(moduleSpirvs[i], rectLoweredSpirv) &&
+                    !rectLoweredSpirv.empty()) {
+                    moduleSpirvs[i] = Move(rectLoweredSpirv);
+                }
+            }
+
             // GL apps depend on cross-program position invariance for multi-pass equality
             // depth tests (MC 26.3's OIT re-draws the cloud geometry with GEQUAL against the
             // depth its own first pass wrote); decorate Position outputs Invariant so
@@ -2068,4 +2514,42 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         return entry;
     }
+
+    void ProgramFactory::OnFrameBoundary() {
+        ++m_frameCounter;
+
+        // Sweep cadence and retire age mirror VkRenderPassManager::OnPresent: an entry
+        // idle for more than kRetireAgeFrames frame boundaries cannot be referenced by
+        // any in-flight command buffer (frames-in-flight <= MOBILEGL_MAGMA_FRAMESINFLIGHT),
+        // so its shader modules and layouts are destroyed immediately - no deferred-
+        // destroy machinery needed. Eviction is content-based, never tied to
+        // glDeleteProgram: the cache is content-hash-shared across GL programs, so a
+        // delete-driven erase could free an entry another live program still resolves.
+        // An evicted entry self-heals - the frontend program keeps its generated
+        // SPIR-V, so the next GetOrCreateProgram rebuilds it (this also covers the
+        // renderer's internal blit/depth-mipmap programs).
+        constexpr Uint64 kSweepInterval = 256;
+        constexpr Uint64 kRetireAgeFrames = 1024;
+        if ((m_frameCounter % kSweepInterval) != 0) {
+            return;
+        }
+
+        for (auto it = m_cache.begin(); it != m_cache.end();) {
+            if (m_frameCounter - it->second.lastUsedFrame > kRetireAgeFrames) {
+                const HashType hash = it->first;
+                const VkDescriptorSetLayout descriptorSetLayout = it->second.descriptorSetLayout;
+                MGLOG_D("ProgramFactory::OnFrameBoundary: evicting idle program entry hash=0x%llx",
+                        static_cast<unsigned long long>(hash));
+                // erase runs ~VkProgramObject (modules/layouts destroyed); notify after
+                // so an observer never observes a half-destroyed entry through a lookup.
+                // Observers only need the handle values to purge their keyed caches.
+                it = m_cache.erase(it);
+                if (m_evictionObserver != nullptr) {
+                    m_evictionObserver->OnProgramEvicted(hash, descriptorSetLayout);
+                }
+            } else {
+                ++it;
+            }
+        }
+    }
 } // namespace MobileGL::MG_Backend::DirectVulkan

MobileGL/MG_Backend/DirectVulkan/Renderer/ProgramFactory.h

@@ -42,6 +42,16 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             SurfaceRotate90 = 1 << 2,
             SurfaceRotate180 = 1 << 3,
             SurfaceRotate270 = 1 << 4,
+            // Rewrites the fragment stage's implicit-LOD image samples to explicit LOD 0.
+            // Only ever set for a draw whose every sampler binding is clamped to a single mip
+            // level, which makes the two forms produce identical texels (the implicit lambda is
+            // clamped into [minLod, maxLod] = [0, 0] regardless of derivatives or bias).
+            ExplicitLod0Sampling = 1 << 5,
+            // Decorates the last vertex-processing stage's captured varyings with
+            // XfbBuffer/XfbStride/Offset (VK_EXT_transform_feedback). Set only for draws
+            // recorded while GL transform feedback is active, so plain draws keep the
+            // undecorated variant.
+            XfbCapture = 1 << 6,
         };
         using CompileOptionFlags = Flags<CompileOptionBit>;
         using HashType = Uint64;
@@ -88,6 +98,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // gl_FragDepth); shader-computed depth is immune to the cross-pipeline
             // position-invariance quirk (see PipelineFactory::ShouldSuppressDepthWrite).
             Bool fragmentReplacesDepth = false;
+            // Frame-boundary counter value of the last GetOrCreateProgram hit; drives
+            // cache eviction (see OnFrameBoundary).
+            Uint64 lastUsedFrame = 0;
 
             static inline VkDevice s_device = VK_NULL_HANDLE;
 
@@ -124,6 +137,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 producerOutputComponentCount = other.producerOutputComponentCount;
                 fragmentInputComponentCount = other.fragmentInputComponentCount;
                 fragmentReplacesDepth = other.fragmentReplacesDepth;
+                lastUsedFrame = other.lastUsedFrame;
                 other.hash = 0;
                 other.descriptorSetLayout = VK_NULL_HANDLE;
                 other.pipelineLayout = VK_NULL_HANDLE;
@@ -135,6 +149,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 other.producerOutputComponentCount = 0;
                 other.fragmentInputComponentCount = 0;
                 other.fragmentReplacesDepth = false;
+                other.lastUsedFrame = 0;
             }
             VkProgramObject& operator=(VkProgramObject&& other) noexcept {
                 if (this == &other) {
@@ -170,6 +185,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 producerOutputComponentCount = other.producerOutputComponentCount;
                 fragmentInputComponentCount = other.fragmentInputComponentCount;
                 fragmentReplacesDepth = other.fragmentReplacesDepth;
+                lastUsedFrame = other.lastUsedFrame;
                 other.hash = 0;
                 other.descriptorSetLayout = VK_NULL_HANDLE;
                 other.pipelineLayout = VK_NULL_HANDLE;
@@ -181,6 +197,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 other.producerOutputComponentCount = 0;
                 other.fragmentInputComponentCount = 0;
                 other.fragmentReplacesDepth = false;
+                other.lastUsedFrame = 0;
                 return *this;
             }
 
@@ -210,6 +227,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             }
         };
 
+        // Notified when the OnFrameBoundary sweep destroys an aged-out cache entry,
+        // carrying the entry's content hash and the VkDescriptorSetLayout it owned.
+        // Dependent caches (compute pipelines, PipelineFactory entries, UniformManager's
+        // per-layout descriptor sets) must purge in the same step: after vkDestroy the
+        // layout handle value may be recycled for an unrelated layout, and the program
+        // hash may be re-inserted by a later rebuild of the same content.
+        class IEvictionObserver {
+        public:
+            virtual ~IEvictionObserver() = default;
+            virtual void OnProgramEvicted(HashType programHash, VkDescriptorSetLayout descriptorSetLayout) = 0;
+        };
+
         explicit ProgramFactory(VkDevice device, const VulkanRendererConfig& config, Uint32 maxBindings = 16,
                                 Bool shaderDrawParametersEnabled = false,
                                 Bool unformattedFloatStorageImagesEnabled = false)
@@ -225,6 +254,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         const VkProgramObject& GetOrCreateProgram(
             const MG_State::GLState::ProgramObject& program, CompileOptionFlags flags);
 
+        // Observer may be null (no notifications). Not owned.
+        void SetEvictionObserver(IEvictionObserver* observer) { m_evictionObserver = observer; }
+        // Frame boundary hook: ages the program cache and evicts long-unused entries
+        // (their command buffers retired many frames ago), mirroring
+        // VkRenderPassManager::OnPresent's sweep.
+        void OnFrameBoundary();
+
         static VkShaderStageFlagBits ToVkStage(ShaderStage stage);
         static VkFormat ConvertSpirvImageFormatToVkFormat(SpvImageFormat format);
         static SamplerNumericDomain UniformTypeToSamplerNumericDomain(GLenum glType);
@@ -266,6 +302,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // shaderStorageImageReadWithoutFormat and shaderStorageImageWriteWithoutFormat.
         Bool m_unformattedFloatStorageImagesEnabled = false;
         mutable ProgramLookupCache m_lastLookup;
+        // Monotonic frame-boundary counter (bumped in OnFrameBoundary) for cache aging.
+        Uint64 m_frameCounter = 0;
+        IEvictionObserver* m_evictionObserver = nullptr;
         static inline XXH64_state_t* m_hashState = XXH64_createState();
     };
 } // namespace MobileGL::MG_Backend::DirectVulkan

MobileGL/MG_Backend/DirectVulkan/Renderer/SwapchainObject.cpp

@@ -247,6 +247,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         m_surfaceFormat = {createInfo.imageFormat, createInfo.imageColorSpace};
         m_extent = createInfo.imageExtent;
+        // The surface-space extent this swapchain was built from, i.e. before the
+        // quarter-turn swap above. Out-of-date checks must compare in THIS space: comparing a
+        // freshly queried currentExtent against the swapped m_extent flips axes every rotation
+        // and makes the comparison alternate forever.
+        m_surfaceExtent = defaultFramebufferExtent;
         m_preTransform = createInfo.preTransform;
 
         VK_VERIFY(vkCreateSwapchainKHR(device, &createInfo, nullptr, &m_swapchain));
@@ -257,6 +262,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_images.resize(imageCount, VK_NULL_HANDLE);
         VK_VERIFY(vkGetSwapchainImagesKHR(device, m_swapchain, &imageCount, m_images.data()));
         m_imageLayouts.assign(imageCount, VK_IMAGE_LAYOUT_UNDEFINED);
+        // Fresh swapchain images hold garbage until a render pass stores into them.
+        m_imageContentDefined.assign(imageCount, false);
+        m_depthStencilContentDefined.assign(imageCount, false);
 
         CreateImageViews(device);
         CreateDepthStencilResources(device, physicalDevice);
@@ -428,9 +436,39 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         m_images.clear();
         m_imageLayouts.clear();
+        m_imageContentDefined.clear();
+        m_depthStencilContentDefined.clear();
         m_preTransform = VK_SURFACE_TRANSFORM_IDENTITY_BIT_KHR;
     }
 
+    Bool SwapchainObject::IsImageContentDefined(Uint32 index) const {
+        MOBILEGL_ASSERT(index < m_imageContentDefined.size(), "Swapchain image content index out of range");
+        return m_imageContentDefined[index];
+    }
+
+    void SwapchainObject::SetImageContentDefined(Uint32 index, Bool defined) {
+        MOBILEGL_ASSERT(index < m_imageContentDefined.size(), "Swapchain image content index out of range");
+        m_imageContentDefined[index] = defined;
+    }
+
+    Bool SwapchainObject::IsDepthStencilContentDefined(Uint32 index) const {
+        MOBILEGL_ASSERT(index < m_depthStencilContentDefined.size(),
+                        "Swapchain depth/stencil content index out of range");
+        return m_depthStencilContentDefined[index];
+    }
+
+    void SwapchainObject::SetDepthStencilContentDefined(Uint32 index, Bool defined) {
+        MOBILEGL_ASSERT(index < m_depthStencilContentDefined.size(),
+                        "Swapchain depth/stencil content index out of range");
+        m_depthStencilContentDefined[index] = defined;
+    }
+
+    void SwapchainObject::SetAllDepthStencilContentUndefined() {
+        for (SizeT i = 0; i < m_depthStencilContentDefined.size(); ++i) {
+            m_depthStencilContentDefined[i] = false;
+        }
+    }
+
     VkImage SwapchainObject::GetImage(Uint32 index) const {
         MOBILEGL_ASSERT(index < m_images.size(), "Swapchain image index out of range");
         return m_images[index];

MobileGL/MG_Backend/DirectVulkan/Renderer/SwapchainObject.h

@@ -35,6 +35,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         VkSwapchainKHR GetHandle() const { return m_swapchain; }
         const VkSurfaceFormatKHR& GetSurfaceFormat() const { return m_surfaceFormat; }
         VkExtent2D GetExtent() const { return m_extent; }
+        // Surface-space extent (before the pre-rotation quarter-turn swap) this swapchain was
+        // created from - the value to compare a freshly queried currentExtent against.
+        VkExtent2D GetSurfaceExtent() const { return m_surfaceExtent; }
         VkSurfaceTransformFlagBitsKHR GetPreTransform() const { return m_preTransform; }
         const Vector<VkImage>& GetImages() const { return m_images; }
         const Vector<VkImageView>& GetImageViews() const { return m_imageViews; }
@@ -49,6 +52,21 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         void SetImageLayout(Uint32 index, VkImageLayout layout);
         SizeT GetImageCount() const { return m_images.size(); }
 
+        // EGL content-validity tracking for the default framebuffer. A color
+        // buffer's content is undefined once its image has been presented
+        // (EGL_BUFFER_DESTROYED swap behaviour, the implementation default),
+        // and every ancillary (depth/stencil) buffer's content is undefined
+        // after ANY swap regardless of swap behaviour (EGL 1.5 §3.10.1). The
+        // render-pass manager turns an undefined attachment's tile load into
+        // LOAD_OP_DONT_CARE. Flags start false (a fresh swapchain image holds
+        // garbage) and a render pass storing into an attachment sets it back
+        // to defined.
+        Bool IsImageContentDefined(Uint32 index) const;
+        void SetImageContentDefined(Uint32 index, Bool defined);
+        Bool IsDepthStencilContentDefined(Uint32 index) const;
+        void SetDepthStencilContentDefined(Uint32 index, Bool defined);
+        void SetAllDepthStencilContentUndefined();
+
     private:
         void CreateImageViews(VkDevice device);
         void CreateDepthStencilResources(VkDevice device, VkPhysicalDevice physicalDevice);
@@ -63,6 +81,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         VkSwapchainKHR m_swapchain = VK_NULL_HANDLE;
         VkSurfaceFormatKHR m_surfaceFormat{};
         VkExtent2D m_extent{};
+        VkExtent2D m_surfaceExtent{};
         VkSurfaceTransformFlagBitsKHR m_preTransform = VK_SURFACE_TRANSFORM_IDENTITY_BIT_KHR;
         Vector<VkImage> m_images;
         Vector<VkImageView> m_imageViews;
@@ -73,5 +92,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         Vector<VkDeviceMemory> m_depthStencilImageMemories;
         Vector<VkImageView> m_depthStencilImageViews;
         Vector<VkImageLayout> m_depthStencilImageLayouts;
+        Vector<Bool> m_imageContentDefined;
+        Vector<Bool> m_depthStencilContentDefined;
     };
 } // namespace MobileGL::MG_Backend::DirectVulkan

MobileGL/MG_Backend/DirectVulkan/Renderer/UniformManager.cpp

@@ -18,6 +18,7 @@
 #include "MG_Util/Converters/MGToVk/TextureEnumConverter.h"
 #include "MG_Util/Metrics/TextureMetrics.h"
 #include <Config.h>
+#include <algorithm>
 #include <cstdio>
 #include <cstdlib>
 #include <cstring>
@@ -204,6 +205,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // The frame's descriptor sets are recycled above, so last frame's reuse target
         // is gone: start the per-draw descriptor-reuse cache fresh this frame.
         m_hasLastDescriptor = false;
+        m_lastBindValid = false;
         // Re-fingerprint the bound sampler set fresh this frame so any GL object address
         // reuse cannot outlive a single frame (see SamplerResolveMemo).
         for (auto& memo : m_samplerResolveMemo) {
@@ -211,6 +213,40 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
     }
 
+    void UniformManager::OnDescriptorSetLayoutDestroyed(VkDescriptorSetLayout descriptorSetLayout) {
+        SizeT purgedSets = 0;
+        for (auto& frame : m_frames) {
+            const auto it = frame.descriptorSetCacheByLayout.find(descriptorSetLayout);
+            if (it == frame.descriptorSetCacheByLayout.end()) {
+                continue;
+            }
+            // Free the sets back to their pools and credit the bucket accounting, so
+            // program churn recycles pool capacity instead of abandoning the slots.
+            // GPU-safe: the layout only dies after >1024 idle frame boundaries, so no
+            // in-flight command buffer references these sets.
+            for (const auto& cached : it->second.sets) {
+                if (cached.set == VK_NULL_HANDLE) {
+                    continue;
+                }
+                vkFreeDescriptorSets(m_device, cached.pool, 1, &cached.set);
+                const auto bucket = std::find_if(
+                    frame.descriptorPools.begin(), frame.descriptorPools.end(),
+                    [&cached](const DescriptorPoolBucket& candidate) { return candidate.handle == cached.pool; });
+                if (bucket != frame.descriptorPools.end() && bucket->allocatedSets > 0) {
+                    --bucket->allocatedSets;
+                }
+            }
+            purgedSets += it->second.sets.size();
+            frame.descriptorSetCacheByLayout.erase(it);
+        }
+        if (purgedSets > 0) {
+            // The per-draw reuse memo folds the layout handle into its signature; drop
+            // it so a recycled handle value cannot revive a purged set mid-frame.
+            m_hasLastDescriptor = false;
+            MGLOG_D("UniformDescriptorBinder: freed %zu descriptor sets for destroyed layout", purgedSets);
+        }
+    }
+
     Bool UniformManager::ResolveSamplerDescriptor(VkCommandBuffer commandBuffer,
                                                             const MG_State::GLState::ProgramObject& program,
                                                             const ProgramFactory::VkProgramObject& programObj,
@@ -230,12 +266,24 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         const auto& samplerOverride = textureUnit.GetSamplerObject();
         const auto preferredTarget = programObj.samplerTextureTargetByBinding[binding];
         SharedPtr<MG_State::GLState::ITextureObject> fallbackHolder;
+        // A texture that fails the completeness rules for the filter in effect reads
+        // (0, 0, 0, 1), which is exactly what the fallback texture holds - so it takes the
+        // same route as a sampler with nothing bound.
+        if (texture != nullptr &&
+            MG_State::GLState::SamplesAsIncompleteTexture(
+                texture, samplerOverride ? samplerOverride.get() : texture->GetSamplerObject().get())) {
+            texture = nullptr;
+        }
         if (texture == nullptr) {
             fallbackHolder = GetFallbackTexture(preferredTarget);
             texture = fallbackHolder.get();
-            MOBILEGL_ASSERT(texture != nullptr,
-                            "ResolveSamplerDescriptor: no fallback texture available for binding=%u location=%d unit=%d target=%d",
-                            binding, location, unit, static_cast<Int>(preferredTarget));
+            if (texture == nullptr) {
+                MGLOG_E("ResolveSamplerDescriptor: no fallback texture available for binding=%u ('%s') "
+                        "location=%d unit=%d target=%d",
+                        binding, programObj.samplerNameByBinding[binding].c_str(), location, unit,
+                        static_cast<Int>(preferredTarget));
+                return false;
+            }
             MGLOG_W(
                 "ResolveSamplerDescriptor: using fallback texture for unbound sampler binding=%u ('%s') location=%d unit=%d target=%d",
                 binding, programObj.samplerNameByBinding[binding].c_str(), location, unit,
@@ -350,24 +398,28 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             const Uint16 samplerVersion = samplerToUse->GetVersion();
             const Uint64 textureLifetimeId = texture->GetLifetimeId();
             const Uint16 textureParamsVersion = texture->GetTextureParamsVersion();
+            // The sampler's LOD clamp depends on how many levels the sampled view exposes, and that
+            // follows uploads as well as GL parameters - so it belongs in the memo key too.
+            const Uint32 viewLevelCount = resource->sampledLevelCount;
             if (memo.valid && memo.samplerLifetimeId == samplerLifetimeId && memo.samplerVersion == samplerVersion &&
                 memo.textureLifetimeId == textureLifetimeId && memo.textureParamsVersion == textureParamsVersion &&
-                memo.forceNearestFiltering == forceNearestFiltering) {
+                memo.forceNearestFiltering == forceNearestFiltering && memo.viewLevelCount == viewLevelCount) {
                 resolvedSampler = memo.sampler;
             } else {
-                resolvedSampler =
-                    m_samplerManager->GetOrCreateSampler(*samplerToUse, *texture, forceNearestFiltering);
+                resolvedSampler = m_samplerManager->GetOrCreateSampler(*samplerToUse, *texture,
+                                                                       forceNearestFiltering, viewLevelCount);
                 memo.samplerLifetimeId = samplerLifetimeId;
                 memo.samplerVersion = samplerVersion;
                 memo.textureLifetimeId = textureLifetimeId;
                 memo.textureParamsVersion = textureParamsVersion;
                 memo.forceNearestFiltering = forceNearestFiltering;
+                memo.viewLevelCount = viewLevelCount;
                 memo.sampler = resolvedSampler;
                 memo.valid = true;
             }
         } else {
-            resolvedSampler =
-                m_samplerManager->GetOrCreateSampler(*samplerToUse, *texture, forceNearestFiltering);
+            resolvedSampler = m_samplerManager->GetOrCreateSampler(*samplerToUse, *texture, forceNearestFiltering,
+                                                                   resource->sampledLevelCount);
         }
         outImageInfo = {
             .sampler = resolvedSampler,
@@ -408,6 +460,48 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         return outImageInfo.sampler != VK_NULL_HANDLE;
     }
 
+    Bool UniformManager::ProgramSamplesOnlySingleLevelTextures(
+        const MG_State::GLState::ProgramObject& program, const ProgramFactory::VkProgramObject& programObj) {
+        Bool sawSampler = false;
+        for (Uint32 binding = 0; binding < programObj.bindingKinds.size(); ++binding) {
+            if (programObj.bindingKinds[binding] != ProgramFactory::DescriptorBindingKind::CombinedImageSampler) {
+                continue;
+            }
+            const auto* texture = ResolveSamplerTextureRaw(program, programObj, binding);
+            if (texture == nullptr) return false;
+            const auto& levelRange = texture->GetLevelRange();
+            if (levelRange.x() != levelRange.y()) return false;
+
+            // An explicit-LOD sample is a single filtered tap, so it also gives up anisotropic
+            // filtering - which a single-level view can still have. Resolve the sampler exactly
+            // the way ResolveSamplerDescriptor does and bail if anisotropy would apply.
+            const Int location = programObj.samplerUniformLocationByBinding[binding];
+            const Int unit = ResolveSamplerUnitIndex(program, location, binding);
+            const auto& samplerOverride = MG_State::pGLContext->GetTextureUnitObject(unit).GetSamplerObject();
+            const auto* effectiveSampler =
+                samplerOverride ? samplerOverride.get() : texture->GetSamplerObject().get();
+            if (effectiveSampler == nullptr) return false;
+            if (effectiveSampler->GetMaxAnisotropy() > 1.0f &&
+                effectiveSampler->GetMinFilter() == SamplerFilterMode::Linear &&
+                effectiveSampler->GetMagFilter() == SamplerFilterMode::Linear) {
+                return false;
+            }
+
+            // An explicit LOD 0 makes lambda exactly 0, which is the magnification side of the
+            // min/mag decision. That only matches the implicit form when lambda could not have been
+            // positive anyway (the LOD clamp already pins it at or below 0), or when the two
+            // filters are the same and the choice cannot be observed.
+            const Float effectiveMaxLod = effectiveSampler->GetMipmapMode() == SamplerMipmapMode::None
+                                              ? 0.0f
+                                              : effectiveSampler->GetMaxLod();
+            if (effectiveMaxLod > 0.0f && effectiveSampler->GetMinFilter() != effectiveSampler->GetMagFilter()) {
+                return false;
+            }
+            sawSampler = true;
+        }
+        return sawSampler;
+    }
+
     Bool UniformManager::ResolveSamplerTexture(const MG_State::GLState::ProgramObject& program,
                                                          const ProgramFactory::VkProgramObject& programObj, Uint32 binding,
                                                          SharedPtr<MG_State::GLState::ITextureObject>& outTexture) {
@@ -510,7 +604,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         const VkDeviceSize texelSize =
             static_cast<VkDeviceSize>(MG_Util::GetSizedInternalFormatSizeInBytes(internalFormat));
-        VkDeviceSize viewRange = slice.size;
+        // glTextureBufferRange addresses a window of the buffer, not all of it; the whole-buffer
+        // forms report the buffer's current size here, so both go through the same clamp.
+        const VkDeviceSize rangeOffset = static_cast<VkDeviceSize>(textureBuffer->GetBufferRangeOffset());
+        const VkDeviceSize rangeSize = static_cast<VkDeviceSize>(textureBuffer->GetBufferRangeSizeInBytes());
+        VkDeviceSize viewRange = std::min(rangeSize, slice.size > rangeOffset ? slice.size - rangeOffset : 0);
         if (texelSize > 0) {
             viewRange = (viewRange / texelSize) * texelSize;
         }
@@ -523,7 +621,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         viewInfo.sType = VK_STRUCTURE_TYPE_BUFFER_VIEW_CREATE_INFO;
         viewInfo.buffer = slice.buffer;
         viewInfo.format = vkFormat;
-        viewInfo.offset = slice.offset;
+        viewInfo.offset = slice.offset + rangeOffset;
         viewInfo.range = viewRange;
 
         VkBufferView bufferView = VK_NULL_HANDLE;
@@ -568,6 +666,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return false;
         }
 
+        // The shader may write this buffer, and those writes land in GPU memory behind the
+        // frontend's CPU shadow - which is what MapBuffer and GetBufferSubData read.
+        // Host-visible coherent GPU residency makes the shadow BE that memory, so the
+        // results are visible without a readback path, exactly as for a capture buffer.
+        bufferObject->EnsureGpuResidentStorage();
+        // ... and the read that follows has to wait for this draw or dispatch to retire.
+        bufferObject->MarkGpuWritten();
+
         BufferSlice slice{};
         if (!m_bufferManager->AcquireResidentSlice(BufferKind::ShaderStorage, bufferObject, slice) || !slice.IsValid()) {
             MGLOG_E("ResolveStorageBufferDescriptor: failed to sync GL buffer %u for block '%s'",
@@ -670,15 +776,27 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     }
 
     SharedPtr<MG_State::GLState::ITextureObject> UniformManager::GetFallbackTexture(TextureTarget target) const {
-        MOBILEGL_ASSERT(target == TextureTarget::Texture2D || target == TextureTarget::TextureRectangle,
-                        "UniformManager::GetFallbackTexture: unsupported fallback target=%d",
-                        static_cast<Int>(target));
+        // The fallback is a single-sampled 2D image, so it can only stand in for a sampler that
+        // would accept one. A multisample sampler in particular cannot: its descriptor demands a
+        // multisample view, and handing it this one is invalid Vulkan, not a degraded picture.
+        // Report that there is no fallback and let the caller decline the draw - aborting the
+        // process over an unbound sampler is never the right answer.
+        if (target != TextureTarget::Texture2D && target != TextureTarget::TextureRectangle) {
+            MGLOG_E("UniformManager::GetFallbackTexture: no fallback exists for target=%d",
+                    static_cast<Int>(target));
+            return nullptr;
+        }
 
         if (m_fallbackTexture2D == nullptr) {
             auto fallbackTexture = MakeShared<MG_State::GLState::TextureObject2D>(kFallbackTexture2DExternalIndex);
             fallbackTexture->SetInternalFormat(TextureInternalFormat::RGBA8);
             fallbackTexture->AllocateStorage(TextureUploadTarget::Texture2D, 0,
                                              {.texelSize = {1, 1, 1}, .byteSize = 4});
+            // (0, 0, 0, 1): what GL reads from a texture that is not complete, and the only
+            // sensible answer for a sampler with nothing bound.
+            static Uint8 kOpaqueBlackTexel[4] = {0, 0, 0, 255};
+            fallbackTexture->UpdateMipmapSubData(TextureUploadTarget::Texture2D, 0,
+                                                 {kOpaqueBlackTexel, sizeof(kOpaqueBlackTexel)});
             fallbackTexture->MarkStorageDirty(TextureUploadTarget::Texture2D, 0, true);
             m_fallbackTexture2D = fallbackTexture;
         }
@@ -911,6 +1029,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         VkDescriptorPoolCreateInfo poolInfo{};
         poolInfo.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_POOL_CREATE_INFO;
+        // FREE_DESCRIPTOR_SET_BIT lets a destroyed layout's cached sets be freed back
+        // (OnDescriptorSetLayoutDestroyed) so program churn recycles pool capacity.
+        // The cost is on set allocation only, which happens when a layout's per-frame
+        // cache grows - never on the per-draw reuse path.
+        poolInfo.flags = VK_DESCRIPTOR_POOL_CREATE_FREE_DESCRIPTOR_SET_BIT;
         poolInfo.maxSets = maxSets;
         poolInfo.poolSizeCount = static_cast<Uint32>(std::size(poolSizes));
         poolInfo.pPoolSizes = poolSizes;
@@ -990,7 +1113,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         auto& frame = m_frames[frameIndex];
         auto& cache = frame.descriptorSetCacheByLayout[programObj.descriptorSetLayout];
         if (cache.cursor < cache.sets.size()) {
-            outDescriptorSet = cache.sets[cache.cursor++];
+            outDescriptorSet = cache.sets[cache.cursor++].set;
         } else {
             VkResult allocResult = AllocateDescriptorSetsFromActivePool(frameIndex, programObj, outDescriptorSet);
             if (allocResult == VK_ERROR_OUT_OF_POOL_MEMORY || allocResult == VK_ERROR_FRAGMENTED_POOL) {
@@ -1004,7 +1127,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 return allocResult;
             }
 
-            cache.sets.push_back(outDescriptorSet);
+            // The successful allocation came from the bucket the alloc helper left
+            // active; record it so a layout-destroyed purge can free the set back.
+            cache.sets.push_back({outDescriptorSet, frame.descriptorPools[frame.activeDescriptorPoolIndex].handle});
             ++cache.cursor;
             MGLOG_D("UniformDescriptorBinder: cached descriptor set count for frame=%u grew to %zu", frameIndex,
                     cache.sets.size());
@@ -1102,16 +1227,47 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                         bufferInfo.range = ubo.range;
                         dynOffset = static_cast<Uint32>(ubo.dynamicOffset);
                     } else {
-                        BufferSlice slice{};
-                        if (!m_bufferManager->UploadTransient(BufferKind::Uniform, frameIndex, ubo.payload,
-                                                              ubo.payloadSize, m_minDynamicOffsetAlignment, slice)) {
-                            MOBILEGL_ASSERT(false, "UniformDescriptorBinder::BindProgramUniformBuffers failed: UBO upload failed on binding %u element %u",
-                                    binding, element);
-                            return false;
+                        // Global-UBO slice reuse (see GlobalUboSliceMemo): unchanged
+                        // uniform bytes re-use the slice already uploaded this frame.
+                        const Bool isGlobalUbo =
+                            programObj.globalUboBinding == static_cast<Int>(binding) && element == 0;
+                        const Uint64 uboFrameSerial = m_bufferManager->GetFrameSerial();
+                        const Uint64 uboProgramLifetimeId = program.GetLifetimeId();
+                        const Uint32 uboContentVersion = program.GetUBOContentVersion();
+                        Bool reusedSlice = false;
+                        if (isGlobalUbo) {
+                            for (const auto& memo : m_globalUboMemo) {
+                                if (memo.buffer != VK_NULL_HANDLE &&
+                                    memo.programLifetimeId == uboProgramLifetimeId &&
+                                    memo.frameSerial == uboFrameSerial &&
+                                    memo.uboContentVersion == uboContentVersion &&
+                                    memo.range == static_cast<VkDeviceSize>(ubo.payloadSize)) {
+                                    bufferInfo.buffer = memo.buffer;
+                                    bufferInfo.range = memo.range;
+                                    dynOffset = static_cast<Uint32>(memo.offset);
+                                    reusedSlice = true;
+                                    break;
+                                }
+                            }
+                        }
+                        if (!reusedSlice) {
+                            BufferSlice slice{};
+                            if (!m_bufferManager->UploadTransient(BufferKind::Uniform, frameIndex, ubo.payload,
+                                                                  ubo.payloadSize, m_minDynamicOffsetAlignment, slice)) {
+                                MOBILEGL_ASSERT(false, "UniformDescriptorBinder::BindProgramUniformBuffers failed: UBO upload failed on binding %u element %u",
+                                        binding, element);
+                                return false;
+                            }
+                            bufferInfo.buffer = slice.buffer;
+                            bufferInfo.range = ubo.payloadSize;
+                            dynOffset = static_cast<Uint32>(slice.offset);
+                            if (isGlobalUbo) {
+                                m_globalUboMemo[m_globalUboMemoNext] = GlobalUboSliceMemo{
+                                    uboProgramLifetimeId, uboFrameSerial, uboContentVersion,
+                                    slice.buffer, slice.offset, static_cast<VkDeviceSize>(ubo.payloadSize)};
+                                m_globalUboMemoNext = (m_globalUboMemoNext + 1) % kGlobalUboMemoSize;
+                            }
                         }
-                        bufferInfo.buffer = slice.buffer;
-                        bufferInfo.range = ubo.payloadSize;
-                        dynOffset = static_cast<Uint32>(slice.offset);
                     }
                     bufferInfos.push_back(bufferInfo);
                     // Dynamic offsets are consumed in binding order, then array element order,
@@ -1250,8 +1406,34 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             m_hasLastDescriptor = cacheable;
         }
 
-        vkCmdBindDescriptorSets(commandBuffer, bindPoint, programObj.pipelineLayout, 0, 1,
-                                &descriptorSet, static_cast<Uint32>(dynamicOffsets.size()), dynamicOffsets.data());
+        // Skip the driver call when this exact binding is already live on the
+        // command buffer (see the bind-dedup shadow in the header).
+        const Uint32 offsetCount = static_cast<Uint32>(dynamicOffsets.size());
+        Bool identicalBind = m_lastBindValid && m_lastBindSet == descriptorSet &&
+                             m_lastBindLayout == programObj.pipelineLayout && m_lastBindPoint == bindPoint &&
+                             m_lastBindOffsetCount == offsetCount && offsetCount <= kMaxShadowedDynamicOffsets;
+        if (identicalBind) {
+            for (Uint32 i = 0; i < offsetCount; ++i) {
+                if (m_lastBindOffsets[i] != dynamicOffsets[i]) {
+                    identicalBind = false;
+                    break;
+                }
+            }
+        }
+        if (!identicalBind) {
+            vkCmdBindDescriptorSets(commandBuffer, bindPoint, programObj.pipelineLayout, 0, 1,
+                                    &descriptorSet, offsetCount, dynamicOffsets.data());
+            if (offsetCount <= kMaxShadowedDynamicOffsets) {
+                m_lastBindValid = true;
+                m_lastBindSet = descriptorSet;
+                m_lastBindLayout = programObj.pipelineLayout;
+                m_lastBindPoint = bindPoint;
+                m_lastBindOffsetCount = offsetCount;
+                std::copy_n(dynamicOffsets.data(), offsetCount, m_lastBindOffsets);
+            } else {
+                m_lastBindValid = false;
+            }
+        }
         return true;
     }
 } // namespace MobileGL::MG_Backend::DirectVulkan

MobileGL/MG_Backend/DirectVulkan/Renderer/UniformManager.h

@@ -39,6 +39,20 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         void Shutdown();
 
         void BeginFrame(Uint32 frameIndex);
+        // A command buffer (re)began recording: descriptor bindings recorded into
+        // the previous buffer do not carry over, so drop the bind-dedup shadow.
+        void OnCommandBufferBoundary() { m_lastBindValid = false; }
+        // A ProgramFactory eviction just destroyed this layout: purge every frame
+        // slot's cached descriptor sets for it, so a recycled handle value can never
+        // stale-hit sets written for the dead layout's bindings. The sets are
+        // vkFreeDescriptorSets'd back to their pools (created with
+        // FREE_DESCRIPTOR_SET_BIT) and the pool accounting is credited, so program
+        // churn recycles pool capacity instead of abandoning it. GPU-safe: the layout
+        // only dies after >1024 idle frame boundaries, so no in-flight command buffer
+        // references its sets. This is the only eviction path for the per-layout
+        // caches - a live layout's entry must never be purged (its sets would be
+        // unreachable pool slots), so there is deliberately no age-based sweep here.
+        void OnDescriptorSetLayoutDestroyed(VkDescriptorSetLayout descriptorSetLayout);
         Bool CollectSampledTextures(const MG_State::GLState::ProgramObject& program,
                                     const ProgramFactory::VkProgramObject& programObj,
                                     Vector<MG_State::GLState::ITextureObject*>& outTextures);
@@ -58,6 +72,16 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         static VkFormat ResolveStorageImageViewFormat(VkFormat reflectedFormat, GLenum bindingFormat,
                                                       VkFormat resourceFormat, Bool useBindingFormat);
 
+        // True when the program reads at least one sampler and every one of them is bound to a
+        // texture whose GL level range is a single level. Such a sampler resolves to
+        // minLod = maxLod = 0 (see VkSamplerManager::GetOrCreateSampler), so an implicit-LOD sample
+        // and an explicit LOD 0 sample must read the same texel - which is what makes the
+        // ExplicitLod0Sampling SPIR-V rewrite safe to request. Deliberately conservative: it reads
+        // only GL state, so a texture that ends up single-level for another reason (one uploaded
+        // level under a wide level range) merely misses the rewrite.
+        static Bool ProgramSamplesOnlySingleLevelTextures(const MG_State::GLState::ProgramObject& program,
+                                                          const ProgramFactory::VkProgramObject& programObj);
+
     private:
         struct DescriptorPoolBucket {
             VkDescriptorPool handle = VK_NULL_HANDLE;
@@ -65,8 +89,16 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             Uint32 allocatedSets = 0;
         };
 
+        // A cached descriptor set together with the pool it was allocated from, so a
+        // layout-destroyed purge can vkFreeDescriptorSets it back and credit the
+        // owning bucket's accounting.
+        struct CachedDescriptorSet {
+            VkDescriptorSet set = VK_NULL_HANDLE;
+            VkDescriptorPool pool = VK_NULL_HANDLE;
+        };
+
         struct DescriptorSetCacheEntry {
-            Vector<VkDescriptorSet> sets;
+            Vector<CachedDescriptorSet> sets;
             Uint32 cursor = 0;
         };
 
@@ -156,6 +188,35 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         Uint64 m_lastDescriptorSignature = 0;
         Bool m_hasLastDescriptor = false;
 
+        // vkCmdBindDescriptorSets dedup: consecutive draws with a static uniform
+        // block resolve to the same set AND the same dynamic offsets, so the
+        // driver call can be skipped outright. Command-buffer-scope state; reset
+        // via OnCommandBufferBoundary whenever a recording (re)begins. Keyed on
+        // layout+bind point, so a pipeline-layout switch always rebinds.
+        static constexpr Uint32 kMaxShadowedDynamicOffsets = 8;
+        Bool m_lastBindValid = false;
+        VkDescriptorSet m_lastBindSet = VK_NULL_HANDLE;
+        VkPipelineLayout m_lastBindLayout = VK_NULL_HANDLE;
+        VkPipelineBindPoint m_lastBindPoint = VK_PIPELINE_BIND_POINT_GRAPHICS;
+        Uint32 m_lastBindOffsetCount = 0;
+        Uint32 m_lastBindOffsets[kMaxShadowedDynamicOffsets] = {};
+
+        // Global-UBO transient-slice reuse: MC leaves the default uniform block
+        // untouched across long GUI/terrain runs, so the per-draw re-upload of
+        // the same bytes can reuse the slice uploaded earlier THIS frame (frame
+        // serial guards arena recycling; the content version guards writes).
+        struct GlobalUboSliceMemo {
+            Uint64 programLifetimeId = 0;
+            Uint64 frameSerial = 0;
+            Uint32 uboContentVersion = 0;
+            VkBuffer buffer = VK_NULL_HANDLE;
+            VkDeviceSize offset = 0;
+            VkDeviceSize range = 0;
+        };
+        static constexpr Uint32 kGlobalUboMemoSize = 4;
+        GlobalUboSliceMemo m_globalUboMemo[kGlobalUboMemoSize];
+        Uint32 m_globalUboMemoNext = 0;
+
         // Per-binding fast path over VkSamplerManager's content-hashed sampler cache, which
         // stays the source of truth: its key hashes all sampler+texture state, so two distinct
         // sampler objects with identical state still resolve to one VkSampler. This memo only
@@ -172,6 +233,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             Uint64 samplerLifetimeId = 0;
             Uint64 textureLifetimeId = 0;
             VkSampler sampler = VK_NULL_HANDLE;
+            Uint32 viewLevelCount = 0;
             Uint16 samplerVersion = 0;
             Uint16 textureParamsVersion = 0;
             Bool forceNearestFiltering = false;

MobileGL/MG_Backend/DirectVulkan/Renderer/VertexInputStateFactory.cpp

@@ -32,6 +32,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             XXHASH_VERIFY(XXH64_update(m_hashState, &attr.IsBgra, sizeof(attr.IsBgra)));
             XXHASH_VERIFY(XXH64_update(m_hashState, &attr.Divisor, sizeof(attr.Divisor)));
 
+            // The buffer's heap address is an identity component of the key: a freed
+            // buffer's reused address can alias an old cache entry, but only under a
+            // byte-identical attribute layout - and the entry payload is a pure function
+            // of the hashed inputs, with the draw path re-resolving bindingBufferKeys
+            // against the live VAO attribute pointers, so an aliased hit returns exactly
+            // what a rebuild would. Address drift only grows the map; the OnFrameBoundary
+            // aging sweep bounds that.
             const SizeT bufferKey = reinterpret_cast<SizeT>(attr.Buffer.get());
             XXHASH_VERIFY(XXH64_update(m_hashState, &bufferKey, sizeof(bufferKey)));
         }
@@ -51,14 +58,27 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
     const VertexInputStateFactory::BackendVertexInputState& VertexInputStateFactory::GetOrCreateVertexInputState(
         const MG_State::GLState::VertexArrayObject& vao) {
-        return GetOrCreateVertexInputState(vao, GetOrComputeHash(vao));
+        // Per-draw fast path: the VAO carries a pointer to its resolved entry,
+        // valid while its config version and the cache's eviction epoch both
+        // match - no re-hash, no map lookup.
+        const void* memoState = nullptr;
+        Uint64 memoEpoch = 0;
+        if (vao.GetBackendStateMemo(memoState, memoEpoch) && memoEpoch == m_evictionEpoch) {
+            const auto* entry = static_cast<const BackendVertexInputState*>(memoState);
+            entry->lastUsedFrameBoundary = m_frameBoundaryCounter;
+            return *entry;
+        }
+        const BackendVertexInputState& entry = GetOrCreateVertexInputState(vao, GetOrComputeHash(vao));
+        vao.SetBackendStateMemo(&entry, m_evictionEpoch);
+        return entry;
     }
 
     const VertexInputStateFactory::BackendVertexInputState& VertexInputStateFactory::GetOrCreateVertexInputState(
         const MG_State::GLState::VertexArrayObject& vao, HashType hash) {
         auto it = m_cache.find(hash);
         if (it != m_cache.end()) {
-            return it->second;
+            it->second->lastUsedFrameBoundary = m_frameBoundaryCounter;
+            return *it->second;
         }
 
         VertexInputStateBuilder builder;
@@ -67,6 +87,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         Vector<Uint32> bindingAttributeLocations;
         Vector<Bool> bindingUsesClientMemory;
         Vector<VertexStreamConversion> bindingConversions;
+        Vector<VkVertexInputBindingDivisorDescriptionEXT> bindingDivisors;
         Uint32 unsupportedAttribMask = 0;
 
         for (Uint32 location = 0; location < MG_State::GLState::VertexArrayObject::MAX_VERTEX_ATTRIBS; ++location) {
@@ -160,14 +181,51 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             bindingConversions.push_back(conversion);
             builder.AddBinding(binding, stride, inputRate);
             builder.AddAttribute(location, binding, vkFormat, 0);
+            // Divisor 1 is what VK_VERTEX_INPUT_RATE_INSTANCE already means; only anything
+            // else needs the extension to say it.
+            if (inputRate == VK_VERTEX_INPUT_RATE_INSTANCE && attr.Divisor != 1) {
+                bindingDivisors.push_back({binding, static_cast<Uint32>(attr.Divisor)});
+            }
         }
 
         const auto& state = builder.Build();
 
-        auto& entry = m_cache[hash];
+        auto& slot = m_cache[hash];
+        if (!slot) {
+            slot = MakeUnique<BackendVertexInputState>();
+        }
+        BackendVertexInputState& entry = *slot;
         entry.hash = hash;
+        entry.lastUsedFrameBoundary = m_frameBoundaryCounter;
+        entry.bindingDivisors = Move(bindingDivisors);
         entry.bindings = builder.GetBindings();
         entry.attributes = builder.GetAttributes();
+        // See the layoutHash declaration: hash only the resolved layout, never
+        // buffer identities, so identical layouts across VAOs/buffers agree.
+        XXHASH_VERIFY(XXH64_reset(m_hashState, 0));
+        for (const auto& binding : entry.bindings) {
+            XXHASH_VERIFY(XXH64_update(m_hashState, &binding.binding, sizeof(binding.binding)));
+            XXHASH_VERIFY(XXH64_update(m_hashState, &binding.stride, sizeof(binding.stride)));
+            XXHASH_VERIFY(XXH64_update(m_hashState, &binding.inputRate, sizeof(binding.inputRate)));
+        }
+        for (const auto& attribute : entry.attributes) {
+            XXHASH_VERIFY(XXH64_update(m_hashState, &attribute.location, sizeof(attribute.location)));
+            XXHASH_VERIFY(XXH64_update(m_hashState, &attribute.binding, sizeof(attribute.binding)));
+            XXHASH_VERIFY(XXH64_update(m_hashState, &attribute.format, sizeof(attribute.format)));
+            XXHASH_VERIFY(XXH64_update(m_hashState, &attribute.offset, sizeof(attribute.offset)));
+        }
+        for (const auto& divisor : entry.bindingDivisors) {
+            XXHASH_VERIFY(XXH64_update(m_hashState, &divisor.binding, sizeof(divisor.binding)));
+            XXHASH_VERIFY(XXH64_update(m_hashState, &divisor.divisor, sizeof(divisor.divisor)));
+        }
+        XXHASH_VERIFY(XXH64_update(m_hashState, &unsupportedAttribMask, sizeof(unsupportedAttribMask)));
+        entry.layoutHash = XXH64_digest(m_hashState);
+        entry.attributeLocationMask = 0;
+        for (const auto& attribute : entry.attributes) {
+            if (attribute.location < 32u) {
+                entry.attributeLocationMask |= (1u << attribute.location);
+            }
+        }
         entry.bindingBufferKeys = std::move(bindingBufferKeys);
         entry.bindingBaseOffsets = std::move(bindingBaseOffsets);
         entry.bindingAttributeLocations = std::move(bindingAttributeLocations);
@@ -177,9 +235,43 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         entry.state = state;
         entry.state.pVertexBindingDescriptions = entry.bindings.empty() ? nullptr : entry.bindings.data();
         entry.state.pVertexAttributeDescriptions = entry.attributes.empty() ? nullptr : entry.attributes.data();
+        if (!entry.bindingDivisors.empty()) {
+            entry.divisorState.vertexBindingDivisorCount = static_cast<Uint32>(entry.bindingDivisors.size());
+            entry.divisorState.pVertexBindingDivisors = entry.bindingDivisors.data();
+            entry.state.pNext = &entry.divisorState;
+        } else {
+            entry.state.pNext = nullptr;
+        }
         return entry;
     }
 
+    void VertexInputStateFactory::OnFrameBoundary() {
+        ++m_frameBoundaryCounter;
+
+        // Sweep occasionally; evict entries whose last hit is far in the past.
+        // Erasure happens only here, never mid-frame: the draw path holds a
+        // reference into the current entry across its setup, and unordered_map
+        // erase would invalidate it. Entries are CPU-side only, so no GPU-idle
+        // proof is needed; an evicted entry that is used again is simply rebuilt
+        // from the VAO state (same hash, same content).
+        constexpr Uint64 kSweepInterval = 256;
+        constexpr Uint64 kRetireAgeBoundaries = 1024;
+        if ((m_frameBoundaryCounter % kSweepInterval) != 0) {
+            return;
+        }
+
+        for (auto it = m_cache.begin(); it != m_cache.end();) {
+            if (m_frameBoundaryCounter - it->second->lastUsedFrameBoundary > kRetireAgeBoundaries) {
+                it = m_cache.erase(it);
+                // Invalidate every VAO's state-pointer memo: the erased node's
+                // address may be reused by a future insert.
+                ++m_evictionEpoch;
+            } else {
+                ++it;
+            }
+        }
+    }
+
     VkFormat VertexInputStateFactory::ToVkVertexFormat(DataType type, Int size, Bool normalized, Bool isInteger,
                                                        Bool isBgra) {
         if (isBgra) {

MobileGL/MG_Backend/DirectVulkan/Renderer/VertexInputStateFactory.h

@@ -27,6 +27,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         struct BackendVertexInputState {
             HashType hash = 0;
+            // Hash of the resolved Vulkan vertex layout only (bindings, attributes,
+            // unsupported mask) - NO buffer identities. `hash` mixes buffer heap
+            // addresses so per-chunk VBOs mint a fresh identity per buffer; keying
+            // pipelines on that minted one VkPipeline per chunk section for an
+            // identical layout, defeating pipeline reuse and the per-draw memo.
+            // Pipelines depend only on the layout, so they key on this instead.
+            HashType layoutHash = 0;
+            // Frame boundary of the last cache hit; entries idle past the
+            // OnFrameBoundary retirement age are evicted (CPU heap only).
+            // Mutable: the VAO's state-pointer memo fast path stamps it through
+            // a const entry reference.
+            mutable Uint64 lastUsedFrameBoundary = 0;
             Vector<VkVertexInputBindingDescription> bindings;
             Vector<VkVertexInputAttributeDescription> attributes;
             Vector<SizeT> bindingBufferKeys;
@@ -38,6 +50,17 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // absent from `attributes`, so without this mask the draw path cannot tell them apart from
             // a genuinely disabled array and would silently feed the shader the current attribute value.
             Uint32 unsupportedAttribMask = 0;
+            // Bitmask of `attributes[i].location` - the draw path needs it up to
+            // three times per draw, so it is baked once at build time.
+            Uint32 attributeLocationMask = 0;
+            // Per-binding glVertexAttribDivisor values other than 1. Vulkan's instance input
+            // rate advances once per instance and nothing else, so anything else has to be
+            // stated through VK_EXT_vertex_attribute_divisor. Empty when every instanced
+            // binding uses divisor 1, which is what the plain input rate already means.
+            Vector<VkVertexInputBindingDivisorDescriptionEXT> bindingDivisors;
+            VkPipelineVertexInputDivisorStateCreateInfoEXT divisorState{
+                VK_STRUCTURE_TYPE_PIPELINE_VERTEX_INPUT_DIVISOR_STATE_CREATE_INFO_EXT
+            };
             VkPipelineVertexInputStateCreateInfo state{
                 VK_STRUCTURE_TYPE_PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO
             };
@@ -55,6 +78,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         const BackendVertexInputState& GetOrCreateVertexInputState(
             const MG_State::GLState::VertexArrayObject& vao, HashType hash);
         const BackendVertexInputState& GetOrCreateVertexInputState(const MG_State::GLState::VertexArrayObject& vao);
+        // Frame boundary hook: ages the cache and evicts entries not hit for many
+        // frames. The key mixes buffer heap addresses, so buffer/VAO churn keeps
+        // minting fresh keys; without eviction the map grows for the whole session.
+        // Entries hold no Vulkan handles (pipeline creation copies the descriptions)
+        // and the draw path's entry reference never spans a frame boundary, so
+        // eviction here needs no GPU-idle proof. Self-gated: one counter bump and
+        // compare except on sweep boundaries.
+        void OnFrameBoundary();
         static SizeT GetComponentSize(DataType type);
         // Tightly-packed byte size of one vertex element for this attribute: componentSize * size for
         // normal types, and 4 (one packed word) for the 2_10_10_10 types and GL_BGRA. Returns 0 for
@@ -69,7 +100,19 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         const VulkanRendererConfig& m_config;
         VkPhysicalDevice m_physicalDevice = VK_NULL_HANDLE;
-        UnorderedMap<HashType, BackendVertexInputState> m_cache;
+        // Values are heap-allocated: FastSTL::unordered_map is open-addressing,
+        // so INSERT invalidates references to stored values. The draw path (and
+        // the VAOs' state-pointer memos) hold entry pointers across inserts;
+        // only the unique_ptr cell moves, never the pointee.
+        UnorderedMap<HashType, UniquePtr<BackendVertexInputState>> m_cache;
+        // Monotonic frame-boundary counter (bumped in OnFrameBoundary) for cache aging.
+        Uint64 m_frameBoundaryCounter = 0;
+        // Bumped whenever any cache entry is erased. VAOs memo a raw pointer to
+        // their heap-allocated entry (stable across map insert/rehash by
+        // construction); a memo is honored only while its recorded epoch
+        // matches, so an evicted entry can never be dereferenced through a
+        // stale memo.
+        Uint64 m_evictionEpoch = 1;
         static inline XXH64_state_t* m_hashState = XXH64_createState();
     };
 } // namespace MobileGL::MG_Backend::DirectVulkan

MobileGL/MG_Backend/DirectVulkan/Renderer/VkBufferManager.cpp

@@ -7,6 +7,8 @@
 // End of Source File Header
 
 #include "VkBufferManager.h"
+#include "../DirectVulkan.h"
+#include "VulkanRenderer.h"
 
 namespace MobileGL::MG_Backend::DirectVulkan {
     namespace {
@@ -22,6 +24,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT | VK_BUFFER_USAGE_STORAGE_BUFFER_BIT |
             VK_BUFFER_USAGE_INDIRECT_BUFFER_BIT | VK_BUFFER_USAGE_UNIFORM_TEXEL_BUFFER_BIT |
             VK_BUFFER_USAGE_TRANSFER_SRC_BIT;
+        // Appended to kPersistentBackedUsage when VK_EXT_transform_feedback is enabled
+        // (see VkBufferManagerInitInfo::transformFeedbackUsageEnabled).
+        constexpr VkBufferUsageFlags kTransformFeedbackUsage =
+            VK_BUFFER_USAGE_TRANSFORM_FEEDBACK_BUFFER_BIT_EXT;
         // The app writes into the persistent map with no explicit flush, so its memory must
         // be host-coherent (Adreno host-visible memory is; requiring it keeps us portable).
         constexpr VkMemoryPropertyFlags kPersistentBackedRequiredFlags =
@@ -53,6 +59,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             }
         }
 
+        // The CPU is about to read a buffer a shader wrote. Its bytes live in coherent
+        // host-visible GPU storage (EnsureGpuResidentStorage adopts it when the buffer is
+        // bound as a shader storage buffer), so nothing needs copying - but coherence only
+        // says the writes are visible once they have happened, so the work has to retire
+        // first.
+        void Ops_ReadbackFromGpu(BufferObject& bufferObject) {
+            (void)bufferObject;
+            if (pVulkanRenderer) {
+                pVulkanRenderer->FinishPendingGpuWork();
+            }
+        }
+
         void* Ops_AcquirePersistentMap(BufferObject& bufferObject) {
             if (g_activeBufferManager) {
                 return g_activeBufferManager->AcquirePersistentMap(bufferObject);
@@ -76,6 +94,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             .FlushMappedRange = Ops_FlushMappedRange,
             .OnDestroy = Ops_OnDestroy,
             .AcquirePersistentMap = Ops_AcquirePersistentMap,
+            .ReadbackFromGpu = Ops_ReadbackFromGpu,
         };
     } // namespace
 
@@ -141,6 +160,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_transientUploadArena.BeginFrame(frameIndex);
     }
 
+    void VkBufferManager::CollectAllDeferredReleases() {
+        for (Uint32 frameIndex = 0; frameIndex < m_deferredBufferReleases.size(); ++frameIndex) {
+            CollectDeferredReleases(frameIndex);
+        }
+        for (Uint32 frameIndex = 0; frameIndex < m_transientUploadArena.GetFrameCount(); ++frameIndex) {
+            m_transientUploadArena.CollectDeferredReleases(frameIndex);
+        }
+    }
+
     void VkBufferManager::NotifyDeviceIdle() {
         // Everything submitted so far has completed. Work recorded for the
         // current frame has not been submitted yet, so the current serial
@@ -456,7 +484,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // it from the current shadow - MappedData() is still the shadow here because the
         // frontend adopts (and drops) the shadow only after this returns.
         DeferRelease(std::move(resource->buffer));
-        if (!CreateResidentStorage(*resource, size, kPersistentBackedUsage, kPersistentBackedRequiredFlags)) {
+        const VkBufferUsageFlags persistentUsage =
+            kPersistentBackedUsage |
+            (m_initInfo.transformFeedbackUsageEnabled ? kTransformFeedbackUsage : 0);
+        if (!CreateResidentStorage(*resource, size, persistentUsage, kPersistentBackedRequiredFlags)) {
             resource->persistentMapped = false;
             resource->storageSize = 0;
             resource->usageFlags = 0;
@@ -530,6 +561,16 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         auto resource = GetOrCreateResource(bufferObject);
         bufferObject->SyncPersistentMappedRange();
 
+        // A persistently mapped resource's storage IS the application's copy of the bytes -
+        // the frontend adopted it in place of the shadow and hands out pointers into it, and
+        // a shader can have written bytes the shadow never saw (a transform feedback
+        // capture). Streaming a second copy would feed this draw the stale shadow, and the
+        // downgrade below would release the storage the application still points at,
+        // breaking the "never recreated" promise AcquirePersistentMap makes.
+        if (resource->persistentMapped) {
+            return AcquireResidentSlice(kind, bufferObject, outSlice);
+        }
+
         const VkDeviceSize size = static_cast<VkDeviceSize>(bufferObject->GetSize());
         if (size == 0) {
             MGLOG_E("VkBufferManager::AcquireStreamedSlice failed: buffer size is zero");

MobileGL/MG_Backend/DirectVulkan/Renderer/VkBufferManager.h

@@ -31,6 +31,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         VmaMemoryUsage transientMemoryUsage = VMA_MEMORY_USAGE_AUTO;
         VmaAllocationCreateFlags transientAllocationFlags = VMA_ALLOCATION_CREATE_HOST_ACCESS_SEQUENTIAL_WRITE_BIT;
         Bool transientPersistentMapping = false;
+        // VK_EXT_transform_feedback is enabled: persistent-map storage additionally
+        // carries the transform feedback usage so capture targets can bind directly.
+        Bool transformFeedbackUsageEnabled = false;
     };
 
     // The DirectVulkan storage behind one frontend buffer (pipe_resource analogue).
@@ -77,6 +80,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // Recreate all per-frame transient arenas
         Bool RecreateTransientArenas(Uint32 frameCount);
         void BeginFrame(Uint32 frameIndex);
+        // Drains every frame slot's deferred buffer/resource releases (and the
+        // transient arena's parked superseded blocks). Only valid when the
+        // caller has proven every queue submission complete; used by the
+        // present-less frame-boundary drain.
+        void CollectAllDeferredReleases();
         // All previously submitted GPU work has completed (vkDeviceWaitIdle).
         void NotifyDeviceIdle();
         // A frame slot's submission fence has been waited: every serial up to

MobileGL/MG_Backend/DirectVulkan/Renderer/VkClearManager.cpp

@@ -93,6 +93,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         const std::lock_guard<std::mutex> lock(m_mutex);
         m_pendingClears.clear();
         m_aliveObjects.clear();
+        m_pendingCount.store(static_cast<Uint32>(m_pendingClears.size()), std::memory_order_relaxed);
     }
 
     TextureIdentity VkClearManager::MakeTextureIdentity(MG_State::GLState::ITextureObject* texture) {
@@ -127,6 +128,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             m_pendingClears.erase(key);
         }
         m_aliveObjects.erase(identity);
+        m_pendingCount.store(static_cast<Uint32>(m_pendingClears.size()), std::memory_order_relaxed);
     }
 
     Bool VkClearManager::LockTextureIdentityLocked(const TextureIdentity& identity,
@@ -221,6 +223,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_aliveObjects[MakeTextureIdentity(texture.get())] = texture;
         auto& pending = m_pendingClears[key];
         MergeClearPayload(pending, clearPayload);
+        m_pendingCount.store(static_cast<Uint32>(m_pendingClears.size()), std::memory_order_relaxed);
     }
 
     void VkClearManager::QueueClear(const ClearAttachmentPayload& clearPayload,
@@ -238,6 +241,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_aliveObjects[MakeTextureIdentity(texture.get())] = texture;
         auto& pending = m_pendingClears[key];
         MergeClearPayload(pending, clearPayload);
+        m_pendingCount.store(static_cast<Uint32>(m_pendingClears.size()), std::memory_order_relaxed);
     }
 
     Bool VkClearManager::HasPendingClear(MG_State::GLState::ITextureObject* texture) {
@@ -245,6 +249,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return false;
         }
 
+        if (m_pendingCount.load(std::memory_order_relaxed) == 0) {
+            return false;  // per-draw hot path: nothing pending anywhere
+        }
+
         const Uint64 lifetimeId = texture->GetLifetimeId();
         const std::lock_guard<std::mutex> lock(m_mutex);
         for (auto it = m_pendingClears.begin(); it != m_pendingClears.end(); ++it) {
@@ -260,6 +268,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         if (key.texture == nullptr) {
             return false;
         }
+        if (m_pendingCount.load(std::memory_order_relaxed) == 0) {
+            return false;  // per-draw hot path: nothing pending anywhere
+        }
 
         const std::lock_guard<std::mutex> lock(m_mutex);
         if (m_pendingClears.find(key) == m_pendingClears.end()) {
@@ -287,6 +298,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         if (key.texture == nullptr) {
             return false;
         }
+        if (m_pendingCount.load(std::memory_order_relaxed) == 0) {
+            return false;  // per-draw hot path: nothing pending anywhere
+        }
 
         const std::lock_guard<std::mutex> lock(m_mutex);
         if (!LockTextureLocked(key, outTexture)) {
@@ -325,6 +339,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         if (texture == nullptr) {
             return false;
         }
+        if (m_pendingCount.load(std::memory_order_relaxed) == 0) {
+            return false;  // per-draw hot path: nothing pending anywhere
+        }
 
         const Uint64 lifetimeId = texture->GetLifetimeId();
         const std::lock_guard<std::mutex> lock(m_mutex);
@@ -345,6 +362,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return;
         }
 
+        if (m_pendingCount.load(std::memory_order_relaxed) == 0) {
+            return;  // per-draw hot path: nothing pending anywhere
+        }
         const TextureIdentity identity = MakeTextureIdentity(texture);
         MGLOG_D("%s: Pop all pending clears for texture %d", __func__, texture->GetExternalIndex());
         const std::lock_guard<std::mutex> lock(m_mutex);
@@ -361,6 +381,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             auto it = m_pendingClears.find(key);
             if (it != m_pendingClears.end()) {
                 m_pendingClears.erase(it);
+            m_pendingCount.store(static_cast<Uint32>(m_pendingClears.size()), std::memory_order_relaxed);
             }
         }
 

MobileGL/MG_Backend/DirectVulkan/Renderer/VkClearManager.h

@@ -14,6 +14,7 @@
 #include "MG_Util/Math/VectorTypes.h"
 
 #include <Includes.h>
+#include <atomic>
 #include <unordered_map>
 
 namespace MobileGL::MG_Backend::DirectVulkan {
@@ -120,7 +121,19 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                SharedPtr<MG_State::GLState::ITextureObject>& outTexture);
 
         Uint8 m_gcCounter = 0;
+    public:
+        // Lock-free probe for the consecutive-draw fast path: any pending clear
+        // forces the full SetupDraw path (which materializes/consumes it).
+        Bool HasAnyPendingClears() const { return m_pendingCount.load(std::memory_order_relaxed) != 0; }
+
+    private:
         mutable std::mutex m_mutex;
+        // Lock-free mirror of m_pendingClears.size(), maintained under m_mutex
+        // by every mutation. The per-draw probes (HasPendingClear/GetPending*)
+        // read it before taking the lock: during draw batches the pending set
+        // is almost always empty, so this turns several locked map probes per
+        // draw into one relaxed load.
+        std::atomic<Uint32> m_pendingCount{0};
         std::unordered_map<PendingClearKey, ClearAttachmentPayload, PendingClearKeyHash> m_pendingClears;
         std::unordered_map<TextureIdentity, WeakPtr<MG_State::GLState::ITextureObject>, TextureIdentityHash> m_aliveObjects;
     };

MobileGL/MG_Backend/DirectVulkan/Renderer/VkRenderPassManager.cpp

@@ -16,31 +16,21 @@
 
 namespace MobileGL::MG_Backend::DirectVulkan {
     static Bool TryResolveSampleCountFlagBits(Int requestedSamples, VkSampleCountFlagBits& outSampleCount) {
-        switch (requestedSamples <= 0 ? 1 : requestedSamples) {
-        case 1:
+        // GL promises "at least the requested samples", so a non-power-of-two
+        // request (legal in GL, e.g. 3) rounds up to the next Vulkan bit.
+        if (requestedSamples <= 1) {
             outSampleCount = VK_SAMPLE_COUNT_1_BIT;
             return true;
-        case 2:
-            outSampleCount = VK_SAMPLE_COUNT_2_BIT;
-            return true;
-        case 4:
-            outSampleCount = VK_SAMPLE_COUNT_4_BIT;
-            return true;
-        case 8:
-            outSampleCount = VK_SAMPLE_COUNT_8_BIT;
-            return true;
-        case 16:
-            outSampleCount = VK_SAMPLE_COUNT_16_BIT;
-            return true;
-        case 32:
-            outSampleCount = VK_SAMPLE_COUNT_32_BIT;
-            return true;
-        case 64:
-            outSampleCount = VK_SAMPLE_COUNT_64_BIT;
-            return true;
-        default:
+        }
+        if (requestedSamples > 64) {
             return false;
         }
+        Uint32 bit = 1;
+        while (bit < static_cast<Uint32>(requestedSamples)) {
+            bit <<= 1;
+        }
+        outSampleCount = static_cast<VkSampleCountFlagBits>(bit);
+        return true;
     }
 
     static VkImageAspectFlags ResolveImageAspectMaskForFormat(VkFormat format) {
@@ -166,6 +156,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         if (view != VK_NULL_HANDLE) {
             vkDestroyImageView(device, view, nullptr);
         }
+        if (unormTwinView != VK_NULL_HANDLE) {
+            vkDestroyImageView(device, unormTwinView, nullptr);
+        }
         if (image != VK_NULL_HANDLE && allocation != nullptr) {
             vmaDestroyImage(allocator, image, allocation);
         }
@@ -173,6 +166,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         image = VK_NULL_HANDLE;
         allocation = nullptr;
         view = VK_NULL_HANDLE;
+        unormTwinView = VK_NULL_HANDLE;
         layout = VK_IMAGE_LAYOUT_UNDEFINED;
         format = VK_FORMAT_UNDEFINED;
         aspect = VK_IMAGE_ASPECT_NONE;
@@ -180,6 +174,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         sampleCount = VK_SAMPLE_COUNT_1_BIT;
         internalFormat = TextureInternalFormat::Unknown;
         samples = 0;
+        deadSinceFrame = kNeverObservedDead;
     }
 
     VkRenderPassManager::VkRenderPassManager(VkDevice device,
@@ -206,6 +201,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             resource.Destroy(m_device, m_allocator);
         }
         m_renderbufferResources.clear();
+        CollectDeferredRenderbufferReleases(/*destroyAll=*/true); // caller guarantees device idle
         m_pendingRenderbufferClears.clear();
         RenderPassEntry::s_textureResourcesScratch.clear();
         s_activeRenderPass = {};
@@ -213,22 +209,80 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_rpFastValid = false;
     }
 
-    void VkRenderPassManager::CollectRenderbufferGarbage() {
-        Vector<MG_State::GLState::RenderbufferObject*> deadRenderbuffers;
-        deadRenderbuffers.reserve(m_renderbufferResources.size());
-        for (auto& [renderbuffer, resource] : m_renderbufferResources) {
-            const auto liveRenderbuffer = resource.renderbuffer.lock();
-            if (!liveRenderbuffer || liveRenderbuffer.get() != renderbuffer) {
-                deadRenderbuffers.emplace_back(renderbuffer);
-            }
+    Uint64 VkRenderPassManager::RetireAgeFrames() const {
+        // MaxFramesInFlight + 2 covers the frame ring plus one boundary for the
+        // recording-to-submit gap and one because OnPresent runs ahead of Present's
+        // fence wait; the floor of 8 keeps a margin over the default ring of 3 while
+        // still releasing multi-MB attachment memory promptly (the render-pass cache's
+        // 1024-frame retirement would pin it for no additional safety).
+        return std::max<Uint64>(8, static_cast<Uint64>(m_config.MaxFramesInFlight) + 2);
+    }
+
+    void VkRenderPassManager::DeferRenderbufferBackingRelease(RenderbufferResource& resource) {
+        // The superseded backing may still be referenced by in-flight command buffers
+        // (glRenderbufferStorage can respecify a renderbuffer drawn this very frame),
+        // so it is parked and destroyed only after RetireAgeFrames() boundaries.
+        if (resource.image == VK_NULL_HANDLE && resource.view == VK_NULL_HANDLE) {
+            return;
         }
-        for (auto* renderbuffer : deadRenderbuffers) {
-            auto resourceIt = m_renderbufferResources.find(renderbuffer);
-            if (resourceIt != m_renderbufferResources.end()) {
-                resourceIt->second.Destroy(m_device, m_allocator);
-                m_renderbufferResources.erase(resourceIt);
+        m_deferredRenderbufferReleases.push_back(
+            {resource.image, resource.allocation, resource.view, resource.unormTwinView, m_frameCounter});
+        resource.image = VK_NULL_HANDLE;
+        resource.allocation = nullptr;
+        resource.view = VK_NULL_HANDLE;
+        resource.unormTwinView = VK_NULL_HANDLE;
+    }
+
+    void VkRenderPassManager::CollectDeferredRenderbufferReleases(Bool destroyAll) {
+        if (m_deferredRenderbufferReleases.empty()) {
+            return;
+        }
+        const Uint64 retireAgeFrames = RetireAgeFrames();
+        std::erase_if(m_deferredRenderbufferReleases, [&](DeferredRenderbufferRelease& release) {
+            if (!destroyAll && m_frameCounter - release.deferredAtFrame < retireAgeFrames) {
+                return false;
             }
-            m_pendingRenderbufferClears.erase(renderbuffer);
+            if (release.view != VK_NULL_HANDLE) {
+                vkDestroyImageView(m_device, release.view, nullptr);
+            }
+            if (release.unormTwinView != VK_NULL_HANDLE) {
+                vkDestroyImageView(m_device, release.unormTwinView, nullptr);
+            }
+            if (release.image != VK_NULL_HANDLE) {
+                vmaDestroyImage(m_allocator, release.image, release.allocation);
+            }
+            return true;
+        });
+    }
+
+    void VkRenderPassManager::CollectRenderbufferGarbage() {
+        // Two-phase reclamation: a dead renderbuffer's VkImage may still be referenced by
+        // command buffers submitted up to frames-in-flight frames ago (it was legally
+        // attached and drawn right up to its deletion), so the first observation of an
+        // expired weak reference only stamps the current frame counter; Destroy runs once
+        // enough frame boundaries have passed that the stamping frame's submission fence
+        // has provably been waited (see RetireAgeFrames).
+        const Uint64 retireAgeFrames = RetireAgeFrames();
+        for (auto it = m_renderbufferResources.begin(); it != m_renderbufferResources.end();) {
+            auto& resource = it->second;
+            const auto liveRenderbuffer = resource.renderbuffer.lock();
+            if (liveRenderbuffer && liveRenderbuffer.get() == it->first) {
+                resource.deadSinceFrame = RenderbufferResource::kNeverObservedDead;
+                ++it;
+                continue;
+            }
+            if (resource.deadSinceFrame == RenderbufferResource::kNeverObservedDead) {
+                resource.deadSinceFrame = m_frameCounter;
+                ++it;
+                continue;
+            }
+            if (m_frameCounter - resource.deadSinceFrame < retireAgeFrames) {
+                ++it;
+                continue;
+            }
+            m_pendingRenderbufferClears.erase(it->first);
+            resource.Destroy(m_device, m_allocator);
+            it = m_renderbufferResources.erase(it);
         }
     }
 
@@ -249,7 +303,47 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
 
         const auto internalFormat = renderbuffer->GetInternalFormat();
-        const VkFormat format = MG_Util::ConvertTextureInternalFormatToVkEnum(internalFormat);
+        // Three-channel color formats widen to their RGBA twin exactly like textures do
+        // (VkTextureManager::ResolveTextureFormatInfo): blits/resolves between a
+        // renderbuffer and a texture of the same GL format then see one VkFormat.
+        const VkFormat format = [&]() -> VkFormat {
+            switch (internalFormat) {
+            case TextureInternalFormat::RGB:
+            case TextureInternalFormat::RGB8:
+            case TextureInternalFormat::R3G3B2:
+            case TextureInternalFormat::RGB4:
+            case TextureInternalFormat::RGB5:
+                return VK_FORMAT_R8G8B8A8_UNORM;
+            case TextureInternalFormat::SRGB8:
+                return VK_FORMAT_R8G8B8A8_SRGB;
+            case TextureInternalFormat::RGB8Snorm:
+                return VK_FORMAT_R8G8B8A8_SNORM;
+            case TextureInternalFormat::RGB10:
+            case TextureInternalFormat::RGB12:
+            case TextureInternalFormat::RGB16:
+                return VK_FORMAT_R16G16B16A16_UNORM;
+            case TextureInternalFormat::RGB16Snorm:
+                return VK_FORMAT_R16G16B16A16_SNORM;
+            case TextureInternalFormat::RGB16F:
+                return VK_FORMAT_R16G16B16A16_SFLOAT;
+            case TextureInternalFormat::RGB32F:
+                return VK_FORMAT_R32G32B32A32_SFLOAT;
+            case TextureInternalFormat::RGB8I:
+                return VK_FORMAT_R8G8B8A8_SINT;
+            case TextureInternalFormat::RGB8UI:
+                return VK_FORMAT_R8G8B8A8_UINT;
+            case TextureInternalFormat::RGB16I:
+                return VK_FORMAT_R16G16B16A16_SINT;
+            case TextureInternalFormat::RGB16UI:
+                return VK_FORMAT_R16G16B16A16_UINT;
+            case TextureInternalFormat::RGB32I:
+                return VK_FORMAT_R32G32B32A32_SINT;
+            case TextureInternalFormat::RGB32UI:
+                return VK_FORMAT_R32G32B32A32_UINT;
+            default:
+                return MG_Util::ConvertTextureInternalFormatToVkEnum(internalFormat);
+            }
+        }();
         const VkImageAspectFlags aspect = ResolveImageAspectMaskForFormat(format);
         // Renderbuffers are never sampled (GL has no way to bind one to a sampler), so the
         // usage set is attachment + transfer: transfer covers readback (vkCmdCopyImageToBuffer),
@@ -259,6 +353,46 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                                        : VK_IMAGE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT) |
             VK_IMAGE_USAGE_TRANSFER_SRC_BIT | VK_IMAGE_USAGE_TRANSFER_DST_BIT;
 
+        // GL allows the implementation to allocate more samples than requested
+        // (glRenderbufferStorageMultisample only promises "at least"), and devices
+        // like llvmpipe expose 1x/4x but not 2x. Round the request up to the
+        // nearest supported count for this format.
+        if (renderbuffer->GetSamples() > 0) {
+            auto supportedIt = m_attachmentSampleCountsByFormat.find(format);
+            if (supportedIt == m_attachmentSampleCountsByFormat.end()) {
+                VkImageFormatProperties formatProperties{};
+                VkSampleCountFlags supported = VK_SAMPLE_COUNT_1_BIT;
+                if (vkGetPhysicalDeviceImageFormatProperties(m_physicalDevice, format, VK_IMAGE_TYPE_2D,
+                                                             VK_IMAGE_TILING_OPTIMAL, imageUsage, 0,
+                                                             &formatProperties) == VK_SUCCESS) {
+                    supported = formatProperties.sampleCounts;
+                }
+                supportedIt = m_attachmentSampleCountsByFormat.emplace(format, supported).first;
+            }
+            const VkSampleCountFlags supported = supportedIt->second;
+            if ((supported & sampleCount) == 0) {
+                // Smallest supported count above the request, else the largest below it.
+                Uint32 rounded = 0;
+                for (Uint32 bit = static_cast<Uint32>(sampleCount) << 1; bit <= VK_SAMPLE_COUNT_64_BIT; bit <<= 1) {
+                    if ((supported & bit) != 0) {
+                        rounded = bit;
+                        break;
+                    }
+                }
+                if (rounded == 0) {
+                    for (Uint32 bit = static_cast<Uint32>(sampleCount) >> 1; bit != 0; bit >>= 1) {
+                        if ((supported & bit) != 0) {
+                            rounded = bit;
+                            break;
+                        }
+                    }
+                }
+                if (rounded != 0) {
+                    sampleCount = static_cast<VkSampleCountFlagBits>(rounded);
+                }
+            }
+        }
+
         auto& resource = m_renderbufferResources[renderbuffer.get()];
         const Bool needsCreate =
             resource.image == VK_NULL_HANDLE ||
@@ -270,9 +404,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             resource.samples != renderbuffer->GetSamples();
         if (!needsCreate) {
             resource.renderbuffer = renderbuffer;
+            // A new renderbuffer at a recycled address may adopt a compatible entry that
+            // was already stamped dead; it is alive again, so cancel the aging.
+            resource.deadSinceFrame = RenderbufferResource::kNeverObservedDead;
             return &resource;
         }
 
+        // Respecify: park the old backing for aged destruction instead of destroying
+        // inline - it may still be referenced by in-flight command buffers.
+        DeferRenderbufferBackingRelease(resource);
         resource.Destroy(m_device, m_allocator);
         resource.renderbuffer = renderbuffer;
 
@@ -290,6 +430,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         imageInfo.usage = imageUsage;
         imageInfo.samples = sampleCount;
         imageInfo.sharingMode = VK_SHARING_MODE_EXCLUSIVE;
+        // sRGB renderbuffers attach through their UNORM twin while GL_FRAMEBUFFER_SRGB
+        // is disabled, which needs a format-reinterpreting second view.
+        const Bool hasUnormTwin = ResolveSrgbAttachmentWriteFormat(format, false) != format;
+        if (hasUnormTwin) {
+            imageInfo.flags |= VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT;
+        }
 
         VkImageFormatProperties imageFormatProperties{};
         const VkResult imageFormatResult = vkGetPhysicalDeviceImageFormatProperties(
@@ -324,6 +470,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         viewInfo.subresourceRange.layerCount = 1;
         VK_VERIFY(vkCreateImageView(m_device, &viewInfo, nullptr, &resource.view),
                   "vkCreateImageView(renderbuffer)");
+        if (hasUnormTwin) {
+            viewInfo.format = ResolveSrgbAttachmentWriteFormat(format, false);
+            VK_VERIFY(vkCreateImageView(m_device, &viewInfo, nullptr, &resource.unormTwinView),
+                      "vkCreateImageView(renderbuffer unorm twin)");
+        }
 
         resource.layout = VK_IMAGE_LAYOUT_UNDEFINED;
         resource.format = format;
@@ -420,12 +571,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     }
 
     VkRenderPassManager::HashType VkRenderPassManager::ComputeHash(
-        const MG_State::GLState::FramebufferObject& fbo, Uint32 swapchainImageIndex, Bool includePendingClear) {
+        const MG_State::GLState::FramebufferObject& fbo, Uint32 swapchainImageIndex, Bool includePendingClear,
+        Bool includeDefaultFboDepthStencil) {
         XXHASH_VERIFY(XXH64_reset(m_hashState, m_config.CacheVersion));
         const Bool isDefaultFbo = fbo.IsDefaultFramebuffer();
         if (isDefaultFbo) {
             XXHASH_VERIFY(XXH64_update(m_hashState, &swapchainImageIndex, sizeof(swapchainImageIndex)));
         }
+        // sRGB attachments switch between their sRGB and UNORM-twin views with this
+        // capability (ResolveSrgbAttachmentWriteFormat), changing the render pass formats.
+        const Bool framebufferSrgbEnabled =
+            MG_State::pGLContext->IsCapabilityEnabled(MobileGL::CapabilityInput::FramebufferSrgb);
+        XXHASH_VERIFY(XXH64_update(m_hashState, &framebufferSrgbEnabled, sizeof(framebufferSrgbEnabled)));
         auto& drawBuffers = fbo.GetDrawBuffers();
         XXHASH_VERIFY(XXH64_update(m_hashState, drawBuffers.data(), drawBuffers.size() * sizeof(drawBuffers[0])));
         auto readBuffer = fbo.GetReadBuffer();
@@ -499,9 +656,17 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                          attachment <= FramebufferAttachmentType::BackRight);
                     if (isDefaultColorAttachment) {
                         currentLayout = m_swapchainObject.GetImageLayout(swapchainImageIndex);
+                        // Content validity feeds the attachment's loadOp (see the
+                        // creation path), so it must key the cache as well.
+                        if (!m_swapchainObject.IsImageContentDefined(swapchainImageIndex)) {
+                            currentLayout = VK_IMAGE_LAYOUT_UNDEFINED;
+                        }
                     } else if (attachment == FramebufferAttachmentType::Depth ||
                                attachment == FramebufferAttachmentType::Stencil) {
                         currentLayout = m_swapchainObject.GetDepthStencilImageLayout(swapchainImageIndex);
+                        if (!m_swapchainObject.IsDepthStencilContentDefined(swapchainImageIndex)) {
+                            currentLayout = VK_IMAGE_LAYOUT_UNDEFINED;
+                        }
                     }
                 } else {
                     auto* textureResource = m_textureManager.SyncTextureAndGetDescriptor(*texture);
@@ -556,14 +721,49 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             combineFramebufferAttachmentObjHash(drawbuf);
         }
 
-        combineFramebufferAttachmentObjHash(FramebufferAttachmentType::Depth);
-        combineFramebufferAttachmentObjHash(FramebufferAttachmentType::Stencil);
+        // The depth-less default-FBO flavor omits the depth/stencil attachment
+        // entirely, so it must hash differently from the depth-full flavor.
+        const Bool depthStencilIncluded = !isDefaultFbo || includeDefaultFboDepthStencil;
+        XXHASH_VERIFY(XXH64_update(m_hashState, &depthStencilIncluded, sizeof(depthStencilIncluded)));
+        if (depthStencilIncluded) {
+            combineFramebufferAttachmentObjHash(FramebufferAttachmentType::Depth);
+            combineFramebufferAttachmentObjHash(FramebufferAttachmentType::Stencil);
+        }
 
         return XXH64_digest(m_hashState);
     }
 
     RenderPassEntry& VkRenderPassManager::GetOrCreateRenderPass(const MG_State::GLState::FramebufferObject& fbo,
-                                                                Uint32 swapchainImageIndex) {
+                                                                Uint32 swapchainImageIndex,
+                                                                Bool drawUsesDepthStencil) {
+        // Resolve the default-FBO depth flavor (see the header comment): keep the
+        // depth attachment when the caller needs it, when a depth/stencil clear is
+        // pending, or when the active pass already carries it (escalate-only, so
+        // alternating depth-less draws never split an established depth pass).
+        Bool includeDefaultFboDepthStencil = true;
+        if (fbo.IsDefaultFramebuffer()) {
+            Bool activeDefaultHasDepthStencil = false;
+            if (const auto* active = GetActiveRenderPass()) {
+                Bool activeIsSwapchainPass = false;
+                Bool activeHasSwapchainDepthStencil = false;
+                for (const auto& tracked : active->trackedAttachmentLayouts) {
+                    activeIsSwapchainPass |= tracked.target == TrackedAttachmentTarget::SwapchainColor;
+                    activeHasSwapchainDepthStencil |=
+                        tracked.target == TrackedAttachmentTarget::SwapchainDepthStencil;
+                }
+                activeDefaultHasDepthStencil = activeIsSwapchainPass && activeHasSwapchainDepthStencil;
+            }
+            const auto& defaultDepthAtt = fbo.GetAttachment(FramebufferAttachmentType::Depth);
+            const auto& defaultStencilAtt = fbo.GetAttachment(FramebufferAttachmentType::Stencil);
+            const Bool pendingDepthStencilClear =
+                (defaultDepthAtt.IsTexture() && m_clearManager.HasPendingClear(defaultDepthAtt)) ||
+                HasPendingRenderbufferClear(defaultDepthAtt) ||
+                (defaultStencilAtt.IsTexture() && m_clearManager.HasPendingClear(defaultStencilAtt)) ||
+                HasPendingRenderbufferClear(defaultStencilAtt);
+            includeDefaultFboDepthStencil =
+                drawUsesDepthStencil || activeDefaultHasDepthStencil || pendingDepthStencilClear;
+        }
+
         auto hasPendingClearOnFramebuffer = [&]() -> Bool {
             const auto& drawBuffers = fbo.GetDrawBuffers();
             for (auto attachment : drawBuffers) {
@@ -613,6 +813,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             m_rpFastFboVersion == fbo.GetObjectVersion() && m_rpFastSwapchainIndex == swapchainImageIndex &&
             m_rpFastTexEpoch == m_textureManager.GetTextureImageEpoch() &&
             m_rpFastRbEpoch == m_renderbufferImageEpoch &&
+            (!fbo.IsDefaultFramebuffer() || m_rpFastHadDepthStencil == includeDefaultFboDepthStencil) &&
             m_rpFastRenderPassHash == activeRenderPass->hash && !hasPendingClearOnFramebuffer()) {
             auto activeIt = m_renderPasses.find(activeRenderPass->hash);
             if (activeIt != m_renderPasses.end()) {
@@ -621,7 +822,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             }
         }
 
-        auto compatibilityHash = ComputeHash(fbo, swapchainImageIndex, false);
+        auto compatibilityHash = ComputeHash(fbo, swapchainImageIndex, false, includeDefaultFboDepthStencil);
         if (activeRenderPass != nullptr &&
             activeRenderPass->CompatibleWith(compatibilityHash) &&
             !hasPendingClearOnFramebuffer()) {
@@ -638,10 +839,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             m_rpFastTexEpoch = m_textureManager.GetTextureImageEpoch();
             m_rpFastRbEpoch = m_renderbufferImageEpoch;
             m_rpFastRenderPassHash = activeRenderPass->hash;
+            m_rpFastHadDepthStencil = activeIt->second.hasDepthStencilAttachment;
             activeIt->second.lastUsedFrame = m_frameCounter;
             return activeIt->second;
         }
-        auto hash = ComputeHash(fbo, swapchainImageIndex, true);
+        auto hash = ComputeHash(fbo, swapchainImageIndex, true, includeDefaultFboDepthStencil);
         auto it = m_renderPasses.find(hash);
         if (it != m_renderPasses.end()) {
             it->second.lastUsedFrame = m_frameCounter;
@@ -728,8 +930,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                     }
 
                     const VkImageLayout trackedRbLayout = rbResource->layout;
+                    const Bool rbFramebufferSrgb =
+                        MG_State::pGLContext->IsCapabilityEnabled(MobileGL::CapabilityInput::FramebufferSrgb);
+                    const VkFormat rbAttachmentFormat =
+                        ResolveSrgbAttachmentWriteFormat(rbResource->format, rbFramebufferSrgb);
                     rbDesc.flags = 0;
-                    rbDesc.format = rbResource->format;
+                    rbDesc.format = rbAttachmentFormat;
                     rbDesc.samples = rbResource->sampleCount;
                     rbDesc.loadOp = rbHasClear ? VK_ATTACHMENT_LOAD_OP_CLEAR :
                                     (trackedRbLayout == VK_IMAGE_LAYOUT_UNDEFINED ? VK_ATTACHMENT_LOAD_OP_DONT_CARE
@@ -764,7 +970,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                         .finalLayout = rbDesc.finalLayout,
                     });
                     textureResources.emplace_back(nullptr);
-                    attachmentViews.emplace_back(rbResource->view);
+                    attachmentViews.emplace_back(rbAttachmentFormat != rbResource->format ? rbResource->unormTwinView
+                                                                                          : rbResource->view);
                     MOBILEGL_ASSERT(attachmentViews.back() != VK_NULL_HANDLE,
                                     "GetOrCreateRenderPass: renderbuffer view missing at color attachment %d", i);
 
@@ -833,6 +1040,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                         MOBILEGL_ASSERT(swapchainImageIndex < swapchainViews.size(),
                                         "GetOrCreateRenderPass: swapchain image index out of range");
                         trackedColorLayout = m_swapchainObject.GetImageLayout(swapchainImageIndex);
+                        // EGL: a presented color buffer's content is undefined when its
+                        // image comes back around (EGL_BUFFER_DESTROYED, the default
+                        // swap behaviour) - skip the tile load instead of reloading
+                        // stale pixels nobody may rely on.
+                        if (!hasClear && !m_swapchainObject.IsImageContentDefined(swapchainImageIndex)) {
+                            trackedColorLayout = VK_IMAGE_LAYOUT_UNDEFINED;
+                        }
                         trackedAttachmentLayouts.emplace_back(TrackedAttachmentLayoutInfo {
                             .target = TrackedAttachmentTarget::SwapchainColor,
                             .swapchainImageIndex = swapchainImageIndex,
@@ -846,12 +1060,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                         MOBILEGL_ASSERT(textureResource,
                                         "GetOrCreateRenderPass: SyncTextureAndGetDescriptor failed at color attachment %d", i);
                         textureResources.emplace_back(textureResource);
-                        desc.format = textureResource->format;
+                        desc.format = ResolveSrgbAttachmentWriteFormat(
+                            textureResource->format,
+                            MG_State::pGLContext->IsCapabilityEnabled(MobileGL::CapabilityInput::FramebufferSrgb));
                         attachmentSampleCount = textureResource->sampleCount;
                         trackedColorLayout = textureResource->layout;
                         trackedAttachmentLayouts.emplace_back(TrackedAttachmentLayoutInfo {
                             .target = TrackedAttachmentTarget::Texture,
                             .texture = att.GetTexture(),
+                            .textureRaw = att.GetTexture().get(),
                             .textureMipLevel = attachmentMipLevel,
                             .finalLayout = desc.finalLayout,
                         });
@@ -915,6 +1132,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         };
         const auto* selectedDepthStencilAttachment = isUsableDepthStencilAttachment(depthAtt) ? &depthAtt :
                                                      (isUsableDepthStencilAttachment(stencilAtt) ? &stencilAtt : nullptr);
+        // Depth-less default-FBO flavor: nothing in this pass touches depth/stencil
+        // and their content is undefined anyway (EGL swap), so drop the attachment
+        // and its whole tile load + store.
+        if (isDefaultFbo && !includeDefaultFboDepthStencil) {
+            selectedDepthStencilAttachment = nullptr;
+        }
         const Bool hasDistinctDepthAndStencilAttachments =
             isUsableDepthStencilAttachment(depthAtt) && isUsableDepthStencilAttachment(stencilAtt) &&
             !sameDepthStencilAttachmentObject(depthAtt, stencilAtt);
@@ -933,6 +1156,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             VkImageLayout trackedDepthLayout = isDefaultFbo ?
                 m_swapchainObject.GetDepthStencilImageLayout(swapchainImageIndex) :
                 VK_IMAGE_LAYOUT_DEPTH_STENCIL_ATTACHMENT_OPTIMAL;
+            // EGL 1.5 §3.10.1: every ancillary (depth/stencil) buffer's content is
+            // undefined after a swap, so the first default-FBO pass of a frame can
+            // skip the depth/stencil tile load outright.
+            if (isDefaultFbo && !m_swapchainObject.IsDepthStencilContentDefined(swapchainImageIndex)) {
+                trackedDepthLayout = VK_IMAGE_LAYOUT_UNDEFINED;
+            }
             depthAttachmentDescription.flags = 0;
             VkSampleCountFlagBits depthAttachmentSampleCount = VK_SAMPLE_COUNT_1_BIT;
             Int depthAttachmentId = 0;
@@ -1016,6 +1245,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 trackedAttachmentLayouts.emplace_back(TrackedAttachmentLayoutInfo {
                     .target = TrackedAttachmentTarget::Texture,
                     .texture = selectedDepthStencilAttachment->GetTexture(),
+                    .textureRaw = selectedDepthStencilAttachment->GetTexture().get(),
                     .textureMipLevel = attachmentMipLevel,
                     .finalLayout = depthAttachmentDescription.finalLayout,
                 });
@@ -1061,6 +1291,22 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
         const Bool hasDepthStencilAttachment = depthAttachmentRef.attachment != VK_ATTACHMENT_UNUSED;
 
+        // Declare only the used colour-reference span. The GL draw-buffer array
+        // always spans 8 slots, so passes used to declare colorAttachmentCount=8
+        // with trailing VK_ATTACHMENT_UNUSED holes - and Adreno configures its
+        // per-pixel render-backend/export path from the DECLARED count, so every
+        // fragment of every pass paid the 8-target export cost (measured on
+        // Adreno 650 / MC 26.2: 11.9 -> 7.5 ms of GPU time per frame, with the
+        // single-quad swapchain blit pass alone dropping 1.26 -> 0.40 ms).
+        // Interior GL_NONE holes keep their slots so fragment-output locations
+        // still line up; a fragment output at a location past the trimmed count
+        // is discarded, which is exactly GL's semantic for writing to a draw
+        // buffer set to GL_NONE.
+        while (!colorAttachmentRefs.empty() &&
+               colorAttachmentRefs.back().attachment == VK_ATTACHMENT_UNUSED) {
+            colorAttachmentRefs.pop_back();
+        }
+
         // Subpass
         VkSubpassDescription subpassDesc;
         subpassDesc.flags = 0;
@@ -1178,6 +1424,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     void VkRenderPassManager::OnPresent() {
         ++m_frameCounter;
 
+        // Runs every frame boundary, ahead of the render-pass sweep gate below: the walk
+        // is O(#renderbuffer resources) — single digits in practice — and per-frame
+        // invocation keeps dead-resource reclaim latency at the aging bound instead of
+        // coupling it to renderbuffer *use* (the GetOrCreateRenderbufferResource call
+        // site never runs again once an app stops using renderbuffers).
+        CollectRenderbufferGarbage();
+        CollectDeferredRenderbufferReleases(/*destroyAll=*/false);
+
         // Sweep occasionally; evict entries whose last use is far past every
         // in-flight frame so their VkRenderPass/VkFramebuffer can be destroyed
         // safely (RenderPassEntry's destructor releases the handles).
@@ -1187,6 +1441,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return;
         }
 
+        // Collect the dying handles and notify once after the loop: pipelines hashed
+        // on them share the entries' >kRetireAgeFrames idleness (they are only bound
+        // by draws that hit those entries), so the observer may destroy them
+        // immediately - and a single batched notification costs one pipeline-cache
+        // scan instead of one per evicted pass.
+        Vector<VkRenderPass> destroyedRenderPasses;
         const Uint64 activeHash = s_hasActiveRenderPass ? s_activeRenderPass.hash : 0;
         for (auto it = m_renderPasses.begin(); it != m_renderPasses.end();) {
             const Bool isActive = s_hasActiveRenderPass && it->first == activeHash;
@@ -1194,11 +1454,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 if (m_rpFastValid && m_rpFastRenderPassHash == it->first) {
                     m_rpFastValid = false;
                 }
+                destroyedRenderPasses.push_back(it->second.renderPass);
                 it = m_renderPasses.erase(it);
             } else {
                 ++it;
             }
         }
+        if (!destroyedRenderPasses.empty() && m_evictionObserver != nullptr) {
+            m_evictionObserver->OnRenderPassesDestroyed(destroyedRenderPasses);
+        }
     }
 
     Bool VkRenderPassManager::BeginRenderPass(VkCommandBuffer commandBuffer, RenderPassEntry& renderPassEntry) {
@@ -1251,6 +1515,17 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         renderPassBeginInfo.pClearValues = clearValues.data();
 
         vkCmdBeginRenderPass(commandBuffer, &renderPassBeginInfo, VK_SUBPASS_CONTENTS_INLINE);
+        // Pre-pass stream bookkeeping: this pass's attachment images are now
+        // referenced by the open frame recording.
+        if (s_textureManager != nullptr) {
+            for (const auto& tracked : renderPassEntry.trackedAttachmentLayouts) {
+                if (tracked.target == TrackedAttachmentTarget::Texture) {
+                    if (const auto texture = tracked.texture.lock()) {
+                        s_textureManager->StampTextureRecordingUse(texture.get());
+                    }
+                }
+            }
+        }
         for (const auto& pending: renderPassEntry.pendingClearAttachments) {
             if (pending.hasInlinePayload) {
                 if (s_renderPassManager != nullptr) {
@@ -1303,11 +1578,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                     case TrackedAttachmentTarget::SwapchainColor:
                         MOBILEGL_ASSERT(s_swapchainObject != nullptr, "EndRenderPass: swapchain object is null");
                         s_swapchainObject->SetImageLayout(trackedAttachment.swapchainImageIndex, trackedAttachment.finalLayout);
+                        // The pass stored into the attachment: its content is defined
+                        // until the image is next presented.
+                        s_swapchainObject->SetImageContentDefined(trackedAttachment.swapchainImageIndex, true);
                         break;
                     case TrackedAttachmentTarget::SwapchainDepthStencil:
                         MOBILEGL_ASSERT(s_swapchainObject != nullptr, "EndRenderPass: swapchain object is null");
                         s_swapchainObject->SetDepthStencilImageLayout(trackedAttachment.swapchainImageIndex,
                                                                       trackedAttachment.finalLayout);
+                        s_swapchainObject->SetDepthStencilContentDefined(trackedAttachment.swapchainImageIndex, true);
                         break;
                     default:
                         MOBILEGL_ASSERT(false, "EndRenderPass: unsupported tracked attachment target=%d",

MobileGL/MG_Backend/DirectVulkan/Renderer/VkRenderPassManager.h

@@ -42,6 +42,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     struct TrackedAttachmentLayoutInfo {
         TrackedAttachmentTarget target = TrackedAttachmentTarget::Texture;
         WeakPtr<MG_State::GLState::ITextureObject> texture;
+        // Identity-compare shortcut for the per-draw "does the active pass use
+        // this sampled texture" probe: comparing this against a LIVE texture's
+        // address needs no weak_ptr::lock (two refcount atomics per probe).
+        // May dangle once the texture dies - compare only, never dereference.
+        MG_State::GLState::ITextureObject* textureRaw = nullptr;
         WeakPtr<MG_State::GLState::RenderbufferObject> renderbuffer;
         Uint32 textureMipLevel = 0;
         Uint32 swapchainImageIndex = 0;
@@ -157,19 +162,53 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     class VkRenderPassManager {
     public:
         using HashType = Uint64;
+
+        // Notified once per OnPresent sweep with every aged-out entry's VkRenderPass
+        // value: pipelines are hashed on the raw handle, and once destroyed the value
+        // may be recycled for an incompatible pass, so dependent caches must purge
+        // everything keyed on them before any new pass can be created (the sweep and
+        // the notification run back-to-back with no creation in between; observers
+        // compare the values, never dereference them). Batched so a mass-idle cohort
+        // (shader-pack switch, dimension exit) costs the observer one pipeline-cache
+        // scan, not one per dying pass. The wholesale paths
+        // (Shutdown/RecreateSwapchain) do not notify - their callers already drop
+        // every pipeline outright.
+        class IEvictionObserver {
+        public:
+            virtual ~IEvictionObserver() = default;
+            virtual void OnRenderPassesDestroyed(const Vector<VkRenderPass>& renderPasses) = 0;
+        };
+
         VkRenderPassManager(VkDevice device,
             VkPhysicalDevice physicalDevice, VmaAllocator allocator, const VulkanRendererConfig& config,
             VkClearManager& clearManager, VkTextureManager& textureManager, SwapchainObject& swapchainObject);
         ~VkRenderPassManager();
 
+        // Observer may be null (no notifications). Not owned.
+        void SetEvictionObserver(IEvictionObserver* observer) { m_evictionObserver = observer; }
+
         Bool Initialize();
         void Shutdown();
 
         HashType ComputeHash(
             const MG_State::GLState::FramebufferObject& fbo,
             Uint32 swapchainImageIndex,
-            Bool includePendingClear = true);
-        RenderPassEntry& GetOrCreateRenderPass(const MG_State::GLState::FramebufferObject& fbo, Uint32 swapchainImageIndex);
+            Bool includePendingClear = true,
+            Bool includeDefaultFboDepthStencil = true);
+        // drawUsesDepthStencil: whether the operation about to run inside the pass
+        // reads or writes the depth/stencil buffer (depth test or stencil test
+        // enabled, or a depth/stencil clear). Only consulted for the DEFAULT
+        // framebuffer: EGL undefines its ancillary buffers at every swap, so a
+        // default-FBO pass whose draws provably never touch depth/stencil is
+        // created WITHOUT the depth attachment - on a tiler that skips the whole
+        // depth tile load AND store. The flavor only escalates: once a pass with
+        // depth is active, later depth-less draws keep using it, and a depth-using
+        // draw against a depth-less active pass resolves to a new (incompatible)
+        // entry, which the caller's compatibility check turns into a pass split;
+        // the new pass's depth loads DONT_CARE (content was undefined all along).
+        RenderPassEntry& GetOrCreateRenderPass(const MG_State::GLState::FramebufferObject& fbo,
+                                               Uint32 swapchainImageIndex,
+                                               Bool drawUsesDepthStencil = true);
         void QueueRenderbufferClear(GLbitfield mask, const ClearFramebufferPayload& clearPayload,
                                     const MG_State::GLState::FramebufferObject& drawFbo);
         void QueueRenderbufferClear(const ClearAttachmentPayload& clearPayload,
@@ -192,12 +231,20 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         UnorderedMap<Uint64, RenderPassEntry> m_renderPasses;
         // Monotonic frame counter (bumped in OnPresent) for render-pass cache aging.
         Uint64 m_frameCounter = 0;
+        IEvictionObserver* m_evictionObserver = nullptr;
 
         // Bumped whenever a renderbuffer VkImage is (re)created; together with the texture
         // manager's image epoch this invalidates the render-pass fast path on any attachment
         // image recreation.
         Uint64 m_renderbufferImageEpoch = 1;
 
+    public:
+        // Bumped whenever a renderbuffer backing is (re)created; consecutive-draw
+        // snapshots include it so an attachment respecify forces a re-resolve.
+        Uint64 GetRenderbufferImageEpoch() const { return m_renderbufferImageEpoch; }
+
+    private:
+
         // Per-draw fast-path memo for GetOrCreateRenderPass (dirty-flag state tracking): when the
         // framebuffer state is provably unchanged since the last resolution, the active render pass
         // is reused WITHOUT recomputing the expensive per-draw hash. Invalidated by FBO switch /
@@ -210,13 +257,26 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         Uint64 m_rpFastTexEpoch = 0;
         Uint64 m_rpFastRbEpoch = 0;
         Uint64 m_rpFastRenderPassHash = 0;
+        // Whether the memoized entry carries a depth/stencil attachment; a
+        // default-FBO resolution whose effective depth request differs must
+        // miss the memo (the depth-less/depth-full flavors hash differently).
+        Bool m_rpFastHadDepthStencil = false;
 
     public:
         struct RenderbufferResource {
+            // deadSinceFrame sentinel: the owning weak reference has not been observed
+            // expired. Dead resources age past every in-flight frame before Destroy
+            // (see CollectRenderbufferGarbage); the GPU may still reference the image
+            // for frames-in-flight frames after the GL object dies.
+            static constexpr Uint64 kNeverObservedDead = UINT64_MAX;
+
             WeakPtr<MG_State::GLState::RenderbufferObject> renderbuffer;
             VkImage image = VK_NULL_HANDLE;
             VmaAllocation allocation = nullptr;
             VkImageView view = VK_NULL_HANDLE;
+            // UNORM reinterpretation of an sRGB image, used as the attachment view while
+            // GL_FRAMEBUFFER_SRGB is disabled (raw writes). Null for non-sRGB formats.
+            VkImageView unormTwinView = VK_NULL_HANDLE;
             VkImageLayout layout = VK_IMAGE_LAYOUT_UNDEFINED;
             VkFormat format = VK_FORMAT_UNDEFINED;
             VkImageAspectFlags aspect = VK_IMAGE_ASPECT_NONE;
@@ -224,6 +284,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             VkSampleCountFlagBits sampleCount = VK_SAMPLE_COUNT_1_BIT;
             TextureInternalFormat internalFormat = TextureInternalFormat::Unknown;
             Int samples = 0;
+            // m_frameCounter value at which the weak reference was first seen expired.
+            Uint64 deadSinceFrame = kNeverObservedDead;
 
             void Destroy(VkDevice device, VmaAllocator allocator);
         };
@@ -241,12 +303,32 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             ClearAttachmentPayload payload{};
         };
 
+        // A superseded renderbuffer backing (glRenderbufferStorage respecify) parked
+        // until enough frame boundaries have passed that no in-flight command buffer
+        // can still reference it; destroyed in OnPresent (see RetireAgeFrames).
+        struct DeferredRenderbufferRelease {
+            VkImage image = VK_NULL_HANDLE;
+            VmaAllocation allocation = nullptr;
+            VkImageView view = VK_NULL_HANDLE;
+            VkImageView unormTwinView = VK_NULL_HANDLE;
+            Uint64 deferredAtFrame = 0;
+        };
+
         UnorderedMap<MG_State::GLState::RenderbufferObject*, RenderbufferResource> m_renderbufferResources;
         UnorderedMap<MG_State::GLState::RenderbufferObject*, PendingRenderbufferClear> m_pendingRenderbufferClears;
+        Vector<DeferredRenderbufferRelease> m_deferredRenderbufferReleases;
+        // Supported sample counts per attachment format, so per-draw resource lookups
+        // do not repeat vkGetPhysicalDeviceImageFormatProperties.
+        UnorderedMap<VkFormat, VkSampleCountFlags> m_attachmentSampleCountsByFormat;
 
         Bool HasPendingRenderbufferClear(
             const MG_State::GLState::FramebufferAttachmentObject& attachment) const;
         void CollectRenderbufferGarbage();
+        // Frame-boundary margin after which a resource last referenced by a retired
+        // GL object (or superseded backing) is provably past every in-flight frame.
+        Uint64 RetireAgeFrames() const;
+        void DeferRenderbufferBackingRelease(RenderbufferResource& resource);
+        void CollectDeferredRenderbufferReleases(Bool destroyAll);
 
         static inline XXH64_state_t* m_hashState = XXH64_createState();
         static inline ActiveRenderPassInfo s_activeRenderPass{};

MobileGL/MG_Backend/DirectVulkan/Renderer/VkSamplerManager.cpp

@@ -51,6 +51,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         Float ResolveEffectiveMinLod(const MG_State::GLState::SamplerObject& sampler, Float effectiveMaxLod) {
             return std::min(sampler.GetMinLod(), effectiveMaxLod);
         }
+
+        // A single-level view can only ever deliver the base level, but the LOD clamp must not be
+        // collapsed to exactly 0: both GL and Vulkan pick magFilter over minFilter from the
+        // *clamped* lambda, so maxLod = 0 would make every fragment magnify and quietly retire the
+        // min filter. 0.25 is the value VkSamplerCreateInfo's own note prescribes for emulating
+        // GL's non-mipmapped minification - large enough for lambda to stay positive, small enough
+        // that a NEAREST mip mode still rounds down to level 0. Clamped rather than assigned, so a
+        // texture whose GL_TEXTURE_MAX_LOD really is 0 keeps magnifying as GL says it must.
+        Float ResolveSingleLevelMaxLod(const MG_State::GLState::SamplerObject& sampler, Bool singleLevelView) {
+            const Float maxLod = ResolveEffectiveMaxLod(sampler);
+            return singleLevelView ? std::min(maxLod, 0.25f) : maxLod;
+        }
     } // namespace
 
     Bool VkSamplerManager::Initialize(const InitInfo& initInfo) {
@@ -89,15 +101,43 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         m_device = VK_NULL_HANDLE;
         m_config = nullptr;
+        m_frameBoundaryCounter = 0;
+    }
+
+    void VkSamplerManager::OnFrameBoundary() {
+        ++m_frameBoundaryCounter;
+
+        // Sweep occasionally; destroy samplers whose last use is far past every
+        // in-flight frame. Destroy and erase must stay atomic, or Shutdown would
+        // double-free the handle; an evicted key that recurs simply re-creates
+        // its sampler on the next miss.
+        constexpr Uint64 kSweepInterval = 256;
+        constexpr Uint64 kRetireAgeBoundaries = 1024;
+        if ((m_frameBoundaryCounter % kSweepInterval) != 0) {
+            return;
+        }
+
+        for (auto it = m_samplers.begin(); it != m_samplers.end();) {
+            auto& entry = it->second;
+            if (m_frameBoundaryCounter - entry.lastUsedFrameBoundary > kRetireAgeBoundaries) {
+                if (m_device != VK_NULL_HANDLE && entry.handle != VK_NULL_HANDLE) {
+                    vkDestroySampler(m_device, entry.handle, nullptr);
+                }
+                it = m_samplers.erase(it);
+            } else {
+                ++it;
+            }
+        }
     }
 
     Uint64 VkSamplerManager::BuildSamplerKey(const MG_State::GLState::SamplerObject& sampler,
                                              const MG_State::GLState::ITextureObject& texture,
-                                             Bool forceNearestFiltering) const {
+                                             Bool forceNearestFiltering, Bool singleLevelView) const {
         MOBILEGL_ASSERT(m_config != nullptr, "VkSamplerManager::BuildSamplerKey: m_config is null");
         XXHASH_VERIFY(XXH64_reset(m_hashState, m_config->CacheVersion));
 
         XXHASH_VERIFY(XXH64_update(m_hashState, &forceNearestFiltering, sizeof(forceNearestFiltering)));
+        XXHASH_VERIFY(XXH64_update(m_hashState, &singleLevelView, sizeof(singleLevelView)));
 
         const auto minFilter = sampler.GetMinFilter();
         XXHASH_VERIFY(XXH64_update(m_hashState, &minFilter, sizeof(minFilter)));
@@ -111,7 +151,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         XXHASH_VERIFY(XXH64_update(m_hashState, &wrapT, sizeof(wrapT)));
         const auto wrapR = sampler.GetWrapR();
         XXHASH_VERIFY(XXH64_update(m_hashState, &wrapR, sizeof(wrapR)));
-        const auto maxLod = ResolveEffectiveMaxLod(sampler);
+        const auto maxLod = ResolveSingleLevelMaxLod(sampler, singleLevelView);
         const auto minLod = ResolveEffectiveMinLod(sampler, maxLod);
         XXHASH_VERIFY(XXH64_update(m_hashState, &minLod, sizeof(minLod)));
         XXHASH_VERIFY(XXH64_update(m_hashState, &maxLod, sizeof(maxLod)));
@@ -133,10 +173,20 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
     VkSampler VkSamplerManager::GetOrCreateSampler(const MG_State::GLState::SamplerObject& sampler,
                                                    const MG_State::GLState::ITextureObject& texture,
-                                                   Bool forceNearestFiltering) {
-        const Uint64 key = BuildSamplerKey(sampler, texture, forceNearestFiltering);
+                                                   Bool forceNearestFiltering, Uint32 viewLevelCount) {
+        // A view that exposes a single mip level has no second level to blend with, so GL's
+        // *_MIPMAP_* minification filters degenerate to plain filtering on the base level -
+        // sampling is unchanged by pinning the Vulkan sampler to NEAREST mip mode at LOD 0.
+        // It is not cosmetic: MobileGL backs such a view with a fully allocated mip chain whose
+        // tail is never written, and a LINEAR mip mode lets the texture unit issue the level+1
+        // fetch anyway. On Adreno that fetch lands in uninitialized UBWC pages (or past the
+        // allocation for a genuinely single-level image) and faults the GPU - the same failure
+        // the default-framebuffer blit shader had to work around with an explicit-LOD sample.
+        const Bool singleLevelView = viewLevelCount == 1;
+        const Uint64 key = BuildSamplerKey(sampler, texture, forceNearestFiltering, singleLevelView);
         auto it = m_samplers.find(key);
         if (it != m_samplers.end()) {
+            it->second.lastUsedFrameBoundary = m_frameBoundaryCounter;
             return it->second.handle;
         }
 
@@ -144,8 +194,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         samplerInfo.sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO;
         samplerInfo.magFilter = forceNearestFiltering ? VK_FILTER_NEAREST : ToVkFilter(sampler.GetMagFilter());
         samplerInfo.minFilter = forceNearestFiltering ? VK_FILTER_NEAREST : ToVkFilter(sampler.GetMinFilter());
-        samplerInfo.mipmapMode = forceNearestFiltering ? VK_SAMPLER_MIPMAP_MODE_NEAREST
-                                                       : ToVkMipmapMode(sampler.GetMipmapMode());
+        samplerInfo.mipmapMode = (forceNearestFiltering || singleLevelView)
+                                     ? VK_SAMPLER_MIPMAP_MODE_NEAREST
+                                     : ToVkMipmapMode(sampler.GetMipmapMode());
         samplerInfo.addressModeU = ToVkAddressMode(sampler.GetWrapS());
         samplerInfo.addressModeV = ToVkAddressMode(sampler.GetWrapT());
         samplerInfo.addressModeW = ToVkAddressMode(sampler.GetWrapR());
@@ -157,7 +208,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         samplerInfo.maxAnisotropy = maxAnisotropy;
         samplerInfo.compareEnable = sampler.GetCompareMode() == SamplerCompareMode::CompareToTexture ? VK_TRUE : VK_FALSE;
         samplerInfo.compareOp = ToVkCompareOp(ResolveCompareFunc(sampler, texture));
-        samplerInfo.maxLod = ResolveEffectiveMaxLod(sampler);
+        // Must match BuildSamplerKey's resolution exactly.
+        samplerInfo.maxLod = ResolveSingleLevelMaxLod(sampler, singleLevelView);
         samplerInfo.minLod = ResolveEffectiveMinLod(sampler, samplerInfo.maxLod);
         samplerInfo.borderColor = ResolveVkBorderColor(sampler, texture);
         samplerInfo.unnormalizedCoordinates = VK_FALSE;
@@ -169,6 +221,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         entry.handle = vkSampler;
         entry.externalIndex = sampler.GetExternalIndex();
         entry.version = sampler.GetVersion();
+        entry.lastUsedFrameBoundary = m_frameBoundaryCounter;
         m_samplers[key] = entry;
         return vkSampler;
     }

MobileGL/MG_Backend/DirectVulkan/Renderer/VkSamplerManager.h

@@ -33,20 +33,38 @@ public:
     Bool Initialize(const InitInfo& initInfo);
     void Shutdown();
 
+    // viewLevelCount is the mip-level count of the image view this sampler will be paired
+    // with; 0 means "unknown, do not narrow". See GetOrCreateSampler for why it matters.
     VkSampler GetOrCreateSampler(const MG_State::GLState::SamplerObject& sampler,
                                  const MG_State::GLState::ITextureObject& texture,
-                                 Bool forceNearestFiltering = false);
+                                 Bool forceNearestFiltering = false,
+                                 Uint32 viewLevelCount = 0);
+    // Frame boundary hook: ages the sampler cache and destroys samplers not used
+    // for many frames. The key hashes continuous float state (lodBias, LOD clamps,
+    // anisotropy), so an app animating those would otherwise mint an unbounded
+    // stream of never-destroyed VkSamplers and eventually exhaust the device's
+    // maxSamplerAllocationCount. A sampler idle for over a thousand frame
+    // boundaries cannot be referenced by any in-flight command buffer (frames in
+    // flight are single digits), and every descriptor set the GPU consumes is
+    // written that same frame with live handles (the per-binding resolve memo and
+    // descriptor-set reuse are both frame-reset), so destruction here needs no
+    // fence wait. Self-gated: one counter bump and compare except on sweep
+    // boundaries.
+    void OnFrameBoundary();
 
 private:
     struct SamplerCacheEntry {
         VkSampler handle = VK_NULL_HANDLE;
         Uint externalIndex = 0;
         Uint16 version = 0;
+        // Frame boundary of the last cache hit; entries idle past the
+        // OnFrameBoundary retirement age have their VkSampler destroyed.
+        Uint64 lastUsedFrameBoundary = 0;
     };
 
     Uint64 BuildSamplerKey(const MG_State::GLState::SamplerObject& sampler,
                            const MG_State::GLState::ITextureObject& texture,
-                           Bool forceNearestFiltering) const;
+                           Bool forceNearestFiltering, Bool singleLevelView) const;
     static VkFilter ToVkFilter(SamplerFilterMode mode);
     static VkSamplerMipmapMode ToVkMipmapMode(SamplerMipmapMode mode);
     static VkSamplerAddressMode ToVkAddressMode(SamplerWrapMode mode);
@@ -67,6 +85,8 @@ private:
     Bool m_samplerAnisotropySupported = false;
     Float m_maxSamplerAnisotropy = 1.0f;
     UnorderedMap<Uint64, SamplerCacheEntry> m_samplers;
+    // Monotonic frame-boundary counter (bumped in OnFrameBoundary) for cache aging.
+    Uint64 m_frameBoundaryCounter = 0;
     static inline XXH64_state_t* m_hashState = XXH64_createState();
 };
 } // namespace MobileGL::MG_Backend::DirectVulkan

MobileGL/MG_Backend/DirectVulkan/Renderer/VkTextureManager.cpp

@@ -120,31 +120,21 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     }
 
     static Bool TryResolveSampleCountFlagBits(Int requestedSamples, VkSampleCountFlagBits& outSampleCount) {
-        switch (requestedSamples) {
-        case 1:
+        // GL promises "at least the requested samples", so a non-power-of-two
+        // request (legal in GL, e.g. 3) rounds up to the next Vulkan bit.
+        if (requestedSamples <= 1) {
             outSampleCount = VK_SAMPLE_COUNT_1_BIT;
             return true;
-        case 2:
-            outSampleCount = VK_SAMPLE_COUNT_2_BIT;
-            return true;
-        case 4:
-            outSampleCount = VK_SAMPLE_COUNT_4_BIT;
-            return true;
-        case 8:
-            outSampleCount = VK_SAMPLE_COUNT_8_BIT;
-            return true;
-        case 16:
-            outSampleCount = VK_SAMPLE_COUNT_16_BIT;
-            return true;
-        case 32:
-            outSampleCount = VK_SAMPLE_COUNT_32_BIT;
-            return true;
-        case 64:
-            outSampleCount = VK_SAMPLE_COUNT_64_BIT;
-            return true;
-        default:
+        }
+        if (requestedSamples > 64) {
             return false;
         }
+        Uint32 bit = 1;
+        while (bit < static_cast<Uint32>(requestedSamples)) {
+            bit <<= 1;
+        }
+        outSampleCount = static_cast<VkSampleCountFlagBits>(bit);
+        return true;
     }
 
     static Bool IsCubeMapFaceUploadTarget(TextureUploadTarget target) {
@@ -587,6 +577,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_allocator = initInfo.allocator;
         m_commandPool = initInfo.commandPool;
         m_graphicsQueue = initInfo.graphicsQueue;
+        m_imageFormatListSupported = initInfo.imageFormatListSupported;
         m_currentFrameIndex = 0;
         m_deferredReleases.clear();
         m_deferredReleases.resize(initInfo.frameCount);
@@ -606,9 +597,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     }
 
     void VkTextureManager::Shutdown() {
+        if (m_device != VK_NULL_HANDLE) {
+            ReclaimCompletedUploads(/*waitAll=*/true);
+        }
         DestroyDeferredReleases();
+        ++m_resourceEraseEpoch;  // every memoized resource pointer dies with the map
         m_textureResources.clear();
         m_aliveObjects.clear();
+        m_storageImageTextures.clear();
 
         m_device = VK_NULL_HANDLE;
         m_physicalDevice = VK_NULL_HANDLE;
@@ -627,6 +623,27 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                         frameIndex, m_deferredViewReleases.size());
         m_currentFrameIndex = frameIndex;
         CollectDeferredReleases(frameIndex);
+        ReclaimCompletedUploads();
+
+        // Frame-boundary GC: every 64 frame boundaries (~1 s at 60 fps) bounds the reclaim
+        // latency for dead textures regardless of draw traffic — workloads that churn
+        // textures through clears/readbacks alone never reach the draw-gated
+        // CollectGarbage. Must run after CollectDeferredReleases above: the prune defers
+        // its releases into this frame's slot, which was just drained, so they are
+        // destroyed only after the slot's fence has been waited again one full frame-ring
+        // cycle from now (never while an in-flight frame may still reference them).
+        constexpr Uint32 kGcFrameInterval = 64;
+        ++m_gcFrameCounter;
+        if (m_gcFrameCounter % kGcFrameInterval == 0) {
+            PruneDeadTextures();
+        }
+    }
+
+    void VkTextureManager::CollectAllDeferredReleases() {
+        const SizeT frameCount = std::min(m_deferredReleases.size(), m_deferredViewReleases.size());
+        for (SizeT frameIndex = 0; frameIndex < frameCount; ++frameIndex) {
+            CollectDeferredReleases(static_cast<Uint32>(frameIndex));
+        }
     }
 
     void VkTextureManager::EraseTrackedTexture(const TextureIdentity& identity) {
@@ -636,6 +653,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             m_textureResources.erase(resourceIt);
         }
         m_aliveObjects.erase(identity);
+        m_storageImageTextures.erase(identity);
+        // Invalidate every cross-draw sampled-texture memo: the erased
+        // resource's address may be reused by a future emplace.
+        ++m_resourceEraseEpoch;
     }
 
     void VkTextureManager::PruneStaleTextureAliases(MG_State::GLState::ITextureObject* texture) {
@@ -691,32 +712,63 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             }
         }
 
-        auto aliveIt = m_aliveObjects.find(identity);
-        if (aliveIt != m_aliveObjects.end() && aliveIt->second.expired()) {
-            EraseTrackedTexture(aliveIt->first);
-            aliveIt = m_aliveObjects.end();
-        }
-
-        // Only (re)register and prune when this (texture, lifetime) pair is new: stale
-        // aliases can only come into existence through an address reuse, which by
-        // construction introduces a new identity. Doing this unconditionally made every
-        // sampled-texture sync scan the entire alive-texture map per draw.
-        if (aliveIt == m_aliveObjects.end()) {
-            const auto& liveTexture = MG_State::pGLContext->GetTextureObject(texture.GetExternalIndex());
-            if (liveTexture && liveTexture.get() == &texture) {
-                m_aliveObjects[identity] = WeakPtr<MG_State::GLState::ITextureObject>(liveTexture);
-                PruneStaleTextureAliases(&texture);
+        // Cross-draw memo probe (see SyncedTextureMemoEntry): skips both map
+        // lookups and the (re)registration path for repeat-bound textures.
+        TextureResource* resourcePtr = nullptr;
+        for (Uint32 i = 0; i < kSyncedTextureMemoSize; ++i) {
+            const SyncedTextureMemoEntry& memo = m_syncedTextureMemo[i];
+            if (memo.texture == &texture && memo.lifetimeId == identity.lifetimeId &&
+                memo.eraseEpoch == m_resourceEraseEpoch) {
+                resourcePtr = memo.resource;
+                break;
             }
         }
 
-        auto it = m_textureResources.find(identity);
-        if (it == m_textureResources.end()) {
-            TextureResource initial{};
-            auto [insertIt, _] = m_textureResources.emplace(identity, Move(initial));
-            it = insertIt;
+        if (resourcePtr == nullptr) {
+            auto aliveIt = m_aliveObjects.find(identity);
+            if (aliveIt != m_aliveObjects.end() && aliveIt->second.expired()) {
+                EraseTrackedTexture(aliveIt->first);
+                aliveIt = m_aliveObjects.end();
+            }
+
+            // Only (re)register and prune when this (texture, lifetime) pair is new: stale
+            // aliases can only come into existence through an address reuse, which by
+            // construction introduces a new identity. Doing this unconditionally made every
+            // sampled-texture sync scan the entire alive-texture map per draw.
+            if (aliveIt == m_aliveObjects.end()) {
+                WeakPtr<MG_State::GLState::ITextureObject> aliveTexture;
+                const auto& liveTexture = MG_State::pGLContext->GetTextureObject(texture.GetExternalIndex());
+                if (liveTexture && liveTexture.get() == &texture) {
+                    aliveTexture = liveTexture;
+                } else {
+                    // The name lookup legally fails while the object is alive: the name was
+                    // deleted with the texture still attached to an FBO (the attachment's
+                    // SharedPtr keeps it alive), or the name was reused by a new texture, or
+                    // this is a default texture object (name 0 lives outside the name map).
+                    // Register through the object's own control block so the resource created
+                    // below still participates in weak-expiry GC instead of becoming an
+                    // orphan no reclamation path can reach until Shutdown.
+                    aliveTexture = texture.weak_from_this();
+                }
+                if (!aliveTexture.expired()) {
+                    m_aliveObjects[identity] = Move(aliveTexture);
+                    PruneStaleTextureAliases(&texture);
+                }
+            }
+
+            auto it = m_textureResources.find(identity);
+            if (it == m_textureResources.end()) {
+                TextureResource initial{};
+                auto [insertIt, _] = m_textureResources.emplace(identity, Move(initial));
+                it = insertIt;
+            }
+            resourcePtr = &(it->second);
+            m_syncedTextureMemo[m_syncedTextureMemoNext] =
+                SyncedTextureMemoEntry{&texture, identity.lifetimeId, m_resourceEraseEpoch, resourcePtr};
+            m_syncedTextureMemoNext = (m_syncedTextureMemoNext + 1) % kSyncedTextureMemoSize;
         }
 
-        if (!SyncTexture(texture, it->second)) {
+        if (!SyncTexture(texture, *resourcePtr)) {
             MGLOG_D("%s: Syncing texture %d failed", __func__, texture.GetExternalIndex());
             return nullptr;
         }
@@ -730,11 +782,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 }
             }
             if (!recorded) {
-                m_drawSyncedThisDraw.push_back({identity, &(it->second)});
+                m_drawSyncedThisDraw.push_back({identity, resourcePtr});
             }
         }
 
-        return &(it->second);
+        return resourcePtr;
     }
 
     VkImageView VkTextureManager::GetOrCreateViewAtMipLevel(MG_State::GLState::ITextureObject& texture, Uint32 mipLevel) {
@@ -778,7 +830,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return VK_NULL_HANDLE;
         }
 
-        if (baseArrayLayer == 0 && layerCount == resource->arrayLayers && viewType == resource->viewType) {
+        const Bool framebufferSrgbEnabled =
+            MG_State::pGLContext->IsCapabilityEnabled(MobileGL::CapabilityInput::FramebufferSrgb);
+        const VkFormat attachmentFormat = ResolveSrgbAttachmentWriteFormat(resource->format, framebufferSrgbEnabled);
+
+        if (attachmentFormat == resource->format && baseArrayLayer == 0 && layerCount == resource->arrayLayers &&
+            viewType == resource->viewType) {
             return GetOrCreateViewAtMipLevel(texture, mipLevel);
         }
 
@@ -787,6 +844,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             .baseArrayLayer = baseArrayLayer,
             .layerCount = layerCount,
             .viewType = viewType,
+            .viewFormat = attachmentFormat,
         };
         auto it = resource->attachmentViews.find(key);
         if (it == resource->attachmentViews.end()) {
@@ -797,7 +855,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return attachmentView;
         }
 
-        attachmentView = CreateImageView(resource->image, resource->format, resource->aspect, viewType,
+        attachmentView = CreateImageView(resource->image, attachmentFormat, resource->aspect, viewType,
                                          mipLevel, 1, baseArrayLayer, layerCount);
         if (attachmentView == VK_NULL_HANDLE) {
             MGLOG_D("%s: CreateImageView failed for textureId=%d mipLevel=%u baseArrayLayer=%u layerCount=%u viewType=%d",
@@ -1013,6 +1071,16 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         return view;
     }
 
+    void VkTextureManager::StampTextureRecordingUse(MG_State::GLState::ITextureObject* texture) {
+        if (texture == nullptr) {
+            return;
+        }
+        auto it = m_textureResources.find(MakeTextureIdentity(texture));
+        if (it != m_textureResources.end()) {
+            it->second.lastRecordingGeneration = m_recordingGeneration;
+        }
+    }
+
     void VkTextureManager::UpdateTrackedImageLayout(MG_State::GLState::ITextureObject* texture, VkImageLayout newLayout) {
         MOBILEGL_ASSERT(texture != nullptr, "UpdateTrackedImageLayout: texture is null");
         auto it = m_textureResources.find(MakeTextureIdentity(texture));
@@ -1040,6 +1108,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         MOBILEGL_ASSERT(writtenMipLevel < resource.mipLevels,
                         "UpdateTrackedImageLayoutAfterAttachmentWrite: textureId=%d mipLevel=%u out of range %u",
                         texture->GetExternalIndex(), writtenMipLevel, resource.mipLevels);
+        // Pre-pass stream bookkeeping: the render pass that just ended wrote this image.
+        StampResourceRecordingUse(resource);
 
         if (resource.layout != newLayout && resource.mipLevels > 1) {
             VkPipelineStageFlags srcStageMask = VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT;
@@ -1124,6 +1194,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                               VK_ACCESS_SHADER_READ_BIT, resource->aspect, 0, resource->mipLevels,
                                               resource->arrayLayers);
         MOBILEGL_ASSERT(ok, "TransitionTextureForSampling: transition failed for textureId=%d", texture.GetExternalIndex());
+        // Pre-pass stream bookkeeping: a command referencing the image was recorded.
+        StampResourceRecordingUse(*resource);
         return ok;
     }
 
@@ -1153,11 +1225,44 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                               resource->aspect, 0, resource->mipLevels, resource->arrayLayers);
         MOBILEGL_ASSERT(ok, "TransitionTextureForStorageImage: transition failed for textureId=%d",
                         texture.GetExternalIndex());
+        // Pre-pass stream bookkeeping: a command referencing the image was recorded.
+        StampResourceRecordingUse(*resource);
         return ok;
     }
 
+    void VkTextureManager::MarkStorageImageTexture(MG_State::GLState::ITextureObject& texture) {
+        m_storageImageTextures.insert(MakeTextureIdentity(&texture));
+    }
+
+    Bool VkTextureManager::NeedsStorageUsageUpgrade(MG_State::GLState::ITextureObject& texture) const {
+        const TextureIdentity identity = MakeTextureIdentity(&texture);
+        if (m_storageImageTextures.find(identity) == m_storageImageTextures.end()) {
+            return false;
+        }
+        const auto it = m_textureResources.find(identity);
+        // No image yet: the first sync creates it with STORAGE straight away, so there is nothing
+        // to preserve and nothing to order against.
+        return it != m_textureResources.end() && it->second.image != VK_NULL_HANDLE &&
+               !it->second.storageUsageResolved;
+    }
+
+    Bool VkTextureManager::NeedsMipChainGrowth(MG_State::GLState::ITextureObject& texture) const {
+        const TextureIdentity identity = MakeTextureIdentity(&texture);
+        const auto it = m_textureResources.find(identity);
+        // No image yet: the first sync sizes the chain from the levels the texture already
+        // defines, so nothing is recreated and there is nothing to order against.
+        if (it == m_textureResources.end() || it->second.image == VK_NULL_HANDLE) {
+            return false;
+        }
+        const TextureResource& resource = it->second;
+        const IntVec3 extent = {static_cast<Int>(resource.extent.width), static_cast<Int>(resource.extent.height),
+                                static_cast<Int>(resource.depth)};
+        return resource.mipLevels < ComputeFullMipLevelCount(extent);
+    }
+
     Bool VkTextureManager::NeedsStorageImagePreparation(MG_State::GLState::ITextureObject& texture) const {
-        const auto it = m_textureResources.find(MakeTextureIdentity(&texture));
+        const TextureIdentity identity = MakeTextureIdentity(&texture);
+        const auto it = m_textureResources.find(identity);
         if (it == m_textureResources.end()) {
             return true;
         }
@@ -1165,6 +1270,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         if (resource.image == VK_NULL_HANDLE || resource.layout != VK_IMAGE_LAYOUT_GENERAL) {
             return true;
         }
+        // The image predates this texture's first image-unit binding, so it was created without
+        // STORAGE usage and has to be recreated - which is illegal inside a render pass.
+        if (!resource.storageUsageResolved &&
+            m_storageImageTextures.find(identity) != m_storageImageTextures.end()) {
+            return true;
+        }
         // Mirror SyncTexture's cross-draw skip condition: any version drift means the sync
         // path may upload or rebuild, both of which need the render pass ended first.
         const auto* mipTexture = MG_State::GLState::AsMipmapTexture(&texture);
@@ -1212,10 +1323,22 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     }
 
     SizeT VkTextureManager::CollectGarbage() {
+        // Draw-gated stagger (1 in 256 calls): keeps the per-draw cost at one counter
+        // bump. The guaranteed reclaim path is the frame-boundary prune in BeginFrame;
+        // this remains as a cheap assist so draw-heavy workloads reclaim sooner.
         m_gcCounter++;
         if (m_gcCounter != 0) {
             return 0;
         }
+        return PruneDeadTextures();
+    }
+
+    SizeT VkTextureManager::PruneDeadTextures() {
+        // Erasing entries would dangle the raw TextureResource pointers memoized for the
+        // current draw; every call path (BeginFrame, and CollectGarbage at the top of a
+        // freshly opened draw-sync scope) runs before any memo entry is recorded.
+        MOBILEGL_ASSERT(m_drawSyncedThisDraw.empty(),
+                        "PruneDeadTextures: draw-sync memo holds raw resource pointers an erase would dangle");
 
         Vector<MG_State::GLState::ITextureObject*> expiredTextures;
         expiredTextures.reserve(m_aliveObjects.size());
@@ -1227,7 +1350,25 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         for (auto* texture : expiredTextures) {
             PruneStaleTextureAliases(texture);
         }
-        return expiredTextures.size();
+        SizeT prunedCount = expiredTextures.size();
+
+        // Orphan sweep: after the pass above, m_aliveObjects holds only live entries.
+        // Registration in SyncTextureAndGetDescriptor cannot fail for a SharedPtr-owned
+        // texture (weak_from_this fallback), so a resource whose identity has no alive
+        // entry has no trackable owner: its GL-side object is gone, or was never
+        // shared-owned, in which case recreation on a later sync is the safe fallback.
+        // Destruction goes through the per-frame deferred queues, never immediate.
+        Vector<TextureIdentity> orphanIdentities;
+        for (auto it = m_textureResources.begin(); it != m_textureResources.end(); ++it) {
+            if (m_aliveObjects.find(it->first) == m_aliveObjects.end()) {
+                orphanIdentities.emplace_back(it->first);
+            }
+        }
+        for (const auto& identity : orphanIdentities) {
+            EraseTrackedTexture(identity);
+        }
+        prunedCount += orphanIdentities.size();
+        return prunedCount;
     }
 
     Bool VkTextureManager::SyncTexture(MG_State::GLState::ITextureObject &texture,
@@ -1241,7 +1382,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         const auto* syncingMipTexture = MG_State::GLState::AsMipmapTexture(&texture);
         const Uint32 syncingMipLevelCount =
             syncingMipTexture != nullptr ? syncingMipTexture->GetMipmapLevelCount() : 0u;
-        if (outResource.image != VK_NULL_HANDLE &&
+        // A pending storage-usage upgrade also has to bust the skip: nothing about the texture's
+        // content or params changed, but the image itself must be recreated with STORAGE usage
+        // before it can back an image-unit descriptor.
+        const Bool storageUpgradePending =
+            !outResource.storageUsageResolved &&
+            m_storageImageTextures.find(MakeTextureIdentity(&texture)) != m_storageImageTextures.end();
+        if (outResource.image != VK_NULL_HANDLE && !storageUpgradePending &&
             outResource.syncedContentVersion == syncingContentVersion &&
             outResource.syncedTextureParamsVersion == texture.GetTextureParamsVersion() &&
             outResource.syncedMipLevelCount == syncingMipLevelCount) {
@@ -1311,11 +1458,23 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                                const IntVec3 &texelSize, SizeT byteSize, Uint32 mipLevels,
                                                TextureResource &resource) {
         const TextureFormatInfo formatInfo = ResolveTextureFormatInfo(texture.GetFormat());
-        const VkFormat format = formatInfo.format;
+        VkFormat format = formatInfo.format;
         if (format == VK_FORMAT_UNDEFINED) {
             MGLOG_D("%s: format == VK_FORMAT_UNDEFINED", __func__);
             return false;
         }
+        // X8_D24 lacks optimal-tiling support on several drivers (lavapipe included);
+        // D32_SFLOAT holds every 24-bit depth value exactly, and the upload path
+        // converts the shadow words to float (see the pure-depth branch below).
+        if (format == VK_FORMAT_X8_D24_UNORM_PACK32) {
+            VkFormatProperties formatProperties{};
+            vkGetPhysicalDeviceFormatProperties(m_physicalDevice, format, &formatProperties);
+            constexpr VkFormatFeatureFlags kDepthAttachmentAndSample =
+                VK_FORMAT_FEATURE_DEPTH_STENCIL_ATTACHMENT_BIT | VK_FORMAT_FEATURE_SAMPLED_IMAGE_BIT;
+            if ((formatProperties.optimalTilingFeatures & kDepthAttachmentAndSample) != kDepthAttachmentAndSample) {
+                format = VK_FORMAT_D32_SFLOAT;
+            }
+        }
         if (texelSize.x() <= 0 || texelSize.y() <= 0 /*|| byteSize == 0*/) {
             MGLOG_D("%s: texelSize or byteSize is zero", __func__);
             return false;
@@ -1325,8 +1484,17 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return false;
         }
         const Bool isMultisampleTexture = IsMultisampleTextureUploadTarget(uploadTarget);
+        // A texture that has only ever defined level 0 gets a single-level backing
+        // (ANGLE's model). Preallocating the full chain put every render target
+        // onto Adreno's multi-mip image layout and grew each texture by a third
+        // for levels most textures never define. Once a second level is defined
+        // the backing is recreated ONE time with the full chain (the
+        // preserve-copy path below carries the pixels over), so sequentially-
+        // defined atlas mips do not recreate per level, and glGenerateMipmap -
+        // which defines every level before syncing - works unchanged.
         const Uint32 backingMipLevels =
-            isMultisampleTexture ? 1u : std::max(mipLevels, ComputeFullMipLevelCount(texelSize));
+            isMultisampleTexture ? 1u
+            : (mipLevels > 1 ? std::max(mipLevels, ComputeFullMipLevelCount(texelSize)) : 1u);
         TextureShapeInfo shapeInfo{};
         const Bool supportedShape = TryResolveTextureShapeInfo(texture, uploadTarget, texelSize, shapeInfo);
         MOBILEGL_ASSERT(supportedShape,
@@ -1352,15 +1520,89 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         const VkImageAspectFlags aspect = GetAspectMaskForFormat(format);
         VkFormatProperties formatProperties{};
         vkGetPhysicalDeviceFormatProperties(m_physicalDevice, format, &formatProperties);
-        const Bool supportsStorageImage =
+        // Only textures that have actually been bound to a GL image unit get STORAGE usage (and
+        // the MUTABLE_FORMAT it drags in for format-reinterpreting image views). Requesting it
+        // for every storage-capable colour texture costs real bandwidth: Adreno cannot keep UBWC
+        // compression on an image that may be written through a storage descriptor, so the whole
+        // render target - MC's included - runs uncompressed. MarkStorageImageTexture upgrades a
+        // texture before its first image-unit draw, and the usage below feeds the compatibility
+        // check so the upgrade recreates the image.
+        const Bool markedAsStorageImage =
+            m_storageImageTextures.find(MakeTextureIdentity(
+                const_cast<MG_State::GLState::ITextureObject*>(&texture))) != m_storageImageTextures.end();
+        // Storage-image CAPABILITY (does the format allow it at all) is deliberately separate from
+        // whether this texture actually needs the usage. MUTABLE_FORMAT keys off capability, as
+        // before: format-reinterpreting views are not a storage-only concern - the SAMPLED path
+        // needs them too (GetOrCreateSampledImageView bails out without it, see ~line 892), so
+        // tying MUTABLE_FORMAT to the image-unit mark would break sampled format reinterpretation
+        // for every texture that never becomes a storage image.
+        const Bool storageImageCapable =
             !isMultisampleTexture &&
             (aspect & VK_IMAGE_ASPECT_COLOR_BIT) != 0 &&
             (formatProperties.optimalTilingFeatures & VK_FORMAT_FEATURE_STORAGE_IMAGE_BIT) != 0;
+        const Bool supportsStorageImage = storageImageCapable && markedAsStorageImage;
         VkImageCreateFlags imageCreateFlags = shapeInfo.imageFlags;
-        if (supportsStorageImage && IsMutableStorageImageFormat(format) &&
+        if (storageImageCapable && IsMutableStorageImageFormat(format) &&
             m_mutableFormatUnsupported.find(format) == m_mutableFormatUnsupported.end()) {
             imageCreateFlags |= VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT;
         }
+        // sRGB color images attach through their UNORM twin while GL_FRAMEBUFFER_SRGB is
+        // disabled (see ResolveSrgbAttachmentWriteFormat), which needs format-reinterpreting
+        // views - multisample sRGB render targets included.
+        if (ResolveSrgbAttachmentWriteFormat(format, false) != format &&
+            (aspect & VK_IMAGE_ASPECT_COLOR_BIT) != 0 &&
+            m_mutableFormatUnsupported.find(format) == m_mutableFormatUnsupported.end()) {
+            imageCreateFlags |= VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT;
+        }
+
+        VkImageUsageFlags desiredUsage =
+            VK_IMAGE_USAGE_SAMPLED_BIT |
+            (supportsStorageImage ? VK_IMAGE_USAGE_STORAGE_BIT : 0) |
+            ((aspect & VK_IMAGE_ASPECT_COLOR_BIT) ? VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT : 0) |
+            (((aspect & VK_IMAGE_ASPECT_DEPTH_BIT) || (aspect & VK_IMAGE_ASPECT_STENCIL_BIT)) ?
+                 VK_IMAGE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT :
+                 0);
+        if (!isMultisampleTexture) {
+            desiredUsage |= VK_IMAGE_USAGE_TRANSFER_DST_BIT | VK_IMAGE_USAGE_TRANSFER_SRC_BIT;
+        }
+
+        // Round a multisample request up to a count the device supports for this
+        // format (GL only promises "at least"), mirroring the renderbuffer path.
+        if (isMultisampleTexture && resolvedSampleCount != VK_SAMPLE_COUNT_1_BIT) {
+            auto supportedIt = m_multisampleCountsByFormat.find(format);
+            if (supportedIt == m_multisampleCountsByFormat.end()) {
+                VkImageFormatProperties imageFormatProperties{};
+                VkSampleCountFlags supported = VK_SAMPLE_COUNT_1_BIT;
+                if (vkGetPhysicalDeviceImageFormatProperties(m_physicalDevice, format, shapeInfo.imageType,
+                                                             VK_IMAGE_TILING_OPTIMAL, desiredUsage, imageCreateFlags,
+                                                             &imageFormatProperties) == VK_SUCCESS) {
+                    supported = imageFormatProperties.sampleCounts;
+                }
+                supportedIt = m_multisampleCountsByFormat.emplace(format, supported).first;
+            }
+            const VkSampleCountFlags supported = supportedIt->second;
+            if ((supported & resolvedSampleCount) == 0) {
+                Uint32 rounded = 0;
+                for (Uint32 bit = static_cast<Uint32>(resolvedSampleCount) << 1; bit <= VK_SAMPLE_COUNT_64_BIT;
+                     bit <<= 1) {
+                    if ((supported & bit) != 0) {
+                        rounded = bit;
+                        break;
+                    }
+                }
+                if (rounded == 0) {
+                    for (Uint32 bit = static_cast<Uint32>(resolvedSampleCount) >> 1; bit != 0; bit >>= 1) {
+                        if ((supported & bit) != 0) {
+                            rounded = bit;
+                            break;
+                        }
+                    }
+                }
+                if (rounded != 0) {
+                    resolvedSampleCount = static_cast<VkSampleCountFlagBits>(rounded);
+                }
+            }
+        }
 
         const Bool compatible = resource.image != VK_NULL_HANDLE && resource.format == format &&
                                 resource.extent.width == static_cast<Uint32>(texelSize.x()) &&
@@ -1370,6 +1612,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                 resource.viewType == shapeInfo.viewType &&
                                 resource.sampleCount == resolvedSampleCount &&
                                 resource.imageCreateFlags == imageCreateFlags &&
+                                resource.usageFlags == desiredUsage &&
                                 resource.mipLevels == backingMipLevels;
         if (compatible) {
             if (resource.perMipViews.size() != backingMipLevels) {
@@ -1378,6 +1621,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             if (resource.perMipSampledViews.size() != backingMipLevels) {
                 resource.perMipSampledViews.resize(backingMipLevels, VK_NULL_HANDLE);
             }
+            // Keeping the image is itself the answer to the mark: either it already carries
+            // STORAGE, or this format can never carry it. Either way there is nothing left to
+            // recreate, so stop reporting the texture as needing preparation.
+            resource.storageUsageResolved = markedAsStorageImage;
             return true;
         }
 
@@ -1392,7 +1639,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             resource.sampleCount == resolvedSampleCount &&
             resource.imageCreateFlags == imageCreateFlags &&
             resolvedSampleCount == VK_SAMPLE_COUNT_1_BIT &&
-            resource.mipLevels < backingMipLevels &&
+            // '<=' rather than '<': a storage-usage upgrade recreates the image with an
+            // unchanged mip count, and its contents (a render target's pixels live only on the
+            // GPU) still have to survive. The vkCmdCopyImage below copies min(mipLevels).
+            resource.mipLevels <= backingMipLevels &&
             resource.layout != VK_IMAGE_LAYOUT_UNDEFINED;
 
         std::unique_ptr<TextureResource> preservedResource;
@@ -1414,16 +1664,37 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         imageInfo.format = format;
         imageInfo.tiling = VK_IMAGE_TILING_OPTIMAL;
         imageInfo.initialLayout = VK_IMAGE_LAYOUT_UNDEFINED;
-        imageInfo.usage = VK_IMAGE_USAGE_SAMPLED_BIT |
-                          (supportsStorageImage ? VK_IMAGE_USAGE_STORAGE_BIT : 0) |
-                          ((aspect & VK_IMAGE_ASPECT_COLOR_BIT) ? VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT : 0) |
-                          (((aspect & VK_IMAGE_ASPECT_DEPTH_BIT) || (aspect & VK_IMAGE_ASPECT_STENCIL_BIT)) ?
-                               VK_IMAGE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT :
-                               0);
-        if (!isMultisampleTexture) {
-            imageInfo.usage |= VK_IMAGE_USAGE_TRANSFER_DST_BIT | VK_IMAGE_USAGE_TRANSFER_SRC_BIT;
-        }
+        imageInfo.usage = desiredUsage;
         imageInfo.samples = resolvedSampleCount;
+
+        // Bound the mutability. A blindly-mutable image has to be laid out so that ANY format in
+        // its compatibility class can be viewed, which costs bandwidth compression on tilers;
+        // naming the exact set instead lets the driver keep it. Only safe when that set really is
+        // exhaustive, so it is restricted to textures that are not image-unit bound: sampled views
+        // can only ever ask for ResolveSampledImageViewFormat's output, whereas glBindImageTexture
+        // may name any compatible format, which nothing here can enumerate ahead of time.
+        Vector<VkFormat> viewFormats;
+        VkImageFormatListCreateInfo formatListInfo{};
+        if (m_imageFormatListSupported && !supportsStorageImage &&
+            (imageInfo.flags & VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT) != 0) {
+            viewFormats.push_back(format);
+            for (const SamplerNumericDomain domain : {SamplerNumericDomain::Float,
+                                                      SamplerNumericDomain::SignedInteger,
+                                                      SamplerNumericDomain::UnsignedInteger}) {
+                const VkFormat viewFormat = ResolveSampledImageViewFormat(format, domain);
+                if (viewFormat == VK_FORMAT_UNDEFINED) {
+                    continue;
+                }
+                if (std::find(viewFormats.begin(), viewFormats.end(), viewFormat) == viewFormats.end()) {
+                    viewFormats.push_back(viewFormat);
+                }
+            }
+            formatListInfo.sType = VK_STRUCTURE_TYPE_IMAGE_FORMAT_LIST_CREATE_INFO;
+            formatListInfo.viewFormatCount = static_cast<Uint32>(viewFormats.size());
+            formatListInfo.pViewFormats = viewFormats.data();
+            imageInfo.pNext = &formatListInfo;
+        }
+
         if (isMultisampleTexture || (imageInfo.flags & VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT) != 0) {
             VkImageFormatProperties imageFormatProperties{};
             VkResult imageFormatResult = vkGetPhysicalDeviceImageFormatProperties(
@@ -1462,8 +1733,21 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         VmaAllocationCreateInfo allocationInfo{};
         allocationInfo.usage = VMA_MEMORY_USAGE_AUTO_PREFER_DEVICE;
         allocationInfo.requiredFlags = VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT;
-        VK_VERIFY(vmaCreateImage(m_allocator, &imageInfo, &allocationInfo, &resource.image, &resource.allocation, nullptr),
-                  "vmaCreateImage(texture)");
+        // Soft failure like the unsupported-sample-count path above: a driver can pass the
+        // vkGetPhysicalDeviceImageFormatProperties pre-check yet still refuse the creation
+        // (e.g. multisampled depth on lavapipe); the texture simply stays unbacked.
+        const VkResult createImageResult =
+            vmaCreateImage(m_allocator, &imageInfo, &allocationInfo, &resource.image, &resource.allocation, nullptr);
+        if (createImageResult != VK_SUCCESS) {
+            MGLOG_F("SyncTextureResource: vmaCreateImage failed (%d) textureId=%d extent=%ux%u depth=%u layers=%u "
+                    "mips=%u samples=%d format=%d",
+                    createImageResult, texture.GetExternalIndex(), imageInfo.extent.width, imageInfo.extent.height,
+                    imageInfo.extent.depth, imageInfo.arrayLayers, imageInfo.mipLevels,
+                    static_cast<Int>(imageInfo.samples), static_cast<Int>(imageInfo.format));
+            resource.image = VK_NULL_HANDLE;
+            resource.allocation = nullptr;
+            return false;
+        }
         ++m_textureImageEpoch; // a new attachment image invalidates cached render passes
 
         resource.layout = VK_IMAGE_LAYOUT_UNDEFINED;
@@ -1480,6 +1764,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         resource.viewType = shapeInfo.viewType;
         resource.sampleCount = resolvedSampleCount;
         resource.imageCreateFlags = imageCreateFlags;
+        resource.usageFlags = imageInfo.usage;
+        resource.storageUsageResolved = markedAsStorageImage;
         resource.syncedTextureParamsVersion = 0;
 
         if (preservedResource) {
@@ -1529,6 +1815,28 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_deferredViewReleases[frameIndex].clear();
     }
 
+    void VkTextureManager::ReclaimCompletedUploads(Bool waitAll) {
+        if (m_pendingUploadReclaims.empty()) {
+            return;
+        }
+
+        SizeT completed = 0;
+        for (; completed < m_pendingUploadReclaims.size(); ++completed) {
+            PendingUploadReclaim& entry = m_pendingUploadReclaims[completed];
+            if (waitAll) {
+                VK_VERIFY(vkWaitForFences(m_device, 1, &entry.fence, VK_TRUE, UINT64_MAX),
+                          "vkWaitForFences(texture upload reclaim)");
+            } else if (vkGetFenceStatus(m_device, entry.fence) != VK_SUCCESS) {
+                break;
+            }
+            vkDestroyFence(m_device, entry.fence, nullptr);
+            vkFreeCommandBuffers(m_device, m_commandPool, 1, &entry.commandBuffer);
+            vmaDestroyBuffer(m_allocator, entry.stagingBuffer, entry.stagingAllocation);
+        }
+        m_pendingUploadReclaims.erase(m_pendingUploadReclaims.begin(),
+                                      m_pendingUploadReclaims.begin() + static_cast<std::ptrdiff_t>(completed));
+    }
+
     void VkTextureManager::DestroyDeferredReleases() {
         for (auto& deferredReleases : m_deferredReleases) {
             deferredReleases.clear();
@@ -1730,17 +2038,119 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return true;
         }
 
-        // Combined depth-stencil images need per-aspect de-interleaved copies (VkBufferImageCopy
-        // aspectMask must have exactly one bit set). Until that is implemented, skip the upload
-        // instead of recording an invalid command buffer that kills the process.
+        // Combined depth-stencil images need per-aspect copies (VkBufferImageCopy aspectMask
+        // must have exactly one bit set), so de-interleave the shadow's GL wire format into
+        // a depth plane followed by a stencil plane per upload item.
         const VkImageAspectFlags uploadAspectMask = GetAspectMaskForFormat(outResource.format);
-        if ((uploadAspectMask & VK_IMAGE_ASPECT_DEPTH_BIT) && (uploadAspectMask & VK_IMAGE_ASPECT_STENCIL_BIT)) {
-            MGLOG_E("UploadDirtyMipLevels: skipping unimplemented depth-stencil data upload for textureId=%d",
-                    mipmapTexture.GetExternalIndex());
-            for (const auto& item : uploadItems) {
-                mipmapTexture.MarkStorageDirty(item.target, item.level, false);
+        const Bool isCombinedDepthStencil =
+            (uploadAspectMask & VK_IMAGE_ASPECT_DEPTH_BIT) && (uploadAspectMask & VK_IMAGE_ASPECT_STENCIL_BIT);
+        if (isCombinedDepthStencil) {
+            const Bool srcIsD24S8 = outResource.format == VK_FORMAT_D24_UNORM_S8_UINT;
+            const Bool srcIsD32FS8 = outResource.format == VK_FORMAT_D32_SFLOAT_S8_UINT;
+            if (!srcIsD24S8 && !srcIsD32FS8) {
+                MGLOG_E("UploadDirtyMipLevels: unsupported combined depth-stencil format %d for textureId=%d",
+                        static_cast<Int>(outResource.format), mipmapTexture.GetExternalIndex());
+                for (const auto& item : uploadItems) {
+                    mipmapTexture.MarkStorageDirty(item.target, item.level, false);
+                }
+                return true;
+            }
+            stagingSize = 0;
+            for (auto& item : uploadItems) {
+                const SizeT texelCount = static_cast<SizeT>(item.texelSize.x()) *
+                                         static_cast<SizeT>(item.texelSize.y()) *
+                                         static_cast<SizeT>(std::max(item.texelSize.z(), 1));
+                const SizeT shadowTexelSize = item.uploadByteSize / std::max<SizeT>(texelCount, 1);
+                MOBILEGL_ASSERT(shadowTexelSize == 4 || shadowTexelSize == 8,
+                                "UploadDirtyMipLevels: unexpected depth-stencil shadow texel size %zu for textureId=%d",
+                                shadowTexelSize, mipmapTexture.GetExternalIndex());
+                // Depth plane as the aspect's buffer-copy format (32-bit word for
+                // D24: low 24 bits; float for D32F), then one stencil byte per texel.
+                Vector<Uint8> deinterleaved(texelCount * 4 + texelCount);
+                Uint8* depthPlane = deinterleaved.data();
+                Uint8* stencilPlane = deinterleaved.data() + texelCount * 4;
+                const Uint8* shadow = static_cast<const Uint8*>(item.source);
+                for (SizeT t = 0; t < texelCount; ++t) {
+                    if (shadowTexelSize == 8) {
+                        // GL_FLOAT_32_UNSIGNED_INT_24_8_REV: float depth, then a word
+                        // with stencil in its low 8 bits.
+                        float depthValue;
+                        Uint32 stencilWord;
+                        std::memcpy(&depthValue, shadow + t * 8, sizeof(depthValue));
+                        std::memcpy(&stencilWord, shadow + t * 8 + 4, sizeof(stencilWord));
+                        if (srcIsD32FS8) {
+                            std::memcpy(depthPlane + t * 4, &depthValue, sizeof(depthValue));
+                        } else {
+                            const float clamped = std::min(std::max(depthValue, 0.0f), 1.0f);
+                            const Uint32 depthWord = static_cast<Uint32>(clamped * 16777215.0f + 0.5f);
+                            std::memcpy(depthPlane + t * 4, &depthWord, sizeof(depthWord));
+                        }
+                        stencilPlane[t] = static_cast<Uint8>(stencilWord & 0xFFu);
+                    } else {
+                        // GL_UNSIGNED_INT_24_8: depth in the high 24 bits, stencil low 8.
+                        Uint32 packed;
+                        std::memcpy(&packed, shadow + t * 4, sizeof(packed));
+                        if (srcIsD24S8) {
+                            const Uint32 depthWord = packed >> 8;
+                            std::memcpy(depthPlane + t * 4, &depthWord, sizeof(depthWord));
+                        } else {
+                            const float depthValue = static_cast<float>(packed >> 8) / 16777215.0f;
+                            std::memcpy(depthPlane + t * 4, &depthValue, sizeof(depthValue));
+                        }
+                        stencilPlane[t] = static_cast<Uint8>(packed & 0xFFu);
+                    }
+                }
+                item.expandedData = Move(deinterleaved);
+                item.source = item.expandedData.data();
+                item.uploadByteSize = item.expandedData.size();
+                item.offset = stagingSize;
+                stagingSize += static_cast<VkDeviceSize>(item.uploadByteSize);
+            }
+        }
+
+        // Pure-depth images whose canonical shadow layout differs from the image texel
+        // layout (the shadow keeps a full-scale 16/32-bit unorm word or a float; the
+        // image may be X8_D24 or a D32_SFLOAT fallback) convert per texel here.
+        if (uploadAspectMask == VK_IMAGE_ASPECT_DEPTH_BIT) {
+            const TextureInternalFormat depthInternal = mipmapTexture.GetFormat();
+            const Bool shadowIsFloat = depthInternal == TextureInternalFormat::DepthComponent32F;
+            const Bool dstIsFloat = outResource.format == VK_FORMAT_D32_SFLOAT;
+            const Bool dstIsD24Word = outResource.format == VK_FORMAT_X8_D24_UNORM_PACK32;
+            stagingSize = 0;
+            for (auto& item : uploadItems) {
+                const SizeT texelCount = static_cast<SizeT>(item.texelSize.x()) *
+                                         static_cast<SizeT>(item.texelSize.y()) *
+                                         static_cast<SizeT>(std::max(item.texelSize.z(), 1));
+                const SizeT shadowTexelSize = item.uploadByteSize / std::max<SizeT>(texelCount, 1);
+                const Bool needsConversion =
+                    (dstIsFloat && !shadowIsFloat) || (dstIsD24Word && shadowTexelSize == 4 && !shadowIsFloat);
+                if (needsConversion) {
+                    Vector<Uint8> converted(texelCount * 4);
+                    const Uint8* shadow = static_cast<const Uint8*>(item.source);
+                    for (SizeT t = 0; t < texelCount; ++t) {
+                        Uint32 wide = 0;
+                        if (shadowTexelSize == 2) {
+                            Uint16 raw = 0;
+                            std::memcpy(&raw, shadow + t * 2, sizeof(raw));
+                            wide = (static_cast<Uint32>(raw) << 16) | raw;
+                        } else {
+                            std::memcpy(&wide, shadow + t * 4, sizeof(wide));
+                        }
+                        if (dstIsFloat) {
+                            const float value = static_cast<float>(static_cast<double>(wide) / 4294967295.0);
+                            std::memcpy(converted.data() + t * 4, &value, sizeof(value));
+                        } else { // X8_D24: depth in the low 24 bits of a 32-bit word
+                            const Uint32 word = wide >> 8;
+                            std::memcpy(converted.data() + t * 4, &word, sizeof(word));
+                        }
+                    }
+                    item.expandedData = Move(converted);
+                    item.source = item.expandedData.data();
+                    item.uploadByteSize = item.expandedData.size();
+                }
+                item.offset = stagingSize;
+                stagingSize += static_cast<VkDeviceSize>(item.uploadByteSize);
             }
-            return true;
         }
 
         VkBuffer stagingBuffer = VK_NULL_HANDLE;
@@ -1793,7 +2203,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                         aspectMask, 0, outResource.mipLevels, outResource.arrayLayers);
         MOBILEGL_ASSERT(ok, "TransitionImageLayout to VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL failed");
 
+        // Array textures keep their GL "depth" in VkImage array layers, so the
+        // copy must address layerCount, not imageExtent.depth (which is invalid
+        // for 2D images and silently dropped every layer past the first).
+        const Bool depthSelectsArrayLayer = outResource.viewType == VK_IMAGE_VIEW_TYPE_1D_ARRAY ||
+                                            outResource.viewType == VK_IMAGE_VIEW_TYPE_2D_ARRAY ||
+                                            outResource.viewType == VK_IMAGE_VIEW_TYPE_CUBE_ARRAY;
         for (const auto& item : uploadItems) {
+            const Uint32 depthOrLayers = item.texelSize.z() > 0 ? static_cast<Uint32>(item.texelSize.z()) : 1u;
             VkBufferImageCopy copy{};
             copy.bufferOffset = item.offset;
             copy.bufferRowLength = 0;
@@ -1801,10 +2218,24 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             copy.imageSubresource.aspectMask = aspectMask;
             copy.imageSubresource.mipLevel = item.level;
             copy.imageSubresource.baseArrayLayer = item.baseArrayLayer;
-            copy.imageSubresource.layerCount = 1;
+            copy.imageSubresource.layerCount = depthSelectsArrayLayer ? depthOrLayers : 1;
             copy.imageOffset = {0, 0, 0};
             copy.imageExtent = {static_cast<Uint32>(item.texelSize.x()), static_cast<Uint32>(item.texelSize.y()),
-                                item.texelSize.z() > 0 ? static_cast<Uint32>(item.texelSize.z()) : 1u};
+                                depthSelectsArrayLayer ? 1u : depthOrLayers};
+            if (isCombinedDepthStencil) {
+                const SizeT texelCount = static_cast<SizeT>(item.texelSize.x()) *
+                                         static_cast<SizeT>(item.texelSize.y()) *
+                                         static_cast<SizeT>(std::max(item.texelSize.z(), 1));
+                VkBufferImageCopy depthCopy = copy;
+                depthCopy.imageSubresource.aspectMask = VK_IMAGE_ASPECT_DEPTH_BIT;
+                VkBufferImageCopy stencilCopy = copy;
+                stencilCopy.imageSubresource.aspectMask = VK_IMAGE_ASPECT_STENCIL_BIT;
+                stencilCopy.bufferOffset = item.offset + static_cast<VkDeviceSize>(texelCount) * 4;
+                const VkBufferImageCopy copies[2] = {depthCopy, stencilCopy};
+                vkCmdCopyBufferToImage(commandBuffer, stagingBuffer, outResource.image,
+                                       VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL, 2, copies);
+                continue;
+            }
             vkCmdCopyBufferToImage(commandBuffer, stagingBuffer, outResource.image, VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL,
                                    1, &copy);
         }
@@ -1835,11 +2266,23 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         VK_VERIFY(vkCreateFence(m_device, &fenceInfo, nullptr, &uploadFence), "vkCreateFence(texture upload)");
 
         VK_VERIFY(vkQueueSubmit(m_graphicsQueue, 1, &submitInfo, uploadFence), "vkQueueSubmit(texture)");
-        VK_VERIFY(vkWaitForFences(m_device, 1, &uploadFence, VK_TRUE, UINT64_MAX), "vkWaitForFences(texture upload)");
-        vkDestroyFence(m_device, uploadFence, nullptr);
-        vkFreeCommandBuffers(m_device, m_commandPool, 1, &commandBuffer);
-
-        vmaDestroyBuffer(m_allocator, stagingBuffer, stagingAllocation);
+        // Do NOT wait the fence here: this submit sits behind the previous
+        // frame's rendering on the queue, so a synchronous wait stalls the CPU
+        // until the GPU drains - a per-frame vkQueueWaitIdle for any workload
+        // with animated textures. Ordering against the current frame's draws is
+        // already guaranteed (its command buffer is submitted later, at
+        // present), so only the transient objects need to survive execution;
+        // park them until the fence signals.
+        m_pendingUploadReclaims.push_back({uploadFence, commandBuffer, stagingBuffer, stagingAllocation});
+        ReclaimCompletedUploads();
+        // Backstop for pathological upload storms: bound in-flight staging
+        // memory by blocking on the oldest upload only once the list is deep.
+        constexpr SizeT kMaxPendingTextureUploads = 16;
+        if (m_pendingUploadReclaims.size() > kMaxPendingTextureUploads) {
+            VK_VERIFY(vkWaitForFences(m_device, 1, &m_pendingUploadReclaims.front().fence, VK_TRUE, UINT64_MAX),
+                      "vkWaitForFences(texture upload backstop)");
+            ReclaimCompletedUploads();
+        }
 
         if (!ok) {
             MGLOG_D("%s: texture upload cmd failed", __func__);

MobileGL/MG_Backend/DirectVulkan/Renderer/VkTextureManager.h

@@ -28,6 +28,9 @@ public:
     // manager keys its per-draw fast path on this so an attachment's image recreation
     // invalidates the cached render pass (dirty-flag tracking; portable to Vulkan 1.1).
     Uint64 GetTextureImageEpoch() const { return m_textureImageEpoch; }
+    // Bumped whenever any tracked texture resource is erased; cached
+    // TextureResource pointers are valid only while this is unchanged.
+    Uint64 GetResourceEraseEpoch() const { return m_resourceEraseEpoch; }
 
     struct TextureIdentity {
         MG_State::GLState::ITextureObject* texture = nullptr;
@@ -53,6 +56,9 @@ public:
         VkCommandPool commandPool = VK_NULL_HANDLE;
         VkQueue graphicsQueue = VK_NULL_HANDLE;
         Uint32 frameCount = 0;
+        // VK_KHR_image_format_list is enabled: MUTABLE_FORMAT images can name the exact set of
+        // formats they will be viewed as, which is what lets a tiler keep them compressed.
+        Bool imageFormatListSupported = false;
     };
 
     struct TextureResource {
@@ -61,12 +67,16 @@ public:
             Uint32 baseArrayLayer = 0;
             Uint32 layerCount = 1;
             VkImageViewType viewType = VK_IMAGE_VIEW_TYPE_2D;
+            // May differ from the image format: sRGB images attach through their UNORM
+            // twin while GL_FRAMEBUFFER_SRGB is disabled.
+            VkFormat viewFormat = VK_FORMAT_UNDEFINED;
 
             Bool operator==(const AttachmentViewKey& other) const {
                 return mipLevel == other.mipLevel &&
                        baseArrayLayer == other.baseArrayLayer &&
                        layerCount == other.layerCount &&
-                       viewType == other.viewType;
+                       viewType == other.viewType &&
+                       viewFormat == other.viewFormat;
             }
         };
 
@@ -77,6 +87,8 @@ public:
                 hash ^= std::hash<Uint32>{}(key.layerCount) + 0x9e3779b9u + (hash << 6) + (hash >> 2);
                 hash ^= std::hash<Uint32>{}(static_cast<Uint32>(key.viewType)) +
                         0x9e3779b9u + (hash << 6) + (hash >> 2);
+                hash ^= std::hash<Uint32>{}(static_cast<Uint32>(key.viewFormat)) +
+                        0x9e3779b9u + (hash << 6) + (hash >> 2);
                 return hash;
             }
         };
@@ -157,7 +169,25 @@ public:
         VkImageViewType viewType = VK_IMAGE_VIEW_TYPE_2D;
         VkSampleCountFlagBits sampleCount = VK_SAMPLE_COUNT_1_BIT;
         VkImageCreateFlags imageCreateFlags = 0;
+        // Usage the live image was created with. STORAGE is only requested for textures that
+        // have actually been bound to a GL image unit, because on Adreno a storage-capable
+        // image loses UBWC bandwidth compression; a later image binding upgrades the usage
+        // and recreates the image, so the resolved usage has to be part of the compatibility
+        // check that decides whether the existing image can be kept.
+        VkImageUsageFlags usageFlags = 0;
+        // True once this image was (re)resolved while the texture was already marked as an
+        // image-unit texture. Distinguishes "not upgraded yet" from "cannot be upgraded"
+        // (a format whose optimalTilingFeatures lack STORAGE_IMAGE never gains the bit), so
+        // NeedsStorageImagePreparation cannot ask for a recreate that will never happen.
+        Bool storageUsageResolved = false;
         Uint16 syncedTextureParamsVersion = 0;
+        // Recording generation (VkTextureManager::GetRecordingGeneration) of the last
+        // command referencing this image that was recorded into the CURRENT frame
+        // command buffer. An image untouched by the open recording may have its
+        // out-of-pass work (deferred clears, sampled-layout transitions) recorded
+        // into the frame's PRE command buffer - which executes strictly before the
+        // frame's commands - instead of splitting the active render pass.
+        Uint64 lastRecordingGeneration = 0;
         // Snapshot of ITextureObject::GetContentVersion() at the last successful sync;
         // lets SyncTexture skip the whole re-check/re-upload when content is unchanged.
         Uint64 syncedContentVersion = 0;
@@ -190,7 +220,10 @@ public:
             std::swap(this->viewType, that.viewType);
             std::swap(this->sampleCount, that.sampleCount);
             std::swap(this->imageCreateFlags, that.imageCreateFlags);
+            std::swap(this->usageFlags, that.usageFlags);
+            std::swap(this->storageUsageResolved, that.storageUsageResolved);
             std::swap(this->syncedTextureParamsVersion, that.syncedTextureParamsVersion);
+            std::swap(this->lastRecordingGeneration, that.lastRecordingGeneration);
             std::swap(this->syncedContentVersion, that.syncedContentVersion);
             std::swap(this->syncedMipLevelCount, that.syncedMipLevelCount);
         }
@@ -251,6 +284,8 @@ public:
             viewType = VK_IMAGE_VIEW_TYPE_2D;
             sampleCount = VK_SAMPLE_COUNT_1_BIT;
             imageCreateFlags = 0;
+            usageFlags = 0;
+            storageUsageResolved = false;
             syncedTextureParamsVersion = 0;
             syncedContentVersion = 0;
             syncedMipLevelCount = 0;
@@ -267,6 +302,10 @@ public:
     Bool Initialize(const InitInfo& initInfo);
     void Shutdown();
     void BeginFrame(Uint32 frameIndex);
+    // Drains every frame slot's deferred image/view releases. Only valid when
+    // the caller has proven every queue submission complete; used by the
+    // present-less frame-boundary drain.
+    void CollectAllDeferredReleases();
 
     TextureResource* SyncTextureAndGetDescriptor(
         MG_State::GLState::ITextureObject& texture);
@@ -285,6 +324,36 @@ public:
                                                       VkImageLayout newLayout);
     Bool TransitionTextureForSampling(VkCommandBuffer commandBuffer, MG_State::GLState::ITextureObject& texture);
     Bool TransitionTextureForStorageImage(VkCommandBuffer commandBuffer, MG_State::GLState::ITextureObject& texture);
+
+    // Recording-generation bookkeeping for the pre-pass command stream. The
+    // generation advances every time the frame command buffer (re)begins
+    // recording; a resource whose stamp does not match was not referenced by
+    // any command in the open recording, so its out-of-pass work may safely
+    // execute ahead of the whole recording (in the pre command buffer).
+    void AdvanceRecordingGeneration() { ++m_recordingGeneration; }
+    void StampResourceRecordingUse(TextureResource& resource) const {
+        resource.lastRecordingGeneration = m_recordingGeneration;
+    }
+    // Map-lookup variant for callers that only hold the GL texture object.
+    void StampTextureRecordingUse(MG_State::GLState::ITextureObject* texture);
+    Bool WasTouchedThisRecording(const TextureResource& resource) const {
+        return resource.lastRecordingGeneration == m_recordingGeneration;
+    }
+    // Records that this texture is bound to a GL image unit, so its image must carry
+    // VK_IMAGE_USAGE_STORAGE_BIT. Must be called before NeedsStorageImagePreparation, and
+    // therefore before the render pass is committed: an image that has to be upgraded is
+    // recreated, which is illegal inside a render pass. Sticky for the texture's lifetime -
+    // GL lets an image binding come and go, and re-creating the image every time it does
+    // would cost far more than the compression it wins back.
+    void MarkStorageImageTexture(MG_State::GLState::ITextureObject& texture);
+    // True when this texture is marked but its live image predates the mark, i.e. the next sync
+    // will recreate it with STORAGE usage and copy the old contents forward. Callers use this to
+    // submit their pending recording first, so that copy cannot read pre-flush content.
+    Bool NeedsStorageUsageUpgrade(MG_State::GLState::ITextureObject& texture) const;
+    // The same ordering question for the other recreate-and-preserve trigger: true when this
+    // texture's live image carries a shorter mip chain than a full one, so defining the missing
+    // levels recreates it and copies the old contents forward.
+    Bool NeedsMipChainGrowth(MG_State::GLState::ITextureObject& texture) const;
     // Non-mutating probe for the per-draw storage-image fast path: true when preparing this
     // texture as a storage image may need work that is illegal inside a render pass (resource
     // creation, dirty-content upload, or a layout transition to GENERAL). Unknown state reports
@@ -331,6 +400,9 @@ public:
 private:
     // Bumped in SyncTextureResource right after vmaCreateImage(texture). See GetTextureImageEpoch().
     Uint64 m_textureImageEpoch = 1;
+    // See AdvanceRecordingGeneration. Starts above every resource's default
+    // stamp of 0 so a fresh resource counts as untouched.
+    Uint64 m_recordingGeneration = 1;
 
     Bool SyncTexture(MG_State::GLState::ITextureObject &texture,
                      TextureResource &outResource);
@@ -361,18 +433,28 @@ private:
     void DeferViewRelease(VkImageView view);
     void CollectDeferredReleases(Uint32 frameIndex);
     void DestroyDeferredReleases();
+    // Frees the fence/command buffer/staging buffer of every in-flight texture
+    // upload whose fence has signaled (submission order = completion order on
+    // the single queue, so the scan stops at the first still-pending entry).
+    // waitAll blocks on every entry - Shutdown's drain.
+    void ReclaimCompletedUploads(Bool waitAll = false);
     static TextureIdentity MakeTextureIdentity(MG_State::GLState::ITextureObject* texture);
     void EraseTrackedTexture(const TextureIdentity& identity);
     void PruneStaleTextureAliases(MG_State::GLState::ITextureObject* texture);
+    SizeT PruneDeadTextures();
 
     VkDevice m_device = VK_NULL_HANDLE;
     VkPhysicalDevice m_physicalDevice = VK_NULL_HANDLE;
     VmaAllocator m_allocator = nullptr;
     VkCommandPool m_commandPool = VK_NULL_HANDLE;
     VkQueue m_graphicsQueue = VK_NULL_HANDLE;
+    Bool m_imageFormatListSupported = false;
     Uint32 m_currentFrameIndex = 0;
 
     Uint8 m_gcCounter = 0;
+    // Frame-boundary GC gate: counts BeginFrame calls, not draws, so texture churn
+    // through non-draw paths (FBO clears, readbacks) still reaches the prune.
+    Uint32 m_gcFrameCounter = 0;
     // Active only between BeginDrawSyncScope/EndDrawSyncScope; identities of
     // textures already fully synced in the current draw (small N -> flat scan).
     Bool m_drawSyncScopeActive = false;
@@ -385,12 +467,45 @@ private:
         TextureResource* resource = nullptr;
     };
     Vector<DrawSyncedTexture> m_drawSyncedThisDraw;
+    // Cross-draw sampled-texture memo: the same few textures (atlas, lightmap)
+    // are resolved on every draw, so cache their resource pointers and skip the
+    // alive/resource map lookups. Node-based std::unordered_map keeps the
+    // pointees stable across inserts; erases bump m_resourceEraseEpoch, which
+    // every memo entry must match. SyncTexture still runs on memo hits, so
+    // content/param freshness is unaffected. A dead-then-reused texture address
+    // cannot false-hit: the new object carries a new lifetime id.
+    struct SyncedTextureMemoEntry {
+        const MG_State::GLState::ITextureObject* texture = nullptr;
+        Uint64 lifetimeId = 0;
+        Uint64 eraseEpoch = 0;
+        TextureResource* resource = nullptr;
+    };
+    static constexpr Uint32 kSyncedTextureMemoSize = 8;
+    SyncedTextureMemoEntry m_syncedTextureMemo[kSyncedTextureMemoSize];
+    Uint32 m_syncedTextureMemoNext = 0;
+    Uint64 m_resourceEraseEpoch = 1;
     // Formats whose mutable-image probe failed on this device; their images are created
     // without MUTABLE_FORMAT_BIT so repeat syncs neither re-probe nor flag-mismatch.
     std::unordered_set<VkFormat> m_mutableFormatUnsupported;
     std::unordered_map<TextureIdentity, WeakPtr<MG_State::GLState::ITextureObject>, TextureIdentityHash> m_aliveObjects;
     std::unordered_map<TextureIdentity, TextureResource, TextureIdentityHash> m_textureResources;
+    // Textures that have been bound to a GL image unit (see MarkStorageImageTexture).
+    std::unordered_set<TextureIdentity, TextureIdentityHash> m_storageImageTextures;
+    // Supported multisample counts per format, so repeat texture syncs do not
+    // re-query vkGetPhysicalDeviceImageFormatProperties.
+    std::unordered_map<VkFormat, VkSampleCountFlags> m_multisampleCountsByFormat;
     Vector<Vector<TextureResource>> m_deferredReleases;
     Vector<Vector<VkImageView>> m_deferredViewReleases;
+    // Texture uploads are submitted out-of-band but NOT waited on (waiting
+    // behind the queue serialized the CPU against the previous frame's GPU
+    // work every time an animated atlas re-uploaded). Their transient objects
+    // are parked here and reclaimed once the upload fence signals.
+    struct PendingUploadReclaim {
+        VkFence fence = VK_NULL_HANDLE;
+        VkCommandBuffer commandBuffer = VK_NULL_HANDLE;
+        VkBuffer stagingBuffer = VK_NULL_HANDLE;
+        VmaAllocation stagingAllocation = nullptr;
+    };
+    Vector<PendingUploadReclaim> m_pendingUploadReclaims;
 };
 } // namespace MobileGL::MG_Backend::DirectVulkan

MobileGL/MG_Backend/DirectVulkan/Renderer/VulkanRenderer.h

@@ -76,6 +76,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         GLenum indexType = GL_UNSIGNED_SHORT;
         SizeT indexByteOffset = 0;
         SizeT indexByteSize = 0;
+        // Interpret indexByteOffset as a raw client pointer even when an element
+        // array buffer is bound (backend-synthesized index lists, e.g. the
+        // GL_LINE_LOOP -> LINE_STRIP rewrite).
+        Bool forceClientMemory = false;
     };
 
     struct DrawIndexedCmd {
@@ -114,7 +118,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
     };
 
-    class VulkanRenderer : public IBufferCopyCommandProvider, public FrameContext::IRecordingObserver {
+    class VulkanRenderer : public IBufferCopyCommandProvider,
+                           public FrameContext::IRecordingObserver,
+                           public VkRenderPassManager::IEvictionObserver,
+                           public ProgramFactory::IEvictionObserver {
     public:
         VulkanRenderer(NativeWindowType window, const VulkanRendererConfig& cfg = {});
         ~VulkanRenderer();
@@ -131,9 +138,31 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // recording, before any render pass.
         void OnFrameCommandRecordingBegan(VkCommandBuffer commandBuffer) override;
 
+        // VkRenderPassManager::IEvictionObserver: the render-pass aging sweep just
+        // destroyed these VkRenderPasses; evict every graphics pipeline hashed on a
+        // dying handle (they share its >1024-boundary idleness, so immediate
+        // destruction is safe) and drop the last-pipeline memo if any went.
+        void OnRenderPassesDestroyed(const Vector<VkRenderPass>& renderPasses) override;
+
+        // ProgramFactory::IEvictionObserver: an aged-out program entry was
+        // destroyed; evict its compute pipeline and graphics pipelines (same
+        // idleness guarantee - they are only bound through draws/dispatches that
+        // stamp the program entry) and purge the descriptor-set cache entries
+        // keyed by its now-recyclable VkDescriptorSetLayout handle.
+        void OnProgramEvicted(ProgramFactory::HashType programHash,
+                              VkDescriptorSetLayout descriptorSetLayout) override;
+
         Bool SetupDraw(FrameContext::FrameData& frame, GLenum mode, Flags<DrawSetupAspect> aspects,
                        const DrawCmdParam& drawParams,
                        const IndexBufferView* pIndexBufferView = nullptr);
+        // ANGLE-style consecutive-draw fast path: SetupDraw snapshots the fully
+        // resolved draw configuration; the next draw whose cheap version/identity
+        // checks all match skips the resolution half (LOD probe, sampled-set
+        // walk, render-pass and pipeline resolution) and jumps straight to the
+        // per-draw tail. Returns false (leaving no side effects that the full
+        // path cannot redo idempotently) whenever anything might have changed.
+        Bool TrySetupDrawFastPath(FrameContext::FrameData& frame, GLenum mode, Flags<DrawSetupAspect> aspects,
+                                  const DrawCmdParam& drawParams, const IndexBufferView* pIndexBufferView);
         void ClearAttachmentsOnActiveRenderPass(VkCommandBuffer commandBuffer,
                                                 const RenderPassEntry& compatibleRenderPassEntry);
 
@@ -171,6 +200,25 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                               GLsizei srcWidth, GLsizei srcHeight, GLsizei srcDepth);
         void GenerateMipmap(GLenum target);
         void ReadPixels(GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, void* pixels);
+        // GL_DEPTH_COMPONENT / GL_DEPTH_STENCIL / GL_STENCIL_INDEX readback from the
+        // read framebuffer's depth/stencil attachment (per-aspect buffer copies with
+        // CPU repacking into the requested client layout).
+        void ReadDepthStencilPixels(MG_State::GLState::FramebufferObject& readFbo, GLint x, GLint y, GLsizei width,
+                                    GLsizei height, GLenum format, GLenum type, void* pixels);
+        // Copy-and-repack core shared by depth-stencil ReadPixels and GetTexImage;
+        // expects command recording to be active and any render pass already ended.
+        void ReadDepthStencilImageToClient(VkImage image, VkFormat vkFormat, VkImageLayout* trackedLayout,
+                                           VkImageAspectFlags imageAspect, Uint32 mipLevel, Uint32 baseArrayLayer,
+                                           GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type,
+                                           void* pixels);
+        // Same-extent depth blit between images of different depth formats: host
+        // round-trip with a per-texel re-encode (see BlitNamedFramebuffer).
+        Bool BlitDepthAcrossFormats(FrameContext::FrameData& frame, VkImage srcImage, VkFormat srcFormat,
+                                    VkImageLayout* srcTrackedLayout, Uint32 srcMipLevel, Uint32 srcBaseArrayLayer,
+                                    VkImage dstImage, VkFormat dstFormat, VkImageLayout* dstTrackedLayout,
+                                    Uint32 dstMipLevel, Uint32 dstBaseArrayLayer, GLint srcX, GLint srcY, GLint dstX,
+                                    GLint dstY, GLint width, GLint height, VkImageLayout srcRestoreLayout,
+                                    VkImageLayout dstRestoreLayout, Bool stencilAspect);
         static SizeT GetReadbackTexelSize(VkFormat sourceFormat);
         static Bool ConvertReadbackPixels(const Uint8* sourcePixels, VkFormat sourceFormat,
                                           GLsizei width, GLsizei height, GLenum destinationFormat,
@@ -256,7 +304,20 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                       const VkTimerQueryManager::TimestampRecord& end) const;
         Uint64 GetTimerQueryTimestampNs(const VkTimerQueryManager::TimestampRecord& record) const;
 
+        // GL_SAMPLES_PASSED occlusion queries: every app draw between Start and Stop is
+        // wrapped in a Vulkan occlusion query slot; the result is the slot sum. Requires
+        // hostQueryReset for slot recycling - Start fails (frontend keeps the query
+        // unsupported) when the device lacks it.
+        Bool StartOcclusionQueryCapture();
+        void StopOcclusionQueryCapture(Vector<Uint32>& outSlots);
+        // Flushes pending commands, waits, sums the slots, and recycles them.
+        Bool ResolveOcclusionQueryResult(const Vector<Uint32>& slots, Uint64& outSamples);
+
         void RequestSwapchainResize(Uint32 width, Uint32 height);
+        // Re-query the surface and report whether the live swapchain no longer matches it
+        // (size or orientation). This - not a VK_SUBOPTIMAL_KHR result - is what decides a
+        // rebuild, so a surface the driver merely considers suboptimal cannot thrash.
+        Bool SwapchainIsOutOfDate();
         // Returns false when the surface is zero-area (minimized/hidden window):
         // no new swapchain is installed and presentation must stay suspended.
         Bool RecreateSwapchain();
@@ -345,16 +406,40 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         VkFence AcquirePooledSubmitFence();
         void DestroySubmitFencePool();
         Bool HasPendingRecordedWork() const;
+        // Frame-boundary housekeeping for paths that never reach Present's
+        // tail (present-less readback loops, suspended presentation, blocking
+        // sync waits): runs the same per-frame drains Present performs, but
+        // only when every queue submission has been observed complete AND no
+        // recorded-but-unsubmitted commands exist - i.e. when CPU-GPU overlap
+        // is provably already zero. Never blocks (non-blocking fence poll
+        // only), so the presenting path's frames-in-flight pipelining is
+        // untouched. Returns true when the drain ran.
+        Bool TryDrainFrameTransients();
 
         Vector<SubmitRecord> m_inFlightSubmits;
         Vector<VkFence> m_freeSubmitFences;
         Uint64 m_submitCounter = 0;
         Uint64 m_completedSubmitCounter = 0;
+        // Drains since the last Present, gating the drain's frame-boundary-equivalent
+        // work (arena rewind + cache aging): a presenting app's mid-frame
+        // readbacks/waits must neither churn the transient caches nor accelerate the
+        // aging clocks, while present-less loops still cross a boundary every few
+        // iterations. Reset in Present.
+        Uint32 m_drainsSinceLastPresent = 0;
 
         NativeWindowType m_window = 0;
         void* m_platformDisplay = nullptr;
         void* m_platformLibrary = nullptr;
         void* m_platformCloseDisplay = nullptr;
+        // Some real ICDs (e.g. NVIDIA's proprietary Linux driver) don't implement
+        // VK_EXT_headless_surface at all. Detected once in CreateInstance() from the
+        // enumerated instance extensions; when false, CreateSurface() falls back to a
+        // hidden Xlib window instead of vkCreateHeadlessSurfaceEXT.
+        Bool m_headlessSurfaceSupported = true;
+        // Set when CreateSurface() had to create its own Xlib window for the fallback
+        // above (rather than being handed one by the caller), so Shutdown() knows it
+        // owns that window and must destroy it.
+        Bool m_ownsFallbackXlibWindow = false;
         VulkanRendererConfig m_config;
         Bool m_swapchainResizeRequested = false;
         // Presentation is suspended while the window is zero-area (minimized): the
@@ -367,6 +452,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         Vector<VkExtensionProperties> m_extensions;
         VkInstance m_instance = VK_NULL_HANDLE;
         VkDebugUtilsMessengerEXT m_debugMessenger = VK_NULL_HANDLE;
+        // Fallback reporting channel for drivers that ship the validation layers but
+        // only expose the older VK_EXT_debug_report (Adreno 650 / Vulkan 1.1.128).
+        VkDebugReportCallbackEXT m_debugReportCallback = VK_NULL_HANDLE;
         PhysicalDevice m_physicalDevice;
         VkDevice m_device = VK_NULL_HANDLE;
         VmaAllocator m_allocator = nullptr;
@@ -404,6 +492,74 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                                                  Uint32 stride);
         static inline PFNDrawIndexedIndirectCountFunc s_vkCmdDrawIndexedIndirectCount = nullptr;
 
+        // VK_EXT_transform_feedback (GL transform feedback capture)
+        Bool m_transformFeedbackFeatureEnabled = false;
+        // VK_EXT_vertex_attribute_divisor: without it every non-zero glVertexAttribDivisor
+        // behaves as 1, because that is all Vulkan's instance input rate can express.
+        Bool m_vertexAttributeDivisorEnabled = false;
+        static inline PFN_vkCmdBindTransformFeedbackBuffersEXT s_vkCmdBindTransformFeedbackBuffersEXT = nullptr;
+        static inline PFN_vkCmdBeginTransformFeedbackEXT s_vkCmdBeginTransformFeedbackEXT = nullptr;
+        static inline PFN_vkCmdEndTransformFeedbackEXT s_vkCmdEndTransformFeedbackEXT = nullptr;
+        // Counter buffers (one 4-byte slot per capture binding) let consecutive
+        // draws within one glBeginTransformFeedback append GL-style. Transform feedback
+        // objects can each hold an open, paused span at the same time, so the counters are
+        // per object: one group of four slots each, handed out on first use.
+        static constexpr SizeT kXfbCounterObjectSlots = 16;
+        VkBufferObject m_xfbCounterBuffer;
+        UnorderedMap<Uint, Uint32> m_xfbCounterSlotByObject;
+        Uint32 m_xfbNextCounterSlot = 0;
+        // Set for a slot once a captured draw has been recorded into its span; selects
+        // counter-buffer resume on the next captured draw of the same span.
+        Array<Bool, kXfbCounterObjectSlots> m_xfbCountersValid{};
+        Array<Uint64, kXfbCounterObjectSlots> m_xfbLastSeenGeneration{};
+        // Counter slot group of the bound transform feedback object.
+        Uint32 CurrentXfbCounterSlot();
+        // Wraps a recorded draw with BeginTransformFeedbackEXT/EndTransformFeedbackEXT
+        // when GL transform feedback is active; binds capture buffers on demand.
+        Bool BeginXfbCaptureForDraw(FrameContext::FrameData& frame);
+        void EndXfbCaptureForDraw(FrameContext::FrameData& frame, Bool began);
+        // Makes the captured bytes visible to whatever reads them next. Deferred rather than
+        // recorded next to the capture, because the capturing draw runs inside a render pass
+        // that declares no self-dependency.
+        void MakeXfbWritesVisible();
+        Bool m_xfbWritesPendingVisibility = false;
+        // Wrap one app draw in an occlusion-query slot while a GL_SAMPLES_PASSED
+        // query is active. Returns whether a slot was begun (End must mirror it).
+        Bool BeginOcclusionForDraw(VkCommandBuffer commandBuffer);
+        void EndOcclusionForDraw(VkCommandBuffer commandBuffer, Bool began);
+        Bool m_occlusionQueryPreciseEnabled = false;
+        Bool m_hostQueryResetEnabled = false;
+        PFN_vkResetQueryPool s_vkResetQueryPool = nullptr;
+        VkQueryPool m_occlusionQueryPool = VK_NULL_HANDLE;
+        static constexpr Uint32 kOcclusionQuerySlots = 8192;
+        Uint32 m_occlusionSlotCursor = 0;
+        Bool m_occlusionCaptureActive = false;
+        Vector<Uint32> m_occlusionActiveSlots;
+        // Transform feedback primitive queries: one pool slot per captured draw yields
+        // the (written, needed) pair; GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN sums the
+        // first, GL_PRIMITIVES_GENERATED the second - exact with geometry shaders,
+        // unlike the CPU fallback accounting.
+        Bool m_xfbQueriesSupported = false;
+        PFN_vkCmdBeginQueryIndexedEXT s_vkCmdBeginQueryIndexedEXT = nullptr;
+        PFN_vkCmdEndQueryIndexedEXT s_vkCmdEndQueryIndexedEXT = nullptr;
+        VkQueryPool m_xfbQueryPool = VK_NULL_HANDLE;
+        static constexpr Uint32 kXfbQuerySlots = 8192;
+        Uint32 m_xfbQuerySlotCursor = 0;
+        Bool m_xfbQueryCaptureActive[2] = {false, false}; // [0]=written, [1]=generated
+        Vector<Uint32> m_xfbQueryActiveSlots[2];
+        Bool m_xfbQuerySlotOpen = false;
+        Uint32 m_xfbQueryOpenSlot = 0;
+
+    public:
+        // kind: 0 = PRIMITIVES_WRITTEN, 1 = PRIMITIVES_GENERATED.
+        Bool StartXfbQueryCapture(Uint32 kind);
+        void StopXfbQueryCapture(Uint32 kind, Vector<Uint32>& outSlots);
+        Bool ResolveXfbQueryResult(const Vector<Uint32>& slots, Bool wantGenerated, Uint64& outPrimitives);
+
+    private:
+        void BeginXfbQueryForDraw(VkCommandBuffer commandBuffer);
+        void EndXfbQueryForDraw(VkCommandBuffer commandBuffer);
+
         VkCommandPool m_commandPool = VK_NULL_HANDLE;
 
         VkBufferManager m_bufferManager;
@@ -416,14 +572,30 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // gather + synthetic vertex-input rebuild + payload hash + lookup) when the full pipeline
         // state is unchanged from the previous draw. The key provably covers every pipeline field.
         // Reset per-frame and on pipeline destruction so the cached handle can never dangle.
-        Bool m_lastPipelineValid = false;
-        GLenum m_lastPipelineMode = 0;
-        Uint64 m_lastPipelineProgramHash = 0;
-        Uint64 m_lastPipelineVertexInputHash = 0;
-        Uint64 m_lastPipelineRenderPassHash = 0;
-        Uint m_lastPipelineRenderStateVersion = 0;
-        ProgramFactory::CompileOptionFlags m_lastPipelineTransformFlags = {};
-        VkPipeline m_lastPipelineResult = VK_NULL_HANDLE;
+        // Small N-way pipeline-resolution memo (round-robin replacement). A
+        // single-entry memo thrashed on draw sequences that alternate a few
+        // pipelines (GUI text/quad program ping-pong), paying the full
+        // payload-hash lookup per draw; eight entries cover such working sets
+        // while keeping the hit path a trivial linear scan.
+        struct PipelineMemoEntry {
+            GLenum mode = 0;
+            Uint64 programHash = 0;
+            Uint64 vertexInputHash = 0;
+            Uint64 renderPassHash = 0;
+            Uint renderStateVersion = 0;
+            ProgramFactory::CompileOptionFlags transformFlags = {};
+            VkPipeline pipeline = VK_NULL_HANDLE;
+        };
+        static constexpr Uint32 kPipelineMemoSize = 8;
+        PipelineMemoEntry m_pipelineMemo[kPipelineMemoSize];
+        Uint32 m_pipelineMemoCount = 0;
+        Uint32 m_pipelineMemoNext = 0;
+        // Drops every memoized pipeline handle. Required at command-buffer
+        // boundaries and whenever any pipeline may have been destroyed.
+        void InvalidatePipelineMemo() {
+            m_pipelineMemoCount = 0;
+            m_pipelineMemoNext = 0;
+        }
         UnorderedMap<ProgramFactory::HashType, VkPipeline> m_computePipelines;
         UniquePtr<ProgramFactory> m_programFactory;
         UniquePtr<UniformManager> m_uniformManager;
@@ -452,9 +624,61 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         ProgramFactory::CompileOptionFlags m_lastSampledSetTransformFlags = {};
         Uint64 m_lastSampledSetBindGeneration = 0;
 
+        // Memo for the per-draw explicit-LOD-0 eligibility probe
+        // (ProgramSamplesOnlySingleLevelTextures): same key family as the
+        // sampled-set memo, plus the sampled textures' params-version sum so a
+        // level-range or filter change re-probes. On a hit the resolved
+        // transform flags are reused, which also collapses the two
+        // GetOrCreateProgram lookups into one.
+        Bool m_lastLodDecisionValid = false;
+        Uint64 m_lastLodProgramLifetimeId = 0;
+        Uint32 m_lastLodProgramVersion = 0;
+        Uint64 m_lastLodBindGeneration = 0;
+        Uint64 m_lastLodParamsSum = 0;
+        ProgramFactory::CompileOptionFlags m_lastLodBaseFlags = {};
+        ProgramFactory::CompileOptionFlags m_lastLodResultFlags = {};
+
+        // Snapshot behind TrySetupDrawFastPath. Values only: the program and
+        // render-pass caches are open-addressing maps whose entries move on
+        // insert, so no pointers into them are cached; the pipeline handle is
+        // protected by the command-buffer-boundary reset plus the mid-frame
+        // pipeline-destruction resets, and monotonic epochs guard everything
+        // that can be destroyed or recreated between draws.
+        struct SetupDrawSnapshot {
+            Bool valid = false;
+            Uint8 aspects = 0;
+            GLenum mode = 0;
+            Uint64 programLifetimeId = 0;
+            Uint32 programVersion = 0;
+            const void* vao = nullptr;
+            Uint32 vaoConfigVersion = 0;
+            const void* drawFbo = nullptr;
+            Uint16 fboVersion = 0;
+            Bool drawFboIsDefault = false;
+            Uint renderStateVersion = 0;
+            Uint64 bindGeneration = 0;
+            Uint32 baseTransformFlags = 0;
+            Uint32 resolvedTransformFlags = 0;
+            Uint64 renderPassHash = 0;
+            Uint32 imageIndex = 0;
+            Uint64 textureEraseEpoch = 0;
+            Uint64 textureImageEpoch = 0;
+            Uint64 renderbufferImageEpoch = 0;
+            Uint64 sampledContentSum = 0;
+            Uint64 sampledParamsSum = 0;
+            IntVec2 renderPassExtent = {0, 0};
+            VkPipeline pipeline = VK_NULL_HANDLE;
+        };
+        SetupDrawSnapshot m_setupDrawSnapshot;
+
         // Per-draw scratch buffers (clear keeps capacity) — these paths run for every
         // draw call and must not allocate.
         Vector<MG_State::GLState::ITextureObject*> m_sampledTexturesScratch;
+        // Parallel to m_sampledTexturesScratch, refilled by every SetupDraw's
+        // first sampled-texture loop: the resolved backend resources, so the
+        // post-transition loop can skip re-resolving textures whose layout is
+        // already sampleable.
+        Vector<VkTextureManager::TextureResource*> m_sampledResourcesScratch;
         Vector<MG_State::GLState::ITextureObject*> m_storageImageTexturesScratch;
         Vector<VkBuffer> m_vertexBuffersScratch;
         Vector<VkDeviceSize> m_vertexOffsetsScratch;
@@ -517,6 +741,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         void CreateInstance();
         VkResult SetupDebugMessenger();
         VkResult DestroyDebugMessenger();
+        VkResult SetupDebugReportCallback();
+        void DestroyDebugReportCallback();
         VkDebugUtilsMessengerCreateInfoEXT PopulateDebugMessengerCreateInfo();
         void CreateSurface();
         void PickPhysicalDevice();
@@ -535,8 +761,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             const RenderPassEntry& renderPassEntry);
         VkPipeline GetOrCreateComputePipeline(const ProgramFactory::VkProgramObject& programObj);
         void DestroyComputePipelines();
+        // Takes the frame rather than a command buffer: a first-time storage-usage upgrade has to
+        // flush the pending recording (see the body), which retires the current command buffer.
         Bool PrepareStorageImageTextures(
-            VkCommandBuffer commandBuffer,
+            FrameContext::FrameData& frame,
             const MG_State::GLState::ProgramObject& program,
             const ProgramFactory::VkProgramObject& programObj);
 
@@ -575,6 +803,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                            VkImageLayout finalLayout);
         Bool SubmitReadbackCommandsAndWait(FrameContext::FrameData& frame);
 
+    public:
+        // Submits whatever is recorded and waits for it. The CPU is about to read memory
+        // a shader wrote (a mapped shader storage buffer), and coherent host-visible
+        // storage only guarantees visibility once the work that produced it has retired.
+        Bool FinishPendingGpuWork();
+
+    private:
+
         void ShutdownSwapchain();
 
         // Static functions
@@ -598,6 +834,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                                  const PhysicalDevice& compareWithDevice,
                                                  PhysicalDevice& outBetterDevice);
         static constexpr const char* s_validationLayerNames[] = {"VK_LAYER_KHRONOS_validation"};
+        // VK_KHR_image_format_list: lets MUTABLE_FORMAT images declare their exact view-format
+        // set so the driver can keep bandwidth compression (see CreateLogicalDeviceAndQueues).
+        Bool m_imageFormatListExtensionEnabled = false;
+
         static constexpr const char* s_deviceExtensionNames[] = {VK_KHR_SWAPCHAIN_EXTENSION_NAME};
         static Bool CheckValidationLayerSupport();
 

MobileGL/MG_Backend/DirectVulkan/VkIncludes.h

@@ -52,19 +52,48 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     }
 } // namespace MobileGL::MG_Backend::DirectVulkan
 
+namespace MobileGL::MG_Backend::DirectVulkan {
+    // GL renders into sRGB color attachments RAW while GL_FRAMEBUFFER_SRGB is disabled
+    // (the core-profile default); Vulkan sRGB attachments always encode on write. The
+    // attachment view (and render pass format) therefore drops to the UNORM twin
+    // whenever the capability is off. Sampled views keep the sRGB format (decode on
+    // sample is unconditional in GL).
+    inline VkFormat ResolveSrgbAttachmentWriteFormat(VkFormat format, bool framebufferSrgbEnabled) {
+        if (framebufferSrgbEnabled) return format;
+        switch (format) {
+        case VK_FORMAT_R8G8B8A8_SRGB:
+            return VK_FORMAT_R8G8B8A8_UNORM;
+        case VK_FORMAT_B8G8R8A8_SRGB:
+            return VK_FORMAT_B8G8R8A8_UNORM;
+        default:
+            return format;
+        }
+    }
+} // namespace MobileGL::MG_Backend::DirectVulkan
+
+// The context line (__VA_ARGS__ = its own format string + args) must be a SEPARATE log
+// call: appending its format to the base format while its arguments precede the base
+// arguments makes every conversion read the wrong slot (a %s pulling an int crashes).
 #define VK_VERIFY(expr, ...)                                                                                           \
     do {                                                                                                               \
         VkResult _vk_verify_result = (expr);                                                                           \
         if (_vk_verify_result != VK_SUCCESS) {                                                                         \
-            MGLOG_F("Vulkan error %s (%d) at %s:%d" __VA_OPT__(" - ") __VA_ARGS__,                                  \
+            __VA_OPT__(MGLOG_F(__VA_ARGS__);)                                                                          \
+            MGLOG_F("Vulkan error %s (%d) at %s:%d",                                                                   \
                     MobileGL::MG_Backend::DirectVulkan::VkResultToString(_vk_verify_result),                          \
                     _vk_verify_result, __FILE__, __LINE__);                                                            \
         }                                                                                                              \
-        MOBILEGL_ASSERT(_vk_verify_result == VK_SUCCESS, "Vulkan error %s (%d) at %s:%d" __VA_OPT__(" - ") __VA_ARGS__, MobileGL::MG_Backend::DirectVulkan::VkResultToString(_vk_verify_result), _vk_verify_result, __FILE__, __LINE__);  \
+        MOBILEGL_ASSERT(_vk_verify_result == VK_SUCCESS, "Vulkan error %s (%d) at %s:%d",                              \
+                        MobileGL::MG_Backend::DirectVulkan::VkResultToString(_vk_verify_result),                       \
+                        _vk_verify_result, __FILE__, __LINE__);                                                        \
     } while (0)
 
 #define XXHASH_VERIFY(expr, ...)                                                                                       \
     do {                                                                                                               \
         XXH_errorcode _xxh_verify_result = (expr);                                                                     \
-        MOBILEGL_ASSERT(_xxh_verify_result == XXH_OK, "XXHash error %d at %s:%d" __VA_OPT__(" - ") __VA_ARGS__, _xxh_verify_result, __FILE__, __LINE__);  \
+        if (_xxh_verify_result != XXH_OK) {                                                                            \
+            __VA_OPT__(MGLOG_F(__VA_ARGS__);)                                                                          \
+        }                                                                                                              \
+        MOBILEGL_ASSERT(_xxh_verify_result == XXH_OK, "XXHash error %d at %s:%d", _xxh_verify_result, __FILE__,        \
+                        __LINE__);                                                                                     \
     } while (0)

MobileGL/MG_Impl/CGLImpl/CGLImpl.cpp

@@ -24,6 +24,7 @@ namespace MobileGL::MG_Impl::CGLImpl {
             GLint Samples = 0;
             GLint Profile = kCGLOGLPVersion_3_2_Core;
             GLint RendererId = 0x4d474c;
+            GLint DisplayMask = 0;
         };
 
         struct ContextObject {
@@ -134,6 +135,9 @@ namespace MobileGL::MG_Impl::CGLImpl {
             case kCGLPFARendererID:
                 pixelFormat.RendererId = value;
                 break;
+            case kCGLPFADisplayMask:
+                pixelFormat.DisplayMask = value;
+                break;
             default:
                 break;
             }
@@ -343,6 +347,9 @@ namespace MobileGL::MG_Impl::CGLImpl {
         case kCGLPFARendererID:
             *value = pixelFormat->RendererId;
             return kCGLNoError;
+        case kCGLPFADisplayMask:
+            *value = pixelFormat->DisplayMask;
+            return kCGLNoError;
         case kCGLPFAOpenGLProfile:
             *value = pixelFormat->Profile;
             return kCGLNoError;
@@ -481,6 +488,32 @@ namespace MobileGL::MG_Impl::CGLImpl {
         return it == currentContexts.end() ? nullptr : it->second;
     }
 
+    CGLError SetVirtualScreen(CGLContextObj ctx, GLint screen) {
+        const std::lock_guard<std::recursive_mutex> lock(RegistryMutex());
+        auto* object = TryGetContext(ctx);
+        if (!object) {
+            return kCGLBadContext;
+        }
+        if (screen != 0) {
+            return kCGLBadValue;
+        }
+        object->VirtualScreen = screen;
+        return kCGLNoError;
+    }
+
+    CGLError GetVirtualScreen(CGLContextObj ctx, GLint* screen) {
+        const std::lock_guard<std::recursive_mutex> lock(RegistryMutex());
+        auto* object = TryGetContext(ctx);
+        if (!object) {
+            return kCGLBadContext;
+        }
+        if (!screen) {
+            return kCGLBadAddress;
+        }
+        *screen = object->VirtualScreen;
+        return kCGLNoError;
+    }
+
     CGLError SetParameter(CGLContextObj ctx, CGLContextParameter pname, const GLint* params) {
         const std::lock_guard<std::recursive_mutex> lock(RegistryMutex());
         auto* object = TryGetContext(ctx);

MobileGL/MG_Impl/CGLImpl/CGLImpl.h

@@ -32,6 +32,8 @@ namespace MobileGL::MG_Impl::CGLImpl {
 
     CGLError SetCurrentContext(CGLContextObj ctx);
     CGLContextObj GetCurrentContext();
+    CGLError SetVirtualScreen(CGLContextObj ctx, GLint screen);
+    CGLError GetVirtualScreen(CGLContextObj ctx, GLint* screen);
     CGLError SetParameter(CGLContextObj ctx, CGLContextParameter pname, const GLint* params);
     CGLError GetParameter(CGLContextObj ctx, CGLContextParameter pname, GLint* params);
     CGLError UpdateContext(CGLContextObj ctx);

MobileGL/MG_Impl/CGLImpl/Exporting/Definitions.cpp

@@ -71,6 +71,14 @@ MOBILEGL_CGL_API CGLContextObj CGLGetCurrentContext(void) {
     return MobileGL::MG_Impl::CGLImpl::GetCurrentContext();
 }
 
+MOBILEGL_CGL_API CGLError CGLSetVirtualScreen(CGLContextObj ctx, GLint screen) {
+    return MobileGL::MG_Impl::CGLImpl::SetVirtualScreen(ctx, screen);
+}
+
+MOBILEGL_CGL_API CGLError CGLGetVirtualScreen(CGLContextObj ctx, GLint* screen) {
+    return MobileGL::MG_Impl::CGLImpl::GetVirtualScreen(ctx, screen);
+}
+
 MOBILEGL_CGL_API CGLError CGLSetParameter(CGLContextObj ctx, CGLContextParameter pname, const GLint* params) {
     return MobileGL::MG_Impl::CGLImpl::SetParameter(ctx, pname, params);
 }

MobileGL/MG_Impl/DyldInterpose/DyldInterpose.cpp

@@ -10,8 +10,12 @@
 
 #if defined(__APPLE__)
 
+#include "MG_Impl/CGLImpl/CGLImpl.h"
 #include "MG_Impl/GetProcAddress.h"
 
+#include <CoreGraphics/CoreGraphics.h>
+#include <CoreVideo/CVDisplayLink.h>
+#include <cstdint>
 #include <dlfcn.h>
 
 namespace {
@@ -47,10 +51,52 @@ namespace {
         return dlsym(handle, symbol);
     }
 
+    CGDirectDisplayID DisplayForMask(GLint displayMask) {
+        constexpr std::uint32_t MaxDisplays = sizeof(CGOpenGLDisplayMask) * 8;
+        CGDirectDisplayID displays[MaxDisplays] = {};
+        std::uint32_t displayCount = 0;
+        if (displayMask != 0 &&
+            CGGetActiveDisplayList(MaxDisplays, displays, &displayCount) == kCGErrorSuccess) {
+            const auto mask = static_cast<CGOpenGLDisplayMask>(displayMask);
+            for (std::uint32_t i = 0; i < displayCount; ++i) {
+                if ((CGDisplayIDToOpenGLDisplayMask(displays[i]) & mask) != 0) {
+                    return displays[i];
+                }
+            }
+        }
+        return CGMainDisplayID();
+    }
+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Wdeprecated-declarations"
+    CVReturn MobileGLCVDisplayLinkSetCurrentCGDisplayFromOpenGLContext(
+        CVDisplayLinkRef displayLink,
+        CGLContextObj context,
+        CGLPixelFormatObj pixelFormat) {
+        GLint virtualScreen = 0;
+        if (MobileGL::MG_Impl::CGLImpl::GetVirtualScreen(context, &virtualScreen) == kCGLNoError) {
+            GLint displayMask = 0;
+            if (!displayLink ||
+                MobileGL::MG_Impl::CGLImpl::DescribePixelFormat(
+                    pixelFormat, virtualScreen, kCGLPFADisplayMask, &displayMask) != kCGLNoError) {
+                return kCVReturnInvalidArgument;
+            }
+            return CVDisplayLinkSetCurrentCGDisplay(displayLink, DisplayForMask(displayMask));
+        }
+
+        using OriginalFunction = CVReturn (*)(CVDisplayLinkRef, CGLContextObj, CGLPixelFormatObj);
+        static const auto original = reinterpret_cast<OriginalFunction>(
+            dlsym(RTLD_NEXT, "CVDisplayLinkSetCurrentCGDisplayFromOpenGLContext"));
+        return original ? original(displayLink, context, pixelFormat) : kCVReturnError;
+    }
+
     __attribute__((used)) static const DyldInterposeEntry kMobileGLDyldInterpose[]
         __attribute__((section("__DATA,__interpose"))) = {
             {reinterpret_cast<const void*>(MobileGLDlsym), reinterpret_cast<const void*>(dlsym)},
+            {reinterpret_cast<const void*>(MobileGLCVDisplayLinkSetCurrentCGDisplayFromOpenGLContext),
+             reinterpret_cast<const void*>(CVDisplayLinkSetCurrentCGDisplayFromOpenGLContext)},
         };
+#pragma clang diagnostic pop
 } // namespace
 
 #endif

MobileGL/MG_Impl/DyldInterpose/ExportedSymbols.txt

@@ -0,0 +1,10 @@
+# Public CGL entry points.
+_CGL*
+
+# Public EGL entry points.
+_egl*
+
+# Public OpenGL and GLX entry points. OpenGL function names always use an
+# uppercase letter or digit after the "gl" prefix; excluding lowercase here
+# deliberately prevents glslang_* from matching this pattern.
+_gl[A-Z0-9]*

MobileGL/MG_Impl/GLImpl/Buffer/GL_Buffer.cpp

@@ -330,12 +330,21 @@ namespace MobileGL::MG_Impl::GLImpl {
                     } else if (access & BufferMappingAccessBit::Write) {
                         *params = GL_WRITE_ONLY;
                     } else {
-                        *params = 0;
+                        *params = GL_READ_WRITE;
                     }
                 } else {
-                    *params = 0;
+                    // Initial value, and what glUnmapBuffer restores (GL 4.6 core table 6.2).
+                    *params = GL_READ_WRITE;
                 }
                 break;
+            case GL_BUFFER_ACCESS_FLAGS:
+                // The MapBufferRange flags verbatim; glMapBuffer's access enum has already been
+                // normalised into the same bits. Zero while the buffer is not mapped.
+                *params = bufferObject->IsMapped()
+                    ? static_cast<GLint>(
+                          MG_Util::ConvertBufferMappingAccessToGLEnum(bufferObject->GetMappingAccess()))
+                    : 0;
+                break;
             case GL_BUFFER_MAPPED:
                 *params = bufferObject->IsMapped() ? GL_TRUE : GL_FALSE;
                 break;
@@ -878,6 +887,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             return;
         }
 
+        bufferObject->SyncGpuWrites();
         bufferObject->DownloadSubData(data, static_cast<SizeT>(offset), static_cast<SizeT>(size));
     }
 
@@ -1351,6 +1361,14 @@ namespace MobileGL::MG_Impl::GLImpl {
         BufferTarget bufferTarget = MG_Util::ConvertGLEnumToBufferTarget(target);
         if (!BufferImpl::ValidateBufferBindingPointTarget(bufferTarget)) return;
         if (!BufferImpl::ValidateBufferBindingPointIndex(bufferTarget, pointIndex)) return;
+        if (bufferTarget == BufferTarget::TransformFeedback && MG_State::pGLContext->IsTransformFeedbackActive()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "Transform feedback buffer bindings cannot change while transform "
+                                             "feedback is active."));
+            return;
+        }
         MG_State::pGLContext->TouchBufferBindingPoint(bufferTarget, pointIndex);
 
         auto& point = MG_State::pGLContext->GetBufferBindingPoint(bufferTarget, pointIndex);
@@ -1358,6 +1376,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (buffer == 0) {
             point.Bind(nullptr);
             point.SetRange(Range1D(0, 0));
+            GetBufferBindingSlot(bufferTarget).Bind(nullptr);
             return;
         }
 
@@ -1376,6 +1395,12 @@ namespace MobileGL::MG_Impl::GLImpl {
         } else {
             point.ClearRange();
         }
+        // The indexed bind also binds to the generic binding point of the same target
+        // (GL 4.6 core 6.1.1). Callers rely on it: the texture_gather tests set up their
+        // SSBO with BindBufferBase and then size it through glBufferData on the generic
+        // target alone, which would otherwise raise GL_INVALID_OPERATION and leave the
+        // buffer with no storage.
+        GetBufferBindingSlot(bufferTarget).Bind(bufferObject);
     }
 
     void BindBufferRange_State(GLenum target, GLuint index, GLuint buffer, GLintptr offset, GLsizeiptr size) {
@@ -1384,6 +1409,14 @@ namespace MobileGL::MG_Impl::GLImpl {
         BufferTarget bufferTarget = MG_Util::ConvertGLEnumToBufferTarget(target);
         if (!BufferImpl::ValidateBufferBindingPointTarget(bufferTarget)) return;
         if (!BufferImpl::ValidateBufferBindingPointIndex(bufferTarget, index)) return;
+        if (bufferTarget == BufferTarget::TransformFeedback && MG_State::pGLContext->IsTransformFeedbackActive()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "Transform feedback buffer bindings cannot change while transform "
+                                             "feedback is active."));
+            return;
+        }
         MG_State::pGLContext->TouchBufferBindingPoint(bufferTarget, index);
 
         auto& point = MG_State::pGLContext->GetBufferBindingPoint(bufferTarget, index);
@@ -1391,6 +1424,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (buffer == 0) {
             point.Bind(nullptr);
             point.SetRange(Range1D(0, 0));
+            GetBufferBindingSlot(bufferTarget).Bind(nullptr);
             return;
         }
 
@@ -1408,6 +1442,8 @@ namespace MobileGL::MG_Impl::GLImpl {
         } else {
             point.ClearRange();
         }
+        // Also the generic binding point, exactly as BindBufferBase (GL 4.6 core 6.1.1).
+        GetBufferBindingSlot(bufferTarget).Bind(bufferObject);
     }
 
     /* @INSERTION_POINT:FUNCTION_IMPLEMENTATION@ */

MobileGL/MG_Impl/GLImpl/Buffer/Validators.cpp

@@ -60,6 +60,11 @@ namespace MobileGL::MG_Impl::GLImpl::BufferImpl {
                 MG_Backend::pActiveBackendObject->GetDynamicParameters().MaxShaderStorageBufferBindings;
             pointCount = std::min(pointCount, static_cast<SizeT>(std::max(backendCount, 0)));
         }
+        if (target == BufferTarget::TransformFeedback) {
+            // GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_ATTRIBS bounds the indexed capture
+            // binding points in GL 3.3 (no ARB_transform_feedback3).
+            pointCount = std::min<SizeT>(pointCount, 4);
+        }
 
         if (index < pointCount) {
             return true;

MobileGL/MG_Impl/GLImpl/Drawing/GL_Drawing.cpp

@@ -11,6 +11,7 @@
 #include <MG_State/GLState/Core.h>
 #include <MG_State/EGLState/Core.h>
 #include <MG_Backend/BackendObjects.h>
+#include "../Getter/GL_Getter.h"
 
 namespace MobileGL::MG_Impl::GLImpl {
     static Bool ValidateCurrentProgramForExecution(const char* functionName) {
@@ -48,7 +49,109 @@ namespace MobileGL::MG_Impl::GLImpl {
         return true;
     }
 
+    // Primitives a draw of `count` vertices in `mode` assembles (0 for
+    // incomplete primitives). Used for the CPU-side transform feedback
+    // primitive accounting.
+    static Uint64 CountPrimitivesForDraw(GLenum mode, GLsizei count) {
+        if (count <= 0) return 0;
+        switch (mode) {
+        case GL_POINTS: return static_cast<Uint64>(count);
+        case GL_LINES: return static_cast<Uint64>(count / 2);
+        case GL_LINE_STRIP: return count >= 2 ? static_cast<Uint64>(count - 1) : 0;
+        case GL_LINE_LOOP: return count >= 2 ? static_cast<Uint64>(count) : 0;
+        case GL_TRIANGLES: return static_cast<Uint64>(count / 3);
+        case GL_TRIANGLE_STRIP:
+        case GL_TRIANGLE_FAN: return count >= 3 ? static_cast<Uint64>(count - 2) : 0;
+        default: return 0;
+        }
+    }
+
+    // Accumulate the transform feedback primitive counter for a captured draw.
+    // Draws without a geometry stage write exactly the primitives they assemble,
+    // clamped by the capture buffers' remaining capacity (a full buffer stops
+    // recording whole primitives, which is what PRIMITIVES_WRITTEN reports).
+    // Geometry amplification is not modelled here.
+    static void AccountTransformFeedbackPrimitives(GLenum mode, GLsizei count) {
+        if (!MG_State::pGLContext->IsTransformFeedbackActive()) return;
+        // A paused span captures nothing, so a draw made while paused contributes to
+        // PRIMITIVES_GENERATED but not to TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN.
+        if (MG_State::pGLContext->IsTransformFeedbackPaused()) {
+            MG_State::pGLContext->AddTransformFeedbackPausedPrimitives(CountPrimitivesForDraw(mode, count));
+            return;
+        }
+        Uint64 primitives = CountPrimitivesForDraw(mode, count);
+        if (primitives == 0) return;
+        MG_State::pGLContext->AddTransformFeedbackInputPrimitives(primitives);
+
+        Uint64 verticesPerPrimitive = 1;
+        switch (mode) {
+        case GL_LINES:
+        case GL_LINE_STRIP:
+        case GL_LINE_LOOP:
+            verticesPerPrimitive = 2;
+            break;
+        case GL_TRIANGLES:
+        case GL_TRIANGLE_STRIP:
+        case GL_TRIANGLE_FAN:
+            verticesPerPrimitive = 3;
+            break;
+        default:
+            break;
+        }
+
+        const auto& program = MG_State::pGLContext->GetTransformFeedbackProgram();
+        if (program != nullptr) {
+            // Capacity in captured vertices = the tightest bound buffer.
+            Uint64 capacityVertices = ~0ull;
+            for (SizeT i = 0; i < program->GetTransformFeedbackBufferCount(); ++i) {
+                const Uint32 stride = program->GetTransformFeedbackStride(static_cast<Uint32>(i));
+                if (stride == 0) continue;
+                const auto& point = MG_State::pGLContext->GetBufferBindingPoint(BufferTarget::TransformFeedback,
+                                                                                static_cast<Uint>(i));
+                const Range1D range = point.GetRange();
+                const Uint64 bytes = range.end > range.start ? static_cast<Uint64>(range.end - range.start) : 0;
+                capacityVertices = std::min<Uint64>(capacityVertices, bytes / stride);
+            }
+            if (capacityVertices != ~0ull) {
+                const Uint64 usedVertices = MG_State::pGLContext->GetTransformFeedbackCapturedVertices();
+                const Uint64 remainingVertices = capacityVertices > usedVertices ? capacityVertices - usedVertices : 0;
+                primitives = std::min<Uint64>(primitives, remainingVertices / verticesPerPrimitive);
+            }
+        }
+        MG_State::pGLContext->AddTransformFeedbackPrimitives(primitives);
+        MG_State::pGLContext->AddTransformFeedbackCapturedVertices(primitives * verticesPerPrimitive);
+    }
+
+    // Every primitive mode a draw command accepts (GL 4.6 core table 10.1, plus
+    // GL_PATCHES for the tessellation pipeline). Anything else is GL_INVALID_ENUM.
+    static Bool IsAcceptedPrimitiveMode(GLenum mode) {
+        switch (mode) {
+        case GL_POINTS:
+        case GL_LINES:
+        case GL_LINE_LOOP:
+        case GL_LINE_STRIP:
+        case GL_LINES_ADJACENCY:
+        case GL_LINE_STRIP_ADJACENCY:
+        case GL_TRIANGLES:
+        case GL_TRIANGLE_STRIP:
+        case GL_TRIANGLE_FAN:
+        case GL_TRIANGLES_ADJACENCY:
+        case GL_TRIANGLE_STRIP_ADJACENCY:
+        case GL_PATCHES:
+            return true;
+        default:
+            return false;
+        }
+    }
+
     static Bool ValidatePrimitiveModeForBackend(const char* functionName, GLenum mode) {
+        if (!IsAcceptedPrimitiveMode(mode)) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName, "mode is not an accepted primitive type."));
+            return false;
+        }
+
         const auto& activeBackendObject = MG_Backend::pActiveBackendObject;
         if (!activeBackendObject) {
             MG_State::pGLContext->RecordError(
@@ -57,15 +160,6 @@ namespace MobileGL::MG_Impl::GLImpl {
             return false;
         }
 
-        if (activeBackendObject->GetBackendType() == BackendType::DirectVulkan && mode == GL_LINE_LOOP) {
-            MG_State::pGLContext->RecordError(
-                ErrorCode::InvalidOperation,
-                MakeUnique<GenericErrorInfo>(
-                    "MG_Impl/GLImpl", functionName,
-                    "Primitive mode GL_LINE_LOOP is not supported by the DirectVulkan backend."));
-            return false;
-        }
-
         const auto& vao = MG_State::pGLContext->GetBoundVertexArray();
         if (vao && vao->GetExternalIndex() == 0 && !MG_State::IsRelaxedSemanticsActive()) {
             MG_State::pGLContext->RecordError(
@@ -75,9 +169,133 @@ namespace MobileGL::MG_Impl::GLImpl {
             return false;
         }
 
+        // A geometry stage only accepts the primitive types that decompose into its declared
+        // input primitive (GL 4.6 core 11.3.1); anything else is INVALID_OPERATION. GL_PATCHES
+        // is the tessellation pipeline's input and reaches the geometry stage already
+        // converted, so it is not constrained here.
+        const auto& currentProgram = MG_State::pGLContext->GetCurrentProgram();
+        const GLenum gsInput = currentProgram ? currentProgram->GetGeometryInputType() : GL_NONE;
+        if (gsInput != GL_NONE && mode != GL_PATCHES) {
+            Bool compatible = false;
+            switch (gsInput) {
+            case GL_POINTS:
+                compatible = mode == GL_POINTS;
+                break;
+            case GL_LINES:
+                compatible = mode == GL_LINES || mode == GL_LINE_STRIP || mode == GL_LINE_LOOP;
+                break;
+            case GL_LINES_ADJACENCY:
+                compatible = mode == GL_LINES_ADJACENCY || mode == GL_LINE_STRIP_ADJACENCY;
+                break;
+            case GL_TRIANGLES:
+                compatible = mode == GL_TRIANGLES || mode == GL_TRIANGLE_STRIP || mode == GL_TRIANGLE_FAN;
+                break;
+            case GL_TRIANGLES_ADJACENCY:
+                compatible = mode == GL_TRIANGLES_ADJACENCY || mode == GL_TRIANGLE_STRIP_ADJACENCY;
+                break;
+            default:
+                break;
+            }
+            if (!compatible) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>(
+                        "MG_Impl/GLImpl", functionName,
+                        "Primitive mode is incompatible with the geometry shader's input primitive type."));
+                return false;
+            }
+        }
+
+        // While transform feedback is active the draw's primitive type must match
+        // the feedback primitive mode (GL 3.3 core 13.2.2). With a geometry shader
+        // the constraint moves to the shader's output primitive type instead, so
+        // the draw mode itself is unconstrained here. A paused span is exempt: it
+        // captures nothing, so there is nothing for the mode to be incompatible with
+        // (GL 4.6 core 13.2.3).
+        if (MG_State::pGLContext->IsTransformFeedbackActive() &&
+            !MG_State::pGLContext->IsTransformFeedbackPaused() &&
+            !(MG_State::pGLContext->GetTransformFeedbackProgram() &&
+              MG_State::pGLContext->GetTransformFeedbackProgram()->GetShaderIndexByStage(ShaderStage::Geometry) >= 0)) {
+            const GLenum feedbackMode = MG_State::pGLContext->GetTransformFeedbackPrimitiveMode();
+            Bool compatible = false;
+            switch (feedbackMode) {
+            case GL_POINTS:
+                compatible = mode == GL_POINTS;
+                break;
+            case GL_LINES:
+                compatible = mode == GL_LINES || mode == GL_LINE_STRIP || mode == GL_LINE_LOOP;
+                break;
+            case GL_TRIANGLES:
+                compatible = mode == GL_TRIANGLES || mode == GL_TRIANGLE_STRIP || mode == GL_TRIANGLE_FAN;
+                break;
+            default:
+                break;
+            }
+            if (!compatible) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>(
+                        "MG_Impl/GLImpl", functionName,
+                        "Primitive mode is incompatible with the active transform feedback primitive mode."));
+                return false;
+            }
+        }
+
         return true;
     }
 
+    // Byte size of the command structures the indirect draws read (GL 4.6 core 10.3.10).
+    constexpr SizeT kDrawArraysIndirectCommandBytes = 4 * sizeof(Uint32);
+    constexpr SizeT kDrawElementsIndirectCommandBytes = 5 * sizeof(Uint32);
+
+    // Shared preconditions of every *Indirect draw: `indirect` is a byte offset into the
+    // buffer bound to GL_DRAW_INDIRECT_BUFFER, must be 4-byte aligned, and the whole
+    // command has to lie inside that buffer.
+    static Bool ValidateIndirectDrawSource(const char* functionName, const void* indirect, SizeT commandBytes) {
+        const auto offset = reinterpret_cast<uintptr_t>(indirect);
+        if (offset % 4 != 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                             "indirect offset must be a multiple of 4."));
+            return false;
+        }
+
+        const auto& buffer = MG_State::pGLContext->GetBufferBindingSlot(BufferTarget::DrawIndirect).GetBoundObject();
+        if (!buffer) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                             "No buffer is bound to GL_DRAW_INDIRECT_BUFFER."));
+            return false;
+        }
+
+        if (offset + commandBytes > buffer->GetSize()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                             "The indirect command extends past the end of the bound "
+                                             "GL_DRAW_INDIRECT_BUFFER."));
+            return false;
+        }
+        return true;
+    }
+
+    // Index type accepted by the DrawElements family (GL 4.6 core 10.3.9).
+    static Bool ValidateDrawElementsIndexType(const char* functionName, GLenum type) {
+        switch (type) {
+        case GL_UNSIGNED_BYTE:
+        case GL_UNSIGNED_SHORT:
+        case GL_UNSIGNED_INT:
+            return true;
+        default:
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName, "type is not an accepted index type."));
+            return false;
+        }
+    }
+
     void Clear_Backend(GLbitfield mask) {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
@@ -272,6 +490,28 @@ namespace MobileGL::MG_Impl::GLImpl {
         dispatchComputeIndirect(indirect);
     }
 
+    void PatchParameteri(GLenum pname, GLint value) {
+        if (pname != GL_PATCH_VERTICES) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "pname must be GL_PATCH_VERTICES."));
+            return;
+        }
+        GLint maxPatchVertices = 32;
+        GetIntegerv(GL_MAX_PATCH_VERTICES, &maxPatchVertices);
+        if (value <= 0 || value > maxPatchVertices) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "value must be in [1, GL_MAX_PATCH_VERTICES]."));
+            return;
+        }
+        MG_State::pGLContext->SetPatchVertices(static_cast<Uint>(value));
+        if (const auto patchParameteri = MG_Backend::gBackendFunctionsTable.GL.PatchParameteri) {
+            patchParameteri(pname, value);
+        }
+    }
+
     void MemoryBarrier(GLbitfield barriers) {
         auto memoryBarrier = MG_Backend::gBackendFunctionsTable.GL.MemoryBarrier;
         if (!memoryBarrier) {
@@ -377,6 +617,8 @@ namespace MobileGL::MG_Impl::GLImpl {
     void DrawElementsIndirect(GLenum mode, GLenum type, const void* indirect) {
         if (!ValidateCurrentProgramForExecution(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
+        if (!ValidateDrawElementsIndexType(__func__, type)) return;
+        if (!ValidateIndirectDrawSource(__func__, indirect, kDrawElementsIndirectCommandBytes)) return;
         DrawElementsIndirect_Backend(mode, type, indirect);
     }
 
@@ -396,18 +638,21 @@ namespace MobileGL::MG_Impl::GLImpl {
     void DrawArraysIndirect(GLenum mode, const void* indirect) {
         if (!ValidateCurrentProgramForExecution(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
+        if (!ValidateIndirectDrawSource(__func__, indirect, kDrawArraysIndirectCommandBytes)) return;
         DrawArraysIndirect_Backend(mode, indirect);
     }
 
     void DrawElementsBaseVertex(GLenum mode, GLsizei count, GLenum type, const void* indices, GLint basevertex) {
         if (!ValidateCurrentProgramForExecution(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
+        AccountTransformFeedbackPrimitives(mode, count);
         DrawElementsBaseVertex_Backend(mode, count, type, indices, basevertex);
     }
 
     void DrawArrays(GLenum mode, GLint first, GLsizei count) {
         if (!ValidateCurrentProgramForExecution(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
+        AccountTransformFeedbackPrimitives(mode, count);
         DrawArrays_Backend(mode, first, count);
     }
 
@@ -444,7 +689,487 @@ namespace MobileGL::MG_Impl::GLImpl {
     void DrawElements(GLenum mode, GLsizei count, GLenum type, const void* indices) {
         if (!ValidateCurrentProgramForExecution(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
+        AccountTransformFeedbackPrimitives(mode, count);
         DrawElements_Backend(mode, count, type, indices);
     }
 
+    void BeginTransformFeedback(GLenum primitiveMode) {
+        if (primitiveMode != GL_POINTS && primitiveMode != GL_LINES && primitiveMode != GL_TRIANGLES) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "primitiveMode must be GL_POINTS, GL_LINES or GL_TRIANGLES."));
+            return;
+        }
+        if (MG_State::pGLContext->IsTransformFeedbackActive()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "Transform feedback is already active."));
+            return;
+        }
+        const auto& program = MG_State::pGLContext->GetCurrentProgram();
+        if (!program || !program->GetLinkStatus() || program->GetTransformFeedbackVaryingCount() == 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>(
+                    "MG_Impl/GLImpl", __func__,
+                    "No program with transform feedback varyings is active."));
+            return;
+        }
+        // Every capture buffer slot the program's mode uses must have a buffer bound. A slot
+        // of stride 0 - two consecutive gl_NextBuffer entries - captures nothing and so needs
+        // no binding.
+        const SizeT usedBufferCount = program->GetTransformFeedbackBufferCount();
+        for (SizeT i = 0; i < usedBufferCount; ++i) {
+            if (program->GetTransformFeedbackStride(static_cast<Uint32>(i)) == 0) continue;
+            const auto& point = MG_State::pGLContext->GetBufferBindingPoint(BufferTarget::TransformFeedback,
+                                                                            static_cast<Uint>(i));
+            if (point.GetBoundObject() == nullptr) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>(
+                        "MG_Impl/GLImpl", __func__,
+                        "Transform feedback buffer binding point " + std::to_string(i) + " has no buffer bound."));
+                return;
+            }
+        }
+        MG_State::pGLContext->BeginTransformFeedback(primitiveMode, program);
+        if (const auto beginXfb = MG_Backend::gBackendFunctionsTable.GL.BeginTransformFeedback) {
+            beginXfb(primitiveMode);
+        }
+    }
+
+    // Vulkan transform feedback captures triangle strips in plain (i, i+1, i+2)
+    // vertex order, but GL decomposes odd strip triangles as (i+1, i, i+2)
+    // (GL 4.6 table 10.1). With the geometry stage's statically-known strip
+    // lengths the captured records are reordered in place: swap the first two
+    // vertex records of every odd triangle within each emitted strip.
+    static void FixupGsStripCaptureOrder(const SharedPtr<MG_State::GLState::ProgramObject>& program,
+                                         Uint64 inputPrimitives) {
+        // Only Vulkan-order captures need this. A backend that runs the capture on its
+        // own GL/ES driver (it owns the span, hence the EndTransformFeedback entry) has
+        // already produced GL's vertex order, and reordering it again would corrupt it.
+        if (MG_Backend::gBackendFunctionsTable.GL.EndTransformFeedback != nullptr) {
+            return;
+        }
+        if (program == nullptr || !program->HasGsTriangleStripCaptureFixup() || inputPrimitives == 0) {
+            return;
+        }
+        const auto& stripTriangles = program->GetGsStripTriangles();
+
+        // Global triangle indices whose leading vertex pair must swap.
+        Vector<Uint64> swapTriangles;
+        Uint64 triangleBase = 0;
+        for (Uint64 input = 0; input < inputPrimitives; ++input) {
+            for (const Uint32 stripLength : stripTriangles) {
+                for (Uint32 t = 1; t < stripLength; t += 2) {
+                    swapTriangles.push_back(triangleBase + t);
+                }
+                triangleBase += stripLength;
+            }
+        }
+        if (swapTriangles.empty()) {
+            return;
+        }
+
+        for (SizeT bufferIndex = 0; bufferIndex < program->GetTransformFeedbackBufferCount(); ++bufferIndex) {
+            const Uint32 stride = program->GetTransformFeedbackStride(static_cast<Uint32>(bufferIndex));
+            if (stride == 0) continue;
+            const auto& bindingPoint =
+                MG_State::pGLContext->GetBufferBindingPoint(BufferTarget::TransformFeedback,
+                                                            static_cast<Uint>(bufferIndex));
+            const auto& buffer = bindingPoint.GetBoundObject();
+            if (buffer == nullptr) continue;
+            const Range1D range = bindingPoint.GetRange();
+            const Uint8* mapped = buffer->MappedData();
+            if (mapped == nullptr) continue;
+            // The geometry stage amplifies, so the CPU vertex counter does not bound
+            // the capture; the binding range's whole-triangle capacity does.
+            const Uint64 rangeBytes = range.end > range.start ? static_cast<Uint64>(range.end - range.start) : 0;
+            const Uint64 capturedTriangles = std::min<Uint64>(triangleBase, (rangeBytes / stride) / 3);
+
+            // Observed Vulkan capture order for odd strip triangles is (i, i+2, i+1)
+            // (winding preserved by swapping the trailing pair); GL wants
+            // (i+1, i, i+2), which is one rotation away: (a,b,c) -> (c,a,b).
+            Vector<Uint8> scratch(stride);
+            for (const Uint64 triangle : swapTriangles) {
+                if (triangle >= capturedTriangles) break;
+                const SizeT v0Offset = static_cast<SizeT>(range.start) + static_cast<SizeT>(triangle * 3) * stride;
+                const SizeT v1Offset = v0Offset + stride;
+                const SizeT v2Offset = v1Offset + stride;
+                Memcpy(scratch.data(), mapped + v2Offset, stride);
+                buffer->WritebackFromBackend({const_cast<Uint8*>(mapped) + v1Offset, stride}, v2Offset);
+                buffer->WritebackFromBackend({const_cast<Uint8*>(mapped) + v0Offset, stride}, v1Offset);
+                buffer->WritebackFromBackend({scratch.data(), stride}, v0Offset);
+            }
+        }
+    }
+
+    void EndTransformFeedback(void) {
+        if (!MG_State::pGLContext->IsTransformFeedbackActive()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "Transform feedback is not active."));
+            return;
+        }
+        const auto capturedProgram = MG_State::pGLContext->GetTransformFeedbackProgram();
+        const Uint64 inputPrimitives = MG_State::pGLContext->GetTransformFeedbackInputPrimitives();
+        // Closed while the capture state is still active: a backend that captures
+        // through its own driver reads the capture program and buffer bindings here.
+        if (const auto endXfb = MG_Backend::gBackendFunctionsTable.GL.EndTransformFeedback) {
+            endXfb();
+        }
+        MG_State::pGLContext->EndTransformFeedback();
+        // Captured results must be visible to MapBuffer/GetBufferSubData after
+        // End; the capture targets are host-coherent GPU memory, so completing
+        // the GPU work is all that is required.
+        auto& backendGL = MG_Backend::gBackendFunctionsTable.GL;
+        if (backendGL.FenceSync && backendGL.ClientWaitSync) {
+            if (auto sync = backendGL.FenceSync()) {
+                backendGL.ClientWaitSync(sync, GL_SYNC_FLUSH_COMMANDS_BIT, ~0ull);
+                if (backendGL.DeleteSync) {
+                    backendGL.DeleteSync(sync);
+                }
+            }
+        }
+        FixupGsStripCaptureOrder(capturedProgram, inputPrimitives);
+    }
+
+    void PauseTransformFeedback(void) {
+        if (!MG_State::pGLContext->IsTransformFeedbackActive() ||
+            MG_State::pGLContext->IsTransformFeedbackPaused()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "Transform feedback is not active, or is already paused."));
+            return;
+        }
+        MG_State::pGLContext->SetTransformFeedbackPaused(true);
+        if (const auto pauseXfb = MG_Backend::gBackendFunctionsTable.GL.PauseTransformFeedback) {
+            pauseXfb();
+        }
+    }
+
+    void ResumeTransformFeedback(void) {
+        if (!MG_State::pGLContext->IsTransformFeedbackActive() ||
+            !MG_State::pGLContext->IsTransformFeedbackPaused()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "Transform feedback is not paused."));
+            return;
+        }
+        MG_State::pGLContext->SetTransformFeedbackPaused(false);
+        if (const auto resumeXfb = MG_Backend::gBackendFunctionsTable.GL.ResumeTransformFeedback) {
+            resumeXfb();
+        }
+    }
+
+    void GenTransformFeedbacks(GLsizei n, GLuint* ids) {
+        if (n < 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "n must be non-negative."));
+            return;
+        }
+        if (n == 0 || ids == nullptr) return;
+        Vector<Uint> names;
+        MG_State::pGLContext->GenTransformFeedbackNames(static_cast<Uint>(n), names);
+        Memcpy(ids, names.data(), static_cast<SizeT>(n) * sizeof(GLuint));
+    }
+
+    void CreateTransformFeedbacks(GLsizei n, GLuint* ids) {
+        if (n < 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "n must be non-negative."));
+            return;
+        }
+        if (n == 0 || ids == nullptr) return;
+        Vector<Uint> names;
+        MG_State::pGLContext->GenTransformFeedbackNames(static_cast<Uint>(n), names);
+        // Unlike glGenTransformFeedbacks, the names are objects immediately: there is no bind step
+        // to create them from (GL 4.6 core 13.2.1).
+        for (const Uint name : names) {
+            MG_State::pGLContext->CreateTransformFeedbackObject(name);
+        }
+        Memcpy(ids, names.data(), static_cast<SizeT>(n) * sizeof(GLuint));
+    }
+
+    namespace {
+        // Shared front half of the by-name transform feedback entry points: the object has to exist
+        // (INVALID_OPERATION otherwise) before anything else about the call is looked at.
+        Bool ValidateNamedTransformFeedback(GLuint xfb, const char* functionName) {
+            if (!MG_State::pGLContext->IsTransformFeedbackObject(xfb)) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                                 std::to_string(xfb) + " is not a transform feedback object."));
+                return false;
+            }
+            return true;
+        }
+
+        Bool ValidateTransformFeedbackBufferIndex(GLuint index, const char* functionName) {
+            if (index >= MG_State::GLState::GLContext::MAX_TRANSFORM_FEEDBACK_BUFFERS) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidValue,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                                 "index exceeds GL_MAX_TRANSFORM_FEEDBACK_BUFFERS."));
+                return false;
+            }
+            return true;
+        }
+
+        // A capture binding may not be changed while the object is capturing (GL 4.6 core 13.2.2).
+        Bool ValidateNamedTransformFeedbackNotActive(GLuint xfb, const char* functionName) {
+            if (MG_State::pGLContext->IsNamedTransformFeedbackActive(xfb)) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                                 "The transform feedback object is capturing."));
+                return false;
+            }
+            return true;
+        }
+
+        SharedPtr<MG_State::GLState::BufferObject> ResolveTransformFeedbackBuffer(GLuint buffer,
+                                                                                 const char* functionName) {
+            if (buffer == 0) return nullptr;
+            if (!MG_State::pGLContext->ValidateBufferName(buffer)) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                                 std::to_string(buffer) + " is not a buffer object."));
+                return nullptr;
+            }
+            return MG_State::pGLContext->GetBufferObject(buffer);
+        }
+    } // namespace
+
+    void TransformFeedbackBufferBase(GLuint xfb, GLuint index, GLuint buffer) {
+        if (!ValidateNamedTransformFeedback(xfb, __func__)) return;
+        if (!ValidateTransformFeedbackBufferIndex(index, __func__)) return;
+        if (!ValidateNamedTransformFeedbackNotActive(xfb, __func__)) return;
+        if (buffer != 0 && !MG_State::pGLContext->ValidateBufferName(buffer)) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             std::to_string(buffer) + " is not a buffer object."));
+            return;
+        }
+        MG_State::pGLContext->SetNamedTransformFeedbackBinding(xfb, index,
+                                                               ResolveTransformFeedbackBuffer(buffer, __func__), {},
+                                                               false);
+    }
+
+    void TransformFeedbackBufferRange(GLuint xfb, GLuint index, GLuint buffer, GLintptr offset, GLsizeiptr size) {
+        if (!ValidateNamedTransformFeedback(xfb, __func__)) return;
+        if (!ValidateTransformFeedbackBufferIndex(index, __func__)) return;
+        if (!ValidateNamedTransformFeedbackNotActive(xfb, __func__)) return;
+        if (offset < 0 || size <= 0 || (offset % 4) != 0 || (size % 4) != 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "offset and size must be non-negative multiples of 4."));
+            return;
+        }
+        if (buffer != 0 && !MG_State::pGLContext->ValidateBufferName(buffer)) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             std::to_string(buffer) + " is not a buffer object."));
+            return;
+        }
+        auto bufferObject = ResolveTransformFeedbackBuffer(buffer, __func__);
+        const Range1D range{static_cast<SizeT>(offset), static_cast<SizeT>(offset) + static_cast<SizeT>(size)};
+        MG_State::pGLContext->SetNamedTransformFeedbackBinding(xfb, index, bufferObject, range,
+                                                               bufferObject != nullptr);
+    }
+
+    void GetTransformFeedbackiv(GLuint xfb, GLenum pname, GLint* param) {
+        if (!ValidateNamedTransformFeedback(xfb, __func__)) return;
+        if (!param) return;
+        switch (pname) {
+        case GL_TRANSFORM_FEEDBACK_ACTIVE:
+            *param = MG_State::pGLContext->IsNamedTransformFeedbackActive(xfb) ? GL_TRUE : GL_FALSE;
+            return;
+        case GL_TRANSFORM_FEEDBACK_PAUSED:
+            *param = MG_State::pGLContext->IsNamedTransformFeedbackPaused(xfb) ? GL_TRUE : GL_FALSE;
+            return;
+        default:
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "pname must be GL_TRANSFORM_FEEDBACK_ACTIVE or _PAUSED."));
+            return;
+        }
+    }
+
+    void GetTransformFeedbacki_v(GLuint xfb, GLenum pname, GLuint index, GLint* param) {
+        if (!ValidateNamedTransformFeedback(xfb, __func__)) return;
+        if (pname != GL_TRANSFORM_FEEDBACK_BUFFER_BINDING) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "pname must be GL_TRANSFORM_FEEDBACK_BUFFER_BINDING."));
+            return;
+        }
+        if (!ValidateTransformFeedbackBufferIndex(index, __func__)) return;
+        if (!param) return;
+        const auto binding = MG_State::pGLContext->GetNamedTransformFeedbackBinding(xfb, index);
+        *param = binding.Buffer ? static_cast<GLint>(binding.Buffer->GetExternalIndex()) : 0;
+    }
+
+    void GetTransformFeedbacki64_v(GLuint xfb, GLenum pname, GLuint index, GLint64* param) {
+        if (!ValidateNamedTransformFeedback(xfb, __func__)) return;
+        if (pname != GL_TRANSFORM_FEEDBACK_BUFFER_START && pname != GL_TRANSFORM_FEEDBACK_BUFFER_SIZE) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "pname must be GL_TRANSFORM_FEEDBACK_BUFFER_START or _SIZE."));
+            return;
+        }
+        if (!ValidateTransformFeedbackBufferIndex(index, __func__)) return;
+        if (!param) return;
+        const auto binding = MG_State::pGLContext->GetNamedTransformFeedbackBinding(xfb, index);
+        // glTransformFeedbackBufferBase leaves both at zero; only the range form sets them
+        // (GL 4.6 core table 23.48).
+        if (!binding.Buffer || !binding.HasExplicitRange) {
+            *param = 0;
+            return;
+        }
+        *param = (pname == GL_TRANSFORM_FEEDBACK_BUFFER_START)
+                     ? static_cast<GLint64>(binding.Range.start)
+                     : static_cast<GLint64>(binding.Range.end - binding.Range.start);
+    }
+
+    void DeleteTransformFeedbacks(GLsizei n, const GLuint* ids) {
+        if (n < 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "n must be non-negative."));
+            return;
+        }
+        if (ids == nullptr) return;
+        for (GLsizei i = 0; i < n; ++i) {
+            const GLuint id = ids[i];
+            // Unknown names and 0 are silently ignored; an object whose capture span is
+            // still open is not (GL 4.6 core 13.2.1).
+            if (id == 0 || !MG_State::pGLContext->ValidateTransformFeedbackName(id)) continue;
+            if (id == MG_State::pGLContext->GetBoundTransformFeedbackName() &&
+                MG_State::pGLContext->IsTransformFeedbackActive()) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                                 "Cannot delete a transform feedback object whose capture is active."));
+                continue;
+            }
+            if (const auto deleteXfb = MG_Backend::gBackendFunctionsTable.GL.DeleteTransformFeedback) {
+                deleteXfb(id);
+            }
+            MG_State::pGLContext->MarkTransformFeedbackObjectForDeletion(id);
+        }
+    }
+
+    void BindTransformFeedback(GLenum target, GLuint id) {
+        if (target != GL_TRANSFORM_FEEDBACK) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "target must be GL_TRANSFORM_FEEDBACK."));
+            return;
+        }
+        // A running capture pins its object; only a paused one may be swapped out.
+        if (MG_State::pGLContext->IsTransformFeedbackActive() &&
+            !MG_State::pGLContext->IsTransformFeedbackPaused()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "Transform feedback is active and not paused."));
+            return;
+        }
+        if (!MG_State::pGLContext->ValidateTransformFeedbackName(id)) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             std::to_string(id) + " is not a transform feedback object name."));
+            return;
+        }
+        MG_State::pGLContext->BindTransformFeedbackObject(id);
+        if (const auto bindXfb = MG_Backend::gBackendFunctionsTable.GL.BindTransformFeedback) {
+            bindXfb(id);
+        }
+    }
+
+    GLboolean IsTransformFeedback(GLuint id) {
+        // Name 0 is the default object, and a name glGenTransformFeedbacks handed out only
+        // becomes the name of an object once it has been bound.
+        return MG_State::pGLContext->IsTransformFeedbackObject(id) ? GL_TRUE : GL_FALSE;
+    }
+
+    // glDrawTransformFeedback[Stream][Instanced]: replays the vertices the named object
+    // captured in its last completed span, as if by glDrawArraysInstanced with that count
+    // (GL 4.6 core 10.3.7).
+    static void DrawTransformFeedbackImpl(const char* functionName, GLenum mode, GLuint id, GLuint stream,
+                                          GLsizei instancecount) {
+        if (!ValidateCurrentProgramForExecution(functionName)) return;
+        if (!ValidatePrimitiveModeForBackend(functionName, mode)) return;
+        if (instancecount < 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName, "instancecount must be non-negative."));
+            return;
+        }
+        if (!MG_State::pGLContext->ValidateTransformFeedbackName(id)) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                             std::to_string(id) + " is not a transform feedback object name."));
+            return;
+        }
+        // GL_MAX_VERTEX_STREAMS is 1, so stream 0 is the only one that exists.
+        if (stream != 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                             "stream must be less than GL_MAX_VERTEX_STREAMS."));
+            return;
+        }
+        // Drawing from an object whose capture is currently open is legal and deliberate:
+        // it is how a transform feedback result is fed straight back into the next span
+        // (ARB_transform_feedback2 lists no such restriction).
+        if (!MG_State::pGLContext->HasTransformFeedbackCompletedSpan(id)) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                             "glEndTransformFeedback has never been called for this object."));
+            return;
+        }
+
+        const Uint64 vertices = MG_State::pGLContext->GetTransformFeedbackRecordedVertices(id);
+        if (vertices == 0) return;
+        const auto count = static_cast<GLsizei>(vertices);
+        AccountTransformFeedbackPrimitives(mode, count);
+        if (instancecount == 1) {
+            DrawArrays_Backend(mode, 0, count);
+        } else {
+            DrawArraysInstanced_Backend(mode, 0, count, instancecount);
+        }
+    }
+
+    void DrawTransformFeedback(GLenum mode, GLuint id) {
+        DrawTransformFeedbackImpl(__func__, mode, id, 0, 1);
+    }
+
+    void DrawTransformFeedbackInstanced(GLenum mode, GLuint id, GLsizei instancecount) {
+        DrawTransformFeedbackImpl(__func__, mode, id, 0, instancecount);
+    }
+
+    void DrawTransformFeedbackStream(GLenum mode, GLuint id, GLuint stream) {
+        DrawTransformFeedbackImpl(__func__, mode, id, stream, 1);
+    }
+
+    void DrawTransformFeedbackStreamInstanced(GLenum mode, GLuint id, GLuint stream, GLsizei instancecount) {
+        DrawTransformFeedbackImpl(__func__, mode, id, stream, instancecount);
+    }
+
 } // namespace MobileGL::MG_Impl::GLImpl

MobileGL/MG_Impl/GLImpl/Drawing/GL_Drawing.h

@@ -11,8 +11,27 @@
 
 namespace MobileGL::MG_Impl::GLImpl {
     /* @INSERTION_POINT:FUNCTION_DECLARATION@ */
+    void BeginTransformFeedback(GLenum primitiveMode);
+    void EndTransformFeedback(void);
+    void PauseTransformFeedback(void);
+    void ResumeTransformFeedback(void);
+    void GenTransformFeedbacks(GLsizei n, GLuint* ids);
+    void CreateTransformFeedbacks(GLsizei n, GLuint* ids);
+    void DeleteTransformFeedbacks(GLsizei n, const GLuint* ids);
+    void TransformFeedbackBufferBase(GLuint xfb, GLuint index, GLuint buffer);
+    void TransformFeedbackBufferRange(GLuint xfb, GLuint index, GLuint buffer, GLintptr offset, GLsizeiptr size);
+    void GetTransformFeedbackiv(GLuint xfb, GLenum pname, GLint* param);
+    void GetTransformFeedbacki_v(GLuint xfb, GLenum pname, GLuint index, GLint* param);
+    void GetTransformFeedbacki64_v(GLuint xfb, GLenum pname, GLuint index, GLint64* param);
+    void BindTransformFeedback(GLenum target, GLuint id);
+    GLboolean IsTransformFeedback(GLuint id);
+    void DrawTransformFeedback(GLenum mode, GLuint id);
+    void DrawTransformFeedbackInstanced(GLenum mode, GLuint id, GLsizei instancecount);
+    void DrawTransformFeedbackStream(GLenum mode, GLuint id, GLuint stream);
+    void DrawTransformFeedbackStreamInstanced(GLenum mode, GLuint id, GLuint stream, GLsizei instancecount);
     void DispatchCompute(GLuint numGroupsX, GLuint numGroupsY, GLuint numGroupsZ);
     void DispatchComputeIndirect(GLintptr indirect);
+    void PatchParameteri(GLenum pname, GLint value);
     void MemoryBarrier(GLbitfield barriers);
     void MemoryBarrierByRegion(GLbitfield barriers);
     void MultiDrawElementsIndirect(GLenum mode, GLenum type, const void* indirect, GLsizei drawcount, GLsizei stride);

MobileGL/MG_Impl/GLImpl/Exporting/Definitions.cpp

@@ -236,12 +236,12 @@ DECLARE_GL_FUNCTION_HEAD(void, DeleteVertexArrays, GLsizei n, const GLuint* arra
 DECLARE_GL_FUNCTION_HEAD(void, GenVertexArrays, GLsizei n, GLuint* arrays) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GenVertexArrays, n, arrays)
 DECLARE_GL_FUNCTION_HEAD(GLboolean, IsVertexArray, GLuint array) DECLARE_GL_FUNCTION_END(GLboolean, IsVertexArray, array)
 DECLARE_GL_FUNCTION_HEAD(void, GetIntegeri_v, GLenum target, GLuint index, GLint* data) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetIntegeri_v, target, index, data)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, BeginTransformFeedback, GLenum primitiveMode) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, BeginTransformFeedback, primitiveMode)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, EndTransformFeedback) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, EndTransformFeedback)
+DECLARE_GL_FUNCTION_HEAD(void, BeginTransformFeedback, GLenum primitiveMode) DECLARE_GL_FUNCTION_END_NO_RETURN(void, BeginTransformFeedback, primitiveMode)
+DECLARE_GL_FUNCTION_HEAD(void, EndTransformFeedback) DECLARE_GL_FUNCTION_END_NO_RETURN(void, EndTransformFeedback)
 DECLARE_GL_FUNCTION_HEAD(void, BindBufferRange, GLenum target, GLuint index, GLuint buffer, GLintptr offset, GLsizeiptr size) DECLARE_GL_FUNCTION_END_NO_RETURN(void, BindBufferRange, target, index, buffer, offset, size)
 DECLARE_GL_FUNCTION_HEAD(void, BindBufferBase, GLenum target, GLuint index, GLuint buffer) DECLARE_GL_FUNCTION_END_NO_RETURN(void, BindBufferBase, target, index, buffer)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, TransformFeedbackVaryings, GLuint program, GLsizei count, const GLchar* const* varyings, GLenum bufferMode) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, TransformFeedbackVaryings, program, count, varyings, bufferMode)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, GetTransformFeedbackVarying, GLuint program, GLuint index, GLsizei bufSize, GLsizei* length, GLsizei* size, GLenum* type, GLchar* name) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetTransformFeedbackVarying, program, index, bufSize, length, size, type, name)
+DECLARE_GL_FUNCTION_HEAD(void, TransformFeedbackVaryings, GLuint program, GLsizei count, const GLchar* const* varyings, GLenum bufferMode) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TransformFeedbackVaryings, program, count, varyings, bufferMode)
+DECLARE_GL_FUNCTION_HEAD(void, GetTransformFeedbackVarying, GLuint program, GLuint index, GLsizei bufSize, GLsizei* length, GLsizei* size, GLenum* type, GLchar* name) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetTransformFeedbackVarying, program, index, bufSize, length, size, type, name)
 DECLARE_GL_FUNCTION_HEAD(void, VertexAttribIPointer, GLuint index, GLint size, GLenum type, GLsizei stride, const void* pointer) DECLARE_GL_FUNCTION_END_NO_RETURN(void, VertexAttribIPointer, index, size, type, stride, pointer)
 DECLARE_GL_FUNCTION_HEAD(void, GetVertexAttribIiv, GLuint index, GLenum pname, GLint* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetVertexAttribIiv, index, pname, params)
 DECLARE_GL_FUNCTION_HEAD(void, GetVertexAttribIuiv, GLuint index, GLenum pname, GLuint* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetVertexAttribIuiv, index, pname, params)
@@ -292,21 +292,15 @@ DECLARE_GL_FUNCTION_HEAD(void, SamplerParameterfv, GLuint sampler, GLenum pname,
 DECLARE_GL_FUNCTION_HEAD(void, GetSamplerParameteriv, GLuint sampler, GLenum pname, GLint* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetSamplerParameteriv, sampler, pname, params)
 DECLARE_GL_FUNCTION_HEAD(void, GetSamplerParameterfv, GLuint sampler, GLenum pname, GLfloat* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetSamplerParameterfv, sampler, pname, params)
 DECLARE_GL_FUNCTION_HEAD(void, VertexAttribDivisor, GLuint index, GLuint divisor) DECLARE_GL_FUNCTION_END_NO_RETURN(void, VertexAttribDivisor, index, divisor)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, BindTransformFeedback, GLenum target, GLuint id) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, BindTransformFeedback, target, id)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, DeleteTransformFeedbacks, GLsizei n, const GLuint* ids) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, DeleteTransformFeedbacks, n, ids)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, GenTransformFeedbacks, GLsizei n, GLuint* ids) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GenTransformFeedbacks, n, ids)
-// Transform feedback objects are not implemented, so no name is ever a live object. The shared
-// stub returns (type)1, telling a probing caller that every id it invents already exists; GL_FALSE
-// is both truthful and what the spec requires for a name that was never generated.
-MOBILEGL_GL_API GLboolean glIsTransformFeedback(GLuint id) {
-    MGLOG_W("Stub function: %s(...)", __FUNCTION__);
-    return GL_FALSE;
-}
-DECLARE_GL_FUNCTION_STUB_HEAD(void, PauseTransformFeedback) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, PauseTransformFeedback)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ResumeTransformFeedback) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ResumeTransformFeedback)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, GetProgramBinary, GLuint program, GLsizei bufSize, GLsizei* length, GLenum* binaryFormat, void* binary) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetProgramBinary, program, bufSize, length, binaryFormat, binary)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramBinary, GLuint program, GLenum binaryFormat, const void* binary, GLsizei length) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramBinary, program, binaryFormat, binary, length)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramParameteri, GLuint program, GLenum pname, GLint value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramParameteri, program, pname, value)
+DECLARE_GL_FUNCTION_HEAD(void, BindTransformFeedback, GLenum target, GLuint id) DECLARE_GL_FUNCTION_END_NO_RETURN(void, BindTransformFeedback, target, id)
+DECLARE_GL_FUNCTION_HEAD(void, DeleteTransformFeedbacks, GLsizei n, const GLuint* ids) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DeleteTransformFeedbacks, n, ids)
+DECLARE_GL_FUNCTION_HEAD(void, GenTransformFeedbacks, GLsizei n, GLuint* ids) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GenTransformFeedbacks, n, ids)
+DECLARE_GL_FUNCTION_HEAD(GLboolean, IsTransformFeedback, GLuint id) DECLARE_GL_FUNCTION_END(GLboolean, IsTransformFeedback, id)
+DECLARE_GL_FUNCTION_HEAD(void, PauseTransformFeedback) DECLARE_GL_FUNCTION_END_NO_RETURN(void, PauseTransformFeedback)
+DECLARE_GL_FUNCTION_HEAD(void, ResumeTransformFeedback) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ResumeTransformFeedback)
+DECLARE_GL_FUNCTION_HEAD(void, GetProgramBinary, GLuint program, GLsizei bufSize, GLsizei* length, GLenum* binaryFormat, void* binary) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetProgramBinary, program, bufSize, length, binaryFormat, binary)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramBinary, GLuint program, GLenum binaryFormat, const void* binary, GLsizei length) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramBinary, program, binaryFormat, binary, length)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramParameteri, GLuint program, GLenum pname, GLint value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramParameteri, program, pname, value)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, InvalidateFramebuffer, GLenum target, GLsizei numAttachments, const GLenum* attachments) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, InvalidateFramebuffer, target, numAttachments, attachments)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, InvalidateSubFramebuffer, GLenum target, GLsizei numAttachments, const GLenum* attachments, GLint x, GLint y, GLsizei width, GLsizei height) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, InvalidateSubFramebuffer, target, numAttachments, attachments, x, y, width, height)
 DECLARE_GL_FUNCTION_HEAD(void, TexStorage2D, GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TexStorage2D, target, levels, internalformat, width, height)
@@ -425,12 +419,12 @@ DECLARE_GL_FUNCTION_HEAD(void, DrawElementsInstancedBaseVertex, GLenum mode, GLs
 DECLARE_GL_FUNCTION_HEAD(void, FramebufferTexture, GLenum target, GLenum attachment, GLuint texture, GLint level) DECLARE_GL_FUNCTION_END_NO_RETURN(void, FramebufferTexture, target, attachment, texture, level)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, PrimitiveBoundingBox, GLfloat minX, GLfloat minY, GLfloat minZ, GLfloat minW, GLfloat maxX, GLfloat maxY, GLfloat maxZ, GLfloat maxW) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, PrimitiveBoundingBox, minX, minY, minZ, minW, maxX, maxY, maxZ, maxW)
 DECLARE_GL_FUNCTION_HEAD(GLenum, GetGraphicsResetStatus) DECLARE_GL_FUNCTION_END(GLenum, GetGraphicsResetStatus)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ReadnPixels, GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, GLsizei bufSize, void* data) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ReadnPixels, x, y, width, height, format, type, bufSize, data)
+DECLARE_GL_FUNCTION_HEAD(void, ReadnPixels, GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, GLsizei bufSize, void* data) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ReadnPixels, x, y, width, height, format, type, bufSize, data)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetnUniformfv, GLuint program, GLint location, GLsizei bufSize, GLfloat* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetnUniformfv, program, location, bufSize, params)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetnUniformiv, GLuint program, GLint location, GLsizei bufSize, GLint* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetnUniformiv, program, location, bufSize, params)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetnUniformuiv, GLuint program, GLint location, GLsizei bufSize, GLuint* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetnUniformuiv, program, location, bufSize, params)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, MinSampleShading, GLfloat value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, MinSampleShading, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, PatchParameteri, GLenum pname, GLint value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, PatchParameteri, pname, value)
+DECLARE_GL_FUNCTION_HEAD(void, PatchParameteri, GLenum pname, GLint value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, PatchParameteri, pname, value)
 DECLARE_GL_FUNCTION_HEAD(void, TexParameterIiv, GLenum target, GLenum pname, const GLint* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TexParameterIiv, target, pname, params)
 DECLARE_GL_FUNCTION_HEAD(void, TexParameterIuiv, GLenum target, GLenum pname, const GLuint* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TexParameterIuiv, target, pname, params)
 DECLARE_GL_FUNCTION_HEAD(void, GetTexParameterIiv, GLenum target, GLenum pname, GLint* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetTexParameterIiv, target, pname, params)
@@ -440,7 +434,7 @@ DECLARE_GL_FUNCTION_HEAD(void, SamplerParameterIuiv, GLuint sampler, GLenum pnam
 DECLARE_GL_FUNCTION_HEAD(void, GetSamplerParameterIiv, GLuint sampler, GLenum pname, GLint* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetSamplerParameterIiv, sampler, pname, params)
 DECLARE_GL_FUNCTION_HEAD(void, GetSamplerParameterIuiv, GLuint sampler, GLenum pname, GLuint* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetSamplerParameterIuiv, sampler, pname, params)
 DECLARE_GL_FUNCTION_HEAD(void, TexBuffer, GLenum target, GLenum internalformat, GLuint buffer) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TexBuffer, target, internalformat, buffer)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, TexBufferRange, GLenum target, GLenum internalformat, GLuint buffer, GLintptr offset, GLsizeiptr size) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, TexBufferRange, target, internalformat, buffer, offset, size)
+DECLARE_GL_FUNCTION_HEAD(void, TexBufferRange, GLenum target, GLenum internalformat, GLuint buffer, GLintptr offset, GLsizeiptr size) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TexBufferRange, target, internalformat, buffer, offset, size)
 DECLARE_GL_FUNCTION_HEAD(void, TexStorage3DMultisample, GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLboolean fixedsamplelocations) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TexStorage3DMultisample, target, samples, internalformat, width, height, depth, fixedsamplelocations)
 DECLARE_GL_FUNCTION_HEAD(void*, MapBufferRange, GLenum target, GLintptr offset, GLsizeiptr length, GLbitfield access) DECLARE_GL_FUNCTION_END(void*, MapBufferRange, target, offset, length, access)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, ClearIndex, GLfloat c) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ClearIndex, c)
@@ -915,24 +909,24 @@ DECLARE_GL_FUNCTION_STUB_HEAD(void, ColorP4ui, GLenum type, GLuint color) DECLAR
 DECLARE_GL_FUNCTION_STUB_HEAD(void, ColorP4uiv, GLenum type, const GLuint* color) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ColorP4uiv, type, color)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, SecondaryColorP3ui, GLenum type, GLuint color) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, SecondaryColorP3ui, type, color)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, SecondaryColorP3uiv, GLenum type, const GLuint* color) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, SecondaryColorP3uiv, type, color)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, Uniform1d, GLint location, GLdouble x) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, Uniform1d, location, x)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, Uniform2d, GLint location, GLdouble x, GLdouble y) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, Uniform2d, location, x, y)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, Uniform3d, GLint location, GLdouble x, GLdouble y, GLdouble z) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, Uniform3d, location, x, y, z)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, Uniform4d, GLint location, GLdouble x, GLdouble y, GLdouble z, GLdouble w) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, Uniform4d, location, x, y, z, w)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, Uniform1dv, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, Uniform1dv, location, count, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, Uniform2dv, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, Uniform2dv, location, count, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, Uniform3dv, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, Uniform3dv, location, count, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, Uniform4dv, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, Uniform4dv, location, count, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, UniformMatrix2dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, UniformMatrix2dv, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, UniformMatrix3dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, UniformMatrix3dv, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, UniformMatrix4dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, UniformMatrix4dv, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, UniformMatrix2x3dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, UniformMatrix2x3dv, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, UniformMatrix2x4dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, UniformMatrix2x4dv, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, UniformMatrix3x2dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, UniformMatrix3x2dv, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, UniformMatrix3x4dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, UniformMatrix3x4dv, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, UniformMatrix4x2dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, UniformMatrix4x2dv, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, UniformMatrix4x3dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, UniformMatrix4x3dv, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, GetUniformdv, GLuint program, GLint location, GLdouble* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetUniformdv, program, location, params)
+DECLARE_GL_FUNCTION_HEAD(void, Uniform1d, GLint location, GLdouble x) DECLARE_GL_FUNCTION_END_NO_RETURN(void, Uniform1d, location, x)
+DECLARE_GL_FUNCTION_HEAD(void, Uniform2d, GLint location, GLdouble x, GLdouble y) DECLARE_GL_FUNCTION_END_NO_RETURN(void, Uniform2d, location, x, y)
+DECLARE_GL_FUNCTION_HEAD(void, Uniform3d, GLint location, GLdouble x, GLdouble y, GLdouble z) DECLARE_GL_FUNCTION_END_NO_RETURN(void, Uniform3d, location, x, y, z)
+DECLARE_GL_FUNCTION_HEAD(void, Uniform4d, GLint location, GLdouble x, GLdouble y, GLdouble z, GLdouble w) DECLARE_GL_FUNCTION_END_NO_RETURN(void, Uniform4d, location, x, y, z, w)
+DECLARE_GL_FUNCTION_HEAD(void, Uniform1dv, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, Uniform1dv, location, count, value)
+DECLARE_GL_FUNCTION_HEAD(void, Uniform2dv, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, Uniform2dv, location, count, value)
+DECLARE_GL_FUNCTION_HEAD(void, Uniform3dv, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, Uniform3dv, location, count, value)
+DECLARE_GL_FUNCTION_HEAD(void, Uniform4dv, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, Uniform4dv, location, count, value)
+DECLARE_GL_FUNCTION_HEAD(void, UniformMatrix2dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, UniformMatrix2dv, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, UniformMatrix3dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, UniformMatrix3dv, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, UniformMatrix4dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, UniformMatrix4dv, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, UniformMatrix2x3dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, UniformMatrix2x3dv, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, UniformMatrix2x4dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, UniformMatrix2x4dv, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, UniformMatrix3x2dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, UniformMatrix3x2dv, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, UniformMatrix3x4dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, UniformMatrix3x4dv, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, UniformMatrix4x2dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, UniformMatrix4x2dv, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, UniformMatrix4x3dv, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, UniformMatrix4x3dv, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, GetUniformdv, GLuint program, GLint location, GLdouble* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetUniformdv, program, location, params)
 DECLARE_GL_FUNCTION_STUB_HEAD(GLint, GetSubroutineUniformLocation, GLuint program, GLenum shadertype, const GLchar* name) DECLARE_GL_FUNCTION_STUB_END(GLint, GetSubroutineUniformLocation, program, shadertype, name)
 DECLARE_GL_FUNCTION_STUB_HEAD(GLuint, GetSubroutineIndex, GLuint program, GLenum shadertype, const GLchar* name) DECLARE_GL_FUNCTION_STUB_END(GLuint, GetSubroutineIndex, program, shadertype, name)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetActiveSubroutineUniformiv, GLuint program, GLenum shadertype, GLuint index, GLenum pname, GLint* values) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetActiveSubroutineUniformiv, program, shadertype, index, pname, values)
@@ -942,28 +936,28 @@ DECLARE_GL_FUNCTION_STUB_HEAD(void, UniformSubroutinesuiv, GLenum shadertype, GL
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetUniformSubroutineuiv, GLenum shadertype, GLint location, GLuint* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetUniformSubroutineuiv, shadertype, location, params)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetProgramStageiv, GLuint program, GLenum shadertype, GLenum pname, GLint* values) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetProgramStageiv, program, shadertype, pname, values)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, PatchParameterfv, GLenum pname, const GLfloat* values) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, PatchParameterfv, pname, values)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, DrawTransformFeedback, GLenum mode, GLuint id) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, DrawTransformFeedback, mode, id)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, DrawTransformFeedbackStream, GLenum mode, GLuint id, GLuint stream) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, DrawTransformFeedbackStream, mode, id, stream)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, BeginQueryIndexed, GLenum target, GLuint index, GLuint id) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, BeginQueryIndexed, target, index, id)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, EndQueryIndexed, GLenum target, GLuint index) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, EndQueryIndexed, target, index)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, GetQueryIndexediv, GLenum target, GLuint index, GLenum pname, GLint* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetQueryIndexediv, target, index, pname, params)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniform1d, GLuint program, GLint location, GLdouble v0) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniform1d, program, location, v0)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniform1dv, GLuint program, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniform1dv, program, location, count, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniform2d, GLuint program, GLint location, GLdouble v0, GLdouble v1) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniform2d, program, location, v0, v1)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniform2dv, GLuint program, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniform2dv, program, location, count, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniform3d, GLuint program, GLint location, GLdouble v0, GLdouble v1, GLdouble v2) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniform3d, program, location, v0, v1, v2)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniform3dv, GLuint program, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniform3dv, program, location, count, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniform4d, GLuint program, GLint location, GLdouble v0, GLdouble v1, GLdouble v2, GLdouble v3) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniform4d, program, location, v0, v1, v2, v3)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniform4dv, GLuint program, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniform4dv, program, location, count, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniformMatrix2dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniformMatrix2dv, program, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniformMatrix3dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniformMatrix3dv, program, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniformMatrix4dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniformMatrix4dv, program, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniformMatrix2x3dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniformMatrix2x3dv, program, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniformMatrix3x2dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniformMatrix3x2dv, program, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniformMatrix2x4dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniformMatrix2x4dv, program, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniformMatrix4x2dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniformMatrix4x2dv, program, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniformMatrix3x4dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniformMatrix3x4dv, program, location, count, transpose, value)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ProgramUniformMatrix4x3dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ProgramUniformMatrix4x3dv, program, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, DrawTransformFeedback, GLenum mode, GLuint id) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DrawTransformFeedback, mode, id)
+DECLARE_GL_FUNCTION_HEAD(void, DrawTransformFeedbackStream, GLenum mode, GLuint id, GLuint stream) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DrawTransformFeedbackStream, mode, id, stream)
+DECLARE_GL_FUNCTION_HEAD(void, BeginQueryIndexed, GLenum target, GLuint index, GLuint id) DECLARE_GL_FUNCTION_END_NO_RETURN(void, BeginQueryIndexed, target, index, id)
+DECLARE_GL_FUNCTION_HEAD(void, EndQueryIndexed, GLenum target, GLuint index) DECLARE_GL_FUNCTION_END_NO_RETURN(void, EndQueryIndexed, target, index)
+DECLARE_GL_FUNCTION_HEAD(void, GetQueryIndexediv, GLenum target, GLuint index, GLenum pname, GLint* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetQueryIndexediv, target, index, pname, params)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniform1d, GLuint program, GLint location, GLdouble v0) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniform1d, program, location, v0)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniform1dv, GLuint program, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniform1dv, program, location, count, value)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniform2d, GLuint program, GLint location, GLdouble v0, GLdouble v1) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniform2d, program, location, v0, v1)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniform2dv, GLuint program, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniform2dv, program, location, count, value)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniform3d, GLuint program, GLint location, GLdouble v0, GLdouble v1, GLdouble v2) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniform3d, program, location, v0, v1, v2)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniform3dv, GLuint program, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniform3dv, program, location, count, value)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniform4d, GLuint program, GLint location, GLdouble v0, GLdouble v1, GLdouble v2, GLdouble v3) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniform4d, program, location, v0, v1, v2, v3)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniform4dv, GLuint program, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniform4dv, program, location, count, value)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniformMatrix2dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniformMatrix2dv, program, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniformMatrix3dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniformMatrix3dv, program, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniformMatrix4dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniformMatrix4dv, program, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniformMatrix2x3dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniformMatrix2x3dv, program, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniformMatrix3x2dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniformMatrix3x2dv, program, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniformMatrix2x4dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniformMatrix2x4dv, program, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniformMatrix4x2dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniformMatrix4x2dv, program, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniformMatrix3x4dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniformMatrix3x4dv, program, location, count, transpose, value)
+DECLARE_GL_FUNCTION_HEAD(void, ProgramUniformMatrix4x3dv, GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProgramUniformMatrix4x3dv, program, location, count, transpose, value)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, VertexAttribL1d, GLuint index, GLdouble x) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, VertexAttribL1d, index, x)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, VertexAttribL2d, GLuint index, GLdouble x, GLdouble y) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, VertexAttribL2d, index, x, y)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, VertexAttribL3d, GLuint index, GLdouble x, GLdouble y, GLdouble z) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, VertexAttribL3d, index, x, y, z)
@@ -988,8 +982,8 @@ DECLARE_GL_FUNCTION_HEAD(void, DrawArraysInstancedBaseInstance, GLenum mode, GLi
 DECLARE_GL_FUNCTION_HEAD(void, DrawElementsInstancedBaseInstance, GLenum mode, GLsizei count, GLenum type, const void* indices, GLsizei instancecount, GLuint baseinstance) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DrawElementsInstancedBaseInstance, mode, count, type, indices, instancecount, baseinstance)
 DECLARE_GL_FUNCTION_HEAD(void, DrawElementsInstancedBaseVertexBaseInstance, GLenum mode, GLsizei count, GLenum type, const void* indices, GLsizei instancecount, GLint basevertex, GLuint baseinstance) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DrawElementsInstancedBaseVertexBaseInstance, mode, count, type, indices, instancecount, basevertex, baseinstance)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetActiveAtomicCounterBufferiv, GLuint program, GLuint bufferIndex, GLenum pname, GLint* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetActiveAtomicCounterBufferiv, program, bufferIndex, pname, params)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, DrawTransformFeedbackInstanced, GLenum mode, GLuint id, GLsizei instancecount) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, DrawTransformFeedbackInstanced, mode, id, instancecount)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, DrawTransformFeedbackStreamInstanced, GLenum mode, GLuint id, GLuint stream, GLsizei instancecount) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, DrawTransformFeedbackStreamInstanced, mode, id, stream, instancecount)
+DECLARE_GL_FUNCTION_HEAD(void, DrawTransformFeedbackInstanced, GLenum mode, GLuint id, GLsizei instancecount) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DrawTransformFeedbackInstanced, mode, id, instancecount)
+DECLARE_GL_FUNCTION_HEAD(void, DrawTransformFeedbackStreamInstanced, GLenum mode, GLuint id, GLuint stream, GLsizei instancecount) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DrawTransformFeedbackStreamInstanced, mode, id, stream, instancecount)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, ClearBufferData, GLenum target, GLenum internalformat, GLenum format, GLenum type, const void* data) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ClearBufferData, target, internalformat, format, type, data)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, ClearBufferSubData, GLenum target, GLenum internalformat, GLintptr offset, GLsizeiptr size, GLenum format, GLenum type, const void* data) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ClearBufferSubData, target, internalformat, offset, size, format, type, data)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetInternalformati64v, GLenum target, GLenum internalformat, GLenum pname, GLsizei count, GLint64* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetInternalformati64v, target, internalformat, pname, count, params)
@@ -1002,7 +996,7 @@ DECLARE_GL_FUNCTION_HEAD(void, MultiDrawElementsIndirect, GLenum mode, GLenum ty
 DECLARE_GL_FUNCTION_HEAD(GLint, GetProgramResourceLocationIndex, GLuint program, GLenum programInterface, const GLchar* name) DECLARE_GL_FUNCTION_END(GLint, GetProgramResourceLocationIndex, program, programInterface, name)
 DECLARE_GL_FUNCTION_HEAD(void, ShaderStorageBlockBinding, GLuint program, GLuint storageBlockIndex, GLuint storageBlockBinding) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ShaderStorageBlockBinding, program, storageBlockIndex, storageBlockBinding)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, TextureView, GLuint texture, GLenum target, GLuint origtexture, GLenum internalformat, GLuint minlevel, GLuint numlevels, GLuint minlayer, GLuint numlayers) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, TextureView, texture, target, origtexture, internalformat, minlevel, numlevels, minlayer, numlayers)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, VertexAttribLFormat, GLuint attribindex, GLint size, GLenum type, GLuint relativeoffset) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, VertexAttribLFormat, attribindex, size, type, relativeoffset)
+DECLARE_GL_FUNCTION_HEAD(void, VertexAttribLFormat, GLuint attribindex, GLint size, GLenum type, GLuint relativeoffset) DECLARE_GL_FUNCTION_END_NO_RETURN(void, VertexAttribLFormat, attribindex, size, type, relativeoffset)
 DECLARE_GL_FUNCTION_HEAD(void, BufferStorage, GLenum target, GLsizeiptr size, const void* data, GLbitfield flags) DECLARE_GL_FUNCTION_END_NO_RETURN(void, BufferStorage, target, size, data, flags)
 DECLARE_GL_FUNCTION_HEAD(void, ClearTexImage, GLuint texture, GLint level, GLenum format, GLenum type, const void* data) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ClearTexImage, texture, level, format, type, data)
 DECLARE_GL_FUNCTION_HEAD(void, ClearTexSubImage, GLuint texture, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const void* data) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ClearTexSubImage, texture, level, xoffset, yoffset, zoffset, width, height, depth, format, type, data)
@@ -1013,12 +1007,12 @@ DECLARE_GL_FUNCTION_HEAD(void, BindSamplers, GLuint first, GLsizei count, const
 DECLARE_GL_FUNCTION_STUB_HEAD(void, BindImageTextures, GLuint first, GLsizei count, const GLuint* textures) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, BindImageTextures, first, count, textures)
 DECLARE_GL_FUNCTION_HEAD(void, BindVertexBuffers, GLuint first, GLsizei count, const GLuint* buffers, const GLintptr* offsets, const GLsizei* strides) DECLARE_GL_FUNCTION_END_NO_RETURN(void, BindVertexBuffers, first, count, buffers, offsets, strides)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, ClipControl, GLenum origin, GLenum depth) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ClipControl, origin, depth)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, CreateTransformFeedbacks, GLsizei n, GLuint* ids) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, CreateTransformFeedbacks, n, ids)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, TransformFeedbackBufferBase, GLuint xfb, GLuint index, GLuint buffer) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, TransformFeedbackBufferBase, xfb, index, buffer)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, TransformFeedbackBufferRange, GLuint xfb, GLuint index, GLuint buffer, GLintptr offset, GLsizeiptr size) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, TransformFeedbackBufferRange, xfb, index, buffer, offset, size)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, GetTransformFeedbackiv, GLuint xfb, GLenum pname, GLint* param) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetTransformFeedbackiv, xfb, pname, param)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, GetTransformFeedbacki_v, GLuint xfb, GLenum pname, GLuint index, GLint* param) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetTransformFeedbacki_v, xfb, pname, index, param)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, GetTransformFeedbacki64_v, GLuint xfb, GLenum pname, GLuint index, GLint64* param) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetTransformFeedbacki64_v, xfb, pname, index, param)
+DECLARE_GL_FUNCTION_HEAD(void, CreateTransformFeedbacks, GLsizei n, GLuint* ids) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CreateTransformFeedbacks, n, ids)
+DECLARE_GL_FUNCTION_HEAD(void, TransformFeedbackBufferBase, GLuint xfb, GLuint index, GLuint buffer) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TransformFeedbackBufferBase, xfb, index, buffer)
+DECLARE_GL_FUNCTION_HEAD(void, TransformFeedbackBufferRange, GLuint xfb, GLuint index, GLuint buffer, GLintptr offset, GLsizeiptr size) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TransformFeedbackBufferRange, xfb, index, buffer, offset, size)
+DECLARE_GL_FUNCTION_HEAD(void, GetTransformFeedbackiv, GLuint xfb, GLenum pname, GLint* param) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetTransformFeedbackiv, xfb, pname, param)
+DECLARE_GL_FUNCTION_HEAD(void, GetTransformFeedbacki_v, GLuint xfb, GLenum pname, GLuint index, GLint* param) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetTransformFeedbacki_v, xfb, pname, index, param)
+DECLARE_GL_FUNCTION_HEAD(void, GetTransformFeedbacki64_v, GLuint xfb, GLenum pname, GLuint index, GLint64* param) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetTransformFeedbacki64_v, xfb, pname, index, param)
 DECLARE_GL_FUNCTION_HEAD(void, CreateBuffers, GLsizei n, GLuint* buffers) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CreateBuffers, n, buffers)
 DECLARE_GL_FUNCTION_HEAD(void, NamedBufferStorage, GLuint buffer, GLsizeiptr size, const void* data, GLbitfield flags) DECLARE_GL_FUNCTION_END_NO_RETURN(void, NamedBufferStorage, buffer, size, data, flags)
 DECLARE_GL_FUNCTION_HEAD(void, NamedBufferData, GLuint buffer, GLsizeiptr size, const void* data, GLenum usage) DECLARE_GL_FUNCTION_END_NO_RETURN(void, NamedBufferData, buffer, size, data, usage)
@@ -1055,8 +1049,8 @@ DECLARE_GL_FUNCTION_HEAD(void, NamedRenderbufferStorage, GLuint renderbuffer, GL
 DECLARE_GL_FUNCTION_HEAD(void, NamedRenderbufferStorageMultisample, GLuint renderbuffer, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height) DECLARE_GL_FUNCTION_END_NO_RETURN(void, NamedRenderbufferStorageMultisample, renderbuffer, samples, internalformat, width, height)
 DECLARE_GL_FUNCTION_HEAD(void, GetNamedRenderbufferParameteriv, GLuint renderbuffer, GLenum pname, GLint* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetNamedRenderbufferParameteriv, renderbuffer, pname, params)
 DECLARE_GL_FUNCTION_HEAD(void, CreateTextures, GLenum target, GLsizei n, GLuint* textures) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CreateTextures, target, n, textures)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, TextureBuffer, GLuint texture, GLenum internalformat, GLuint buffer) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, TextureBuffer, texture, internalformat, buffer)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, TextureBufferRange, GLuint texture, GLenum internalformat, GLuint buffer, GLintptr offset, GLsizeiptr size) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, TextureBufferRange, texture, internalformat, buffer, offset, size)
+DECLARE_GL_FUNCTION_HEAD(void, TextureBuffer, GLuint texture, GLenum internalformat, GLuint buffer) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TextureBuffer, texture, internalformat, buffer)
+DECLARE_GL_FUNCTION_HEAD(void, TextureBufferRange, GLuint texture, GLenum internalformat, GLuint buffer, GLintptr offset, GLsizeiptr size) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TextureBufferRange, texture, internalformat, buffer, offset, size)
 DECLARE_GL_FUNCTION_HEAD(void, TextureStorage1D, GLuint texture, GLsizei levels, GLenum internalformat, GLsizei width) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TextureStorage1D, texture, levels, internalformat, width)
 DECLARE_GL_FUNCTION_HEAD(void, TextureStorage2D, GLuint texture, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TextureStorage2D, texture, levels, internalformat, width, height)
 DECLARE_GL_FUNCTION_HEAD(void, TextureStorage3D, GLuint texture, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TextureStorage3D, texture, levels, internalformat, width, height, depth)
@@ -1096,14 +1090,14 @@ DECLARE_GL_FUNCTION_HEAD(void, VertexArrayVertexBuffers, GLuint vaobj, GLuint fi
 DECLARE_GL_FUNCTION_HEAD(void, VertexArrayAttribBinding, GLuint vaobj, GLuint attribindex, GLuint bindingindex) DECLARE_GL_FUNCTION_END_NO_RETURN(void, VertexArrayAttribBinding, vaobj, attribindex, bindingindex)
 DECLARE_GL_FUNCTION_HEAD(void, VertexArrayAttribFormat, GLuint vaobj, GLuint attribindex, GLint size, GLenum type, GLboolean normalized, GLuint relativeoffset) DECLARE_GL_FUNCTION_END_NO_RETURN(void, VertexArrayAttribFormat, vaobj, attribindex, size, type, normalized, relativeoffset)
 DECLARE_GL_FUNCTION_HEAD(void, VertexArrayAttribIFormat, GLuint vaobj, GLuint attribindex, GLint size, GLenum type, GLuint relativeoffset) DECLARE_GL_FUNCTION_END_NO_RETURN(void, VertexArrayAttribIFormat, vaobj, attribindex, size, type, relativeoffset)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, VertexArrayAttribLFormat, GLuint vaobj, GLuint attribindex, GLint size, GLenum type, GLuint relativeoffset) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, VertexArrayAttribLFormat, vaobj, attribindex, size, type, relativeoffset)
+DECLARE_GL_FUNCTION_HEAD(void, VertexArrayAttribLFormat, GLuint vaobj, GLuint attribindex, GLint size, GLenum type, GLuint relativeoffset) DECLARE_GL_FUNCTION_END_NO_RETURN(void, VertexArrayAttribLFormat, vaobj, attribindex, size, type, relativeoffset)
 DECLARE_GL_FUNCTION_HEAD(void, VertexArrayBindingDivisor, GLuint vaobj, GLuint bindingindex, GLuint divisor) DECLARE_GL_FUNCTION_END_NO_RETURN(void, VertexArrayBindingDivisor, vaobj, bindingindex, divisor)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, GetVertexArrayiv, GLuint vaobj, GLenum pname, GLint* param) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetVertexArrayiv, vaobj, pname, param)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, GetVertexArrayIndexediv, GLuint vaobj, GLuint index, GLenum pname, GLint* param) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetVertexArrayIndexediv, vaobj, index, pname, param)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, GetVertexArrayIndexed64iv, GLuint vaobj, GLuint index, GLenum pname, GLint64* param) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetVertexArrayIndexed64iv, vaobj, index, pname, param)
+DECLARE_GL_FUNCTION_HEAD(void, GetVertexArrayiv, GLuint vaobj, GLenum pname, GLint* param) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetVertexArrayiv, vaobj, pname, param)
+DECLARE_GL_FUNCTION_HEAD(void, GetVertexArrayIndexediv, GLuint vaobj, GLuint index, GLenum pname, GLint* param) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetVertexArrayIndexediv, vaobj, index, pname, param)
+DECLARE_GL_FUNCTION_HEAD(void, GetVertexArrayIndexed64iv, GLuint vaobj, GLuint index, GLenum pname, GLint64* param) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetVertexArrayIndexed64iv, vaobj, index, pname, param)
 DECLARE_GL_FUNCTION_HEAD(void, CreateSamplers, GLsizei n, GLuint* samplers) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CreateSamplers, n, samplers)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, CreateProgramPipelines, GLsizei n, GLuint* pipelines) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, CreateProgramPipelines, n, pipelines)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, CreateQueries, GLenum target, GLsizei n, GLuint* ids) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, CreateQueries, target, n, ids)
+DECLARE_GL_FUNCTION_HEAD(void, CreateQueries, GLenum target, GLsizei n, GLuint* ids) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CreateQueries, target, n, ids)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetQueryBufferObjecti64v, GLuint id, GLuint buffer, GLenum pname, GLintptr offset) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetQueryBufferObjecti64v, id, buffer, pname, offset)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetQueryBufferObjectiv, GLuint id, GLuint buffer, GLenum pname, GLintptr offset) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetQueryBufferObjectiv, id, buffer, pname, offset)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetQueryBufferObjectui64v, GLuint id, GLuint buffer, GLenum pname, GLintptr offset) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetQueryBufferObjectui64v, id, buffer, pname, offset)

MobileGL/MG_Impl/GLImpl/Framebuffer/GL_Framebuffer.cpp

@@ -22,9 +22,21 @@
 
 namespace MobileGL::MG_Impl::GLImpl {
     namespace {
-        Bool IsActiveBackendDirectVulkan() {
+        // GL only requires support for framebuffers whose depth and stencil attachments
+        // are the same image; anything else may be reported GL_FRAMEBUFFER_UNSUPPORTED.
+        // DirectVulkan cannot form two separate attachments at all, and the real ES
+        // drivers behind DirectGLES answer UNSUPPORTED for it too - so saying COMPLETE
+        // and then rendering into a framebuffer the driver refuses produced silently
+        // empty results (KHR-GL3x.packed_depth_stencil.verify_mixed_attachments).
+        Bool ActiveBackendRejectsDistinctDepthStencil() {
             auto* activeBackend = MG_Backend::pActiveBackendObject.get();
-            return activeBackend != nullptr && activeBackend->GetBackendType() == BackendType::DirectVulkan;
+            if (activeBackend == nullptr) {
+                return false;
+            }
+            if (activeBackend->GetBackendType() == BackendType::DirectVulkan) {
+                return true;
+            }
+            return !activeBackend->GetDynamicParameters().SupportsDistinctDepthStencilAttachments;
         }
 
         Bool HasDistinctCompleteDepthStencilTextureAttachments(
@@ -45,10 +57,32 @@ namespace MobileGL::MG_Impl::GLImpl {
                    depthAttachment.GetTextureLevel() != stencilAttachment.GetTextureLevel();
         }
 
-        Bool IsUnsupportedFramebufferForDirectVulkan(
+        // Mirrors the renderer-side gate: distinct depth/stencil renderbuffers (or a
+        // renderbuffer paired with a texture) cannot form one Vulkan depth-stencil
+        // attachment, and GL permits reporting such framebuffers as UNSUPPORTED.
+        Bool HasDistinctCompleteDepthStencilRenderbufferAttachments(
+            const MG_State::GLState::FramebufferObject& framebufferObject) {
+            if (framebufferObject.GetExternalIndex() == 0) {
+                return false;
+            }
+
+            const auto& depthAttachment = framebufferObject.GetAttachment(FramebufferAttachmentType::Depth);
+            const auto& stencilAttachment = framebufferObject.GetAttachment(FramebufferAttachmentType::Stencil);
+            if (!depthAttachment.IsComplete() || !stencilAttachment.IsComplete()) {
+                return false;
+            }
+            if (depthAttachment.IsRenderbuffer() && stencilAttachment.IsRenderbuffer()) {
+                return depthAttachment.GetRenderbuffer().get() != stencilAttachment.GetRenderbuffer().get();
+            }
+            return (depthAttachment.IsRenderbuffer() || stencilAttachment.IsRenderbuffer()) &&
+                   (depthAttachment.IsTexture() || stencilAttachment.IsTexture());
+        }
+
+        Bool HasUnsupportedDistinctDepthStencilAttachments(
             const MG_State::GLState::FramebufferObject& framebufferObject) {
             // TODO: Keep this in sync with DirectVulkan renderbuffer support as color renderbuffer rendering lands.
-            return HasDistinctCompleteDepthStencilTextureAttachments(framebufferObject);
+            return HasDistinctCompleteDepthStencilTextureAttachments(framebufferObject) ||
+                   HasDistinctCompleteDepthStencilRenderbufferAttachments(framebufferObject);
         }
 
         Bool HasDefinedAttachment(const MG_State::GLState::FramebufferObject& framebufferObject) {
@@ -67,7 +101,14 @@ namespace MobileGL::MG_Impl::GLImpl {
         // shared-exponent, SNORM, three-channel norm16/float32/sRGB and three-channel integer formats.
         // Desktop GL treats those as texture-only too (not in the GL 3.3 required-renderable list), so
         // reporting GL_FRAMEBUFFER_UNSUPPORTED for them is legal.
-        Bool IsColorInternalFormatRenderable(TextureInternalFormat format) {
+        //
+        // `capabilityTargetIndex` is the row of the cache the attachment actually lives in;
+        // kFormatCapabilityTargetCount asks about the format in general. Asking per target matters
+        // because a capability recorded for one of them says nothing about the others: DirectGLES
+        // widens three-channel formats to four channels to keep them renderable as *multisample*
+        // storage, and a format that survives only through that substitution is still texture-only
+        // on every ordinary target.
+        Bool IsColorInternalFormatRenderable(TextureInternalFormat format, SizeT capabilityTargetIndex) {
             const SizeT formatIndex = static_cast<SizeT>(format);
             if (MG_Backend::pActiveBackendObject && formatIndex < MG_Backend::kFormatCapabilityFormatCount) {
                 const auto& cache = MG_Backend::pActiveBackendObject->GetFormatCapabilities();
@@ -79,8 +120,11 @@ namespace MobileGL::MG_Impl::GLImpl {
                                                                      MG_Backend::FormatCapability::Creatable);
                 }
                 if (cachePopulated) {
-                    for (SizeT targetIndex = 0; targetIndex < MG_Backend::kFormatCapabilityTargetCount;
-                         ++targetIndex) {
+                    const Bool singleTarget = capabilityTargetIndex < MG_Backend::kFormatCapabilityTargetCount;
+                    const SizeT firstTarget = singleTarget ? capabilityTargetIndex : 0;
+                    const SizeT lastTarget =
+                        singleTarget ? capabilityTargetIndex + 1 : MG_Backend::kFormatCapabilityTargetCount;
+                    for (SizeT targetIndex = firstTarget; targetIndex < lastTarget; ++targetIndex) {
                         if (MG_Backend::HasFormatCapability(cache.FullCaps[targetIndex][formatIndex],
                                                             MG_Backend::FormatCapability::FramebufferRenderable) ||
                             MG_Backend::HasFormatCapability(cache.CaveatCaps[targetIndex][formatIndex],
@@ -129,12 +173,17 @@ namespace MobileGL::MG_Impl::GLImpl {
                 const auto& attachment = attachments[i];
                 if (!attachment.IsValid()) continue;
                 TextureInternalFormat format = TextureInternalFormat::Unknown;
+                SizeT capabilityTargetIndex = MG_Backend::kFormatCapabilityTargetCount;
                 if (attachment.IsTexture() && attachment.GetTexture()) {
                     format = attachment.GetTexture()->GetFormat();
+                    capabilityTargetIndex =
+                        MG_Backend::GetFormatCapabilityTargetIndex(attachment.GetTexture()->GetTarget());
                 } else if (attachment.IsRenderbuffer() && attachment.GetRenderbuffer()) {
                     format = attachment.GetRenderbuffer()->GetInternalFormat();
+                    capabilityTargetIndex = MG_Backend::GetRenderbufferFormatCapabilityTargetIndex();
                 }
-                if (format != TextureInternalFormat::Unknown && !IsColorInternalFormatRenderable(format)) {
+                if (format != TextureInternalFormat::Unknown &&
+                    !IsColorInternalFormatRenderable(format, capabilityTargetIndex)) {
                     return true;
                 }
             }
@@ -147,6 +196,148 @@ namespace MobileGL::MG_Impl::GLImpl {
                 MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName, detail));
         }
 
+        GLint ClassifyAttachmentComponentType(TextureInternalFormat internalFormat,
+                                              FramebufferAttachmentType attachmentType) {
+            // A stencil value is an unsigned integer index regardless of the depth half
+            // of a packed format.
+            if (attachmentType == FramebufferAttachmentType::Stencil) return GL_UNSIGNED_INT;
+            switch (internalFormat) {
+            case TextureInternalFormat::R16F:
+            case TextureInternalFormat::RG16F:
+            case TextureInternalFormat::RGB16F:
+            case TextureInternalFormat::RGBA16F:
+            case TextureInternalFormat::R32F:
+            case TextureInternalFormat::RG32F:
+            case TextureInternalFormat::RGB32F:
+            case TextureInternalFormat::RGBA32F:
+            case TextureInternalFormat::R11FG11FB10F:
+            case TextureInternalFormat::RGB9E5:
+            case TextureInternalFormat::DepthComponent32F:
+            case TextureInternalFormat::Depth32FStencil8:
+                return GL_FLOAT;
+            case TextureInternalFormat::R8I:
+            case TextureInternalFormat::R16I:
+            case TextureInternalFormat::R32I:
+            case TextureInternalFormat::RG8I:
+            case TextureInternalFormat::RG16I:
+            case TextureInternalFormat::RG32I:
+            case TextureInternalFormat::RGB8I:
+            case TextureInternalFormat::RGB16I:
+            case TextureInternalFormat::RGB32I:
+            case TextureInternalFormat::RGBA8I:
+            case TextureInternalFormat::RGBA16I:
+            case TextureInternalFormat::RGBA32I:
+                return GL_INT;
+            case TextureInternalFormat::R8UI:
+            case TextureInternalFormat::R16UI:
+            case TextureInternalFormat::R32UI:
+            case TextureInternalFormat::RG8UI:
+            case TextureInternalFormat::RG16UI:
+            case TextureInternalFormat::RG32UI:
+            case TextureInternalFormat::RGB8UI:
+            case TextureInternalFormat::RGB16UI:
+            case TextureInternalFormat::RGB32UI:
+            case TextureInternalFormat::RGBA8UI:
+            case TextureInternalFormat::RGBA16UI:
+            case TextureInternalFormat::RGBA32UI:
+            case TextureInternalFormat::RGB10A2UI:
+                return GL_UNSIGNED_INT;
+            case TextureInternalFormat::R8Snorm:
+            case TextureInternalFormat::R16Snorm:
+            case TextureInternalFormat::RG8Snorm:
+            case TextureInternalFormat::RG16Snorm:
+            case TextureInternalFormat::RGB8Snorm:
+            case TextureInternalFormat::RGB16Snorm:
+            case TextureInternalFormat::RGBA8Snorm:
+            case TextureInternalFormat::RGBA16Snorm:
+                return GL_SIGNED_NORMALIZED;
+            case TextureInternalFormat::Unknown:
+                return GL_NONE;
+            default:
+                return GL_UNSIGNED_NORMALIZED;
+            }
+        }
+
+        // Handles the format-derived pnames shared by GetFramebufferAttachmentParameteriv
+        // and its DSA variant. Returns true when pname was one of them.
+        Bool TryAnswerAttachmentFormatQuery(const MG_State::GLState::FramebufferAttachmentObject* attachmentObject,
+                                            FramebufferAttachmentType attachmentType, Bool depthStencilAlias,
+                                            GLenum pname, GLint* params, const char* caller) {
+            switch (pname) {
+            case GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE:
+            case GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE:
+            case GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE:
+            case GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE:
+            case GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE:
+            case GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE:
+            case GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE:
+            case GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING:
+                break;
+            default:
+                return false;
+            }
+
+            if (pname == GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE && depthStencilAlias) {
+                // The depth and stencil components have different types, so the combined
+                // attachment name has no single answer.
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>(
+                        "MG_Impl/GLImpl", caller,
+                        "GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE cannot be queried on "
+                        "GL_DEPTH_STENCIL_ATTACHMENT."));
+                return true;
+            }
+
+            if (attachmentObject == nullptr || attachmentObject->IsEmpty() || !attachmentObject->IsValid()) {
+                // With OBJECT_TYPE == GL_NONE only OBJECT_TYPE and OBJECT_NAME may be queried.
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller,
+                                                 "No image is attached to the queried attachment point."));
+                return true;
+            }
+
+            TextureInternalFormat internalFormat = TextureInternalFormat::Unknown;
+            if (attachmentObject->IsTexture() && attachmentObject->GetTexture()) {
+                internalFormat = attachmentObject->GetTexture()->GetFormat();
+            } else if (attachmentObject->IsRenderbuffer() && attachmentObject->GetRenderbuffer()) {
+                internalFormat = attachmentObject->GetRenderbuffer()->GetInternalFormat();
+            }
+            const auto sizes = MG_Util::GetComponentSizesForInternalFormat(internalFormat);
+
+            switch (pname) {
+            case GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE:
+                *params = sizes.Red;
+                break;
+            case GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE:
+                *params = sizes.Green;
+                break;
+            case GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE:
+                *params = sizes.Blue;
+                break;
+            case GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE:
+                *params = sizes.Alpha;
+                break;
+            case GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE:
+                *params = sizes.Depth;
+                break;
+            case GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE:
+                *params = sizes.Stencil;
+                break;
+            case GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING:
+                *params = (internalFormat == TextureInternalFormat::SRGB8 ||
+                           internalFormat == TextureInternalFormat::SRGB8Alpha8)
+                    ? GL_SRGB
+                    : GL_LINEAR;
+                break;
+            case GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE:
+                *params = ClassifyAttachmentComponentType(internalFormat, attachmentType);
+                break;
+            }
+            return true;
+        }
+
         Bool ResolveRepresentableFramebufferTextureUploadTarget(const MG_State::GLState::ITextureObject& textureObject,
                                                                 TextureUploadTarget& outUploadTarget,
                                                                 Bool& outLayered) {
@@ -412,6 +603,27 @@ namespace MobileGL::MG_Impl::GLImpl {
         FramebufferTarget framebufferTarget = MG_Util::ConvertGLEnumToFramebufferTarget(target);
         if (!FramebufferImpl::ValidateFramebufferTarget(framebufferTarget)) return;
 
+        // Default-framebuffer attachment names (GL_DEPTH, GL_STENCIL, GL_FRONT/GL_BACK
+        // variants) alias onto the equivalent attachment points.
+        switch (attachment) {
+        case GL_DEPTH:
+            attachment = GL_DEPTH_ATTACHMENT;
+            break;
+        case GL_STENCIL:
+            attachment = GL_STENCIL_ATTACHMENT;
+            break;
+        case GL_FRONT:
+        case GL_FRONT_LEFT:
+        case GL_FRONT_RIGHT:
+        case GL_BACK:
+        case GL_BACK_LEFT:
+        case GL_BACK_RIGHT:
+            attachment = GL_COLOR_ATTACHMENT0;
+            break;
+        default:
+            break;
+        }
+
         const Bool depthStencilAlias = attachment == GL_DEPTH_STENCIL_ATTACHMENT;
         FramebufferAttachmentType attachmentType = depthStencilAlias
             ? FramebufferAttachmentType::Depth
@@ -428,20 +640,40 @@ namespace MobileGL::MG_Impl::GLImpl {
             return;
         }
 
+        Bool depthStencilMismatch = false;
         const auto* attachmentObject = [&]() -> const MG_State::GLState::FramebufferAttachmentObject* {
             if (!depthStencilAlias) {
                 return &framebufferObject->GetAttachment(attachmentType);
             }
 
             const auto& depthAttachment = framebufferObject->GetAttachment(FramebufferAttachmentType::Depth);
-            if (depthAttachment.IsValid() && !depthAttachment.IsEmpty()) return &depthAttachment;
-
             const auto& stencilAttachment = framebufferObject->GetAttachment(FramebufferAttachmentType::Stencil);
-            if (stencilAttachment.IsValid() && !stencilAttachment.IsEmpty()) return &stencilAttachment;
-
+            const Bool depthLive = depthAttachment.IsValid() && !depthAttachment.IsEmpty();
+            const Bool stencilLive = stencilAttachment.IsValid() && !stencilAttachment.IsEmpty();
+            if (depthLive && stencilLive) {
+                const Bool sameObject = depthAttachment.IsTexture() == stencilAttachment.IsTexture() &&
+                    (!depthAttachment.IsTexture() || depthAttachment.GetTexture() == stencilAttachment.GetTexture()) &&
+                    (!depthAttachment.IsRenderbuffer() ||
+                     depthAttachment.GetRenderbuffer() == stencilAttachment.GetRenderbuffer());
+                depthStencilMismatch = !sameObject;
+            } else {
+                // GL_DEPTH_STENCIL_ATTACHMENT means "both halves"; a lone half does not answer it.
+                depthStencilMismatch = depthLive != stencilLive;
+            }
+            if (depthLive) return &depthAttachment;
+            if (stencilLive) return &stencilAttachment;
             return nullptr;
         }();
 
+        if (depthStencilMismatch) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", "GetFramebufferAttachmentParameteriv_State",
+                                             "GL_DEPTH_STENCIL_ATTACHMENT query with different depth and stencil "
+                                             "attachment images."));
+            return;
+        }
+
         switch (pname) {
         case GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE:
             if (attachmentObject == nullptr || attachmentObject->IsEmpty() || !attachmentObject->IsValid()) {
@@ -505,6 +737,10 @@ namespace MobileGL::MG_Impl::GLImpl {
                 : GL_FALSE;
             break;
         default:
+            if (TryAnswerAttachmentFormatQuery(attachmentObject, attachmentType, depthStencilAlias, pname, params,
+                                               "GetFramebufferAttachmentParameteriv_State")) {
+                return;
+            }
             MG_State::pGLContext->RecordError(
                 ErrorCode::InvalidEnum,
                 MakeUnique<GenericErrorInfo>(
@@ -907,6 +1143,7 @@ namespace MobileGL::MG_Impl::GLImpl {
 
         const FramebufferAttachmentType attachmentType = MG_Util::ConvertGLEnumToFramebufferAttachmentType(attachment);
         if (!FramebufferImpl::ValidateFramebufferAttachmentType(attachmentType)) return;
+        if (!FramebufferImpl::ValidateColorAttachmentInRange(attachmentType, functionName)) return;
         if (!TextureImpl::ValidateTextureName(texture, true)) return;
 
         if (texture == 0) {
@@ -1003,6 +1240,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         FramebufferTarget framebufferTarget = MG_Util::ConvertGLEnumToFramebufferTarget(target);
         RenderbufferTarget rbTarget = MG_Util::ConvertGLEnumToRenderbufferTarget(renderbuffertarget);
         if (!FramebufferImpl::ValidateFramebufferAttachmentType(attachmentType)) return;
+        if (!FramebufferImpl::ValidateColorAttachmentInRange(attachmentType, "FramebufferRenderbuffer_State")) return;
         if (!FramebufferImpl::ValidateFramebufferTarget(framebufferTarget)) return;
         if (!FramebufferImpl::ValidateRenderbufferTarget(rbTarget)) return;
         auto& bindingSlot = MG_State::pGLContext->GetFramebufferBindingSlot(framebufferTarget);
@@ -1047,6 +1285,8 @@ namespace MobileGL::MG_Impl::GLImpl {
         FramebufferAttachmentType attachmentType = MG_Util::ConvertGLEnumToFramebufferAttachmentType(attachment);
         RenderbufferTarget rbTarget = MG_Util::ConvertGLEnumToRenderbufferTarget(renderbuffertarget);
         if (!FramebufferImpl::ValidateFramebufferAttachmentType(attachmentType)) return;
+        if (!FramebufferImpl::ValidateColorAttachmentInRange(attachmentType, "NamedFramebufferRenderbuffer_State"))
+            return;
         if (!FramebufferImpl::ValidateRenderbufferTarget(rbTarget)) return;
 
         if (renderbuffer == 0) {
@@ -1422,8 +1662,8 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (HasNonRenderableColorAttachment(*framebufferObject)) {
             return GL_FRAMEBUFFER_UNSUPPORTED;
         }
-        if (IsActiveBackendDirectVulkan() &&
-            IsUnsupportedFramebufferForDirectVulkan(*framebufferObject)) {
+        if (ActiveBackendRejectsDistinctDepthStencil() &&
+            HasUnsupportedDistinctDepthStencilAttachments(*framebufferObject)) {
             return GL_FRAMEBUFFER_UNSUPPORTED;
         }
         return GL_FRAMEBUFFER_COMPLETE;
@@ -1450,8 +1690,8 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (HasNonRenderableColorAttachment(*framebufferObject)) {
             return GL_FRAMEBUFFER_UNSUPPORTED;
         }
-        if (IsActiveBackendDirectVulkan() &&
-            IsUnsupportedFramebufferForDirectVulkan(*framebufferObject)) {
+        if (ActiveBackendRejectsDistinctDepthStencil() &&
+            HasUnsupportedDistinctDepthStencilAttachments(*framebufferObject)) {
             return GL_FRAMEBUFFER_UNSUPPORTED;
         }
         return GL_FRAMEBUFFER_COMPLETE;
@@ -1468,20 +1708,40 @@ namespace MobileGL::MG_Impl::GLImpl {
             : MG_Util::ConvertGLEnumToFramebufferAttachmentType(attachment);
         if (!FramebufferImpl::ValidateFramebufferAttachmentType(attachmentType)) return;
 
+        Bool depthStencilMismatch = false;
         const auto* attachmentObject = [&]() -> const MG_State::GLState::FramebufferAttachmentObject* {
             if (!depthStencilAlias) {
                 return &framebufferObject->GetAttachment(attachmentType);
             }
 
             const auto& depthAttachment = framebufferObject->GetAttachment(FramebufferAttachmentType::Depth);
-            if (depthAttachment.IsValid() && !depthAttachment.IsEmpty()) return &depthAttachment;
-
             const auto& stencilAttachment = framebufferObject->GetAttachment(FramebufferAttachmentType::Stencil);
-            if (stencilAttachment.IsValid() && !stencilAttachment.IsEmpty()) return &stencilAttachment;
-
+            const Bool depthLive = depthAttachment.IsValid() && !depthAttachment.IsEmpty();
+            const Bool stencilLive = stencilAttachment.IsValid() && !stencilAttachment.IsEmpty();
+            if (depthLive && stencilLive) {
+                const Bool sameObject = depthAttachment.IsTexture() == stencilAttachment.IsTexture() &&
+                    (!depthAttachment.IsTexture() || depthAttachment.GetTexture() == stencilAttachment.GetTexture()) &&
+                    (!depthAttachment.IsRenderbuffer() ||
+                     depthAttachment.GetRenderbuffer() == stencilAttachment.GetRenderbuffer());
+                depthStencilMismatch = !sameObject;
+            } else {
+                // GL_DEPTH_STENCIL_ATTACHMENT means "both halves"; a lone half does not answer it.
+                depthStencilMismatch = depthLive != stencilLive;
+            }
+            if (depthLive) return &depthAttachment;
+            if (stencilLive) return &stencilAttachment;
             return nullptr;
         }();
 
+        if (depthStencilMismatch) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller,
+                                             "GL_DEPTH_STENCIL_ATTACHMENT query with different depth and stencil "
+                                             "attachment images."));
+            return;
+        }
+
         switch (pname) {
         case GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE:
             if (attachmentObject == nullptr || attachmentObject->IsEmpty() || !attachmentObject->IsValid()) {
@@ -1527,6 +1787,10 @@ namespace MobileGL::MG_Impl::GLImpl {
                 : GL_FALSE;
             break;
         default:
+            if (TryAnswerAttachmentFormatQuery(attachmentObject, attachmentType, depthStencilAlias, pname, params,
+                                               caller)) {
+                return;
+            }
             MG_State::pGLContext->RecordError(
                 ErrorCode::InvalidEnum,
                 MakeUnique<GenericErrorInfo>(
@@ -1939,6 +2203,56 @@ namespace MobileGL::MG_Impl::GLImpl {
         ReadPixels_Backend(x, y, width, height, format, type, pixels);
     }
 
+    // Bytes glReadPixels would write for this rectangle under the current GL_PACK_* state
+    // (GL 4.6 core 18.2.8): rows are padded to GL_PACK_ALIGNMENT and laid out GL_PACK_ROW_LENGTH
+    // wide, and the skip parameters offset the first texel. The last row is not padded - nothing
+    // follows it to align - which is what makes a tightly-sized destination legal.
+    static SizeT ComputePackedReadSizeInBytes(GLsizei width, GLsizei height, GLenum format, GLenum type) {
+        const SizeT bytesPerPixel =
+            MG_Util::GetInputBytesPerPixel(MG_Util::ConvertGLEnumToTextureInputFormat(format),
+                                           MG_Util::ConvertGLEnumToTexturePixelDataType(type));
+        if (bytesPerPixel == 0 || width <= 0 || height <= 0) return 0;
+
+        const auto packParam = [](PixelStoreParam param) {
+            return static_cast<SizeT>(std::max(0, MG_State::pGLContext->GetPixelStoreParam(param)));
+        };
+        const SizeT rowLengthInPixels =
+            packParam(PixelStoreParam::PackRowLength) != 0
+                ? packParam(PixelStoreParam::PackRowLength)
+                : static_cast<SizeT>(width);
+        const SizeT alignment = std::max<SizeT>(1, packParam(PixelStoreParam::PackAlignment));
+
+        const SizeT unalignedRowBytes = rowLengthInPixels * bytesPerPixel;
+        const SizeT paddedRowBytes = ((unalignedRowBytes + alignment - 1) / alignment) * alignment;
+        const SizeT skipBytes = packParam(PixelStoreParam::PackSkipRows) * paddedRowBytes +
+                                packParam(PixelStoreParam::PackSkipPixels) * bytesPerPixel;
+
+        return skipBytes + paddedRowBytes * (static_cast<SizeT>(height) - 1) +
+               static_cast<SizeT>(width) * bytesPerPixel;
+    }
+
+    // glReadnPixels is glReadPixels with a bound on how much it may write (GL 4.6 core 18.2.8,
+    // originally GL_ARB_robustness). It is identical in every other respect, so it validates and
+    // reads through exactly the same path once the destination is known to be big enough.
+    void ReadnPixels(GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, GLsizei bufSize,
+                     void* data) {
+        if (bufSize < 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "bufSize must be non-negative."));
+            return;
+        }
+        if (!ReadPixels_State(x, y, width, height, format, type, data)) return;
+        if (ComputePackedReadSizeInBytes(width, height, format, type) > static_cast<SizeT>(bufSize)) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "the data required for this read does not fit in bufSize."));
+            return;
+        }
+        ReadPixels_Backend(x, y, width, height, format, type, data);
+    }
+
     void ClearBufferfi(GLenum buffer, GLint drawbuffer, GLfloat depth, GLint stencil) {
         if (!ValidateClearBufferfi_State(buffer, drawbuffer)) return;
         ClearBufferfi_Backend(buffer, drawbuffer, depth, stencil);

MobileGL/MG_Impl/GLImpl/Framebuffer/GL_Framebuffer.h

@@ -14,6 +14,8 @@
 namespace MobileGL::MG_Impl::GLImpl {
     /* @INSERTION_POINT:FUNCTION_DECLARATION@ */
     void ReadPixels(GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, void* pixels);
+    void ReadnPixels(GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, GLsizei bufSize,
+                     void* data);
     void ClearBufferfi(GLenum buffer, GLint drawbuffer, GLfloat depth, GLint stencil);
     void ClearBufferfv(GLenum buffer, GLint drawbuffer, const GLfloat* value);
     void ClearBufferuiv(GLenum buffer, GLint drawbuffer, const GLuint* value);

MobileGL/MG_Impl/GLImpl/Framebuffer/Validators.cpp

@@ -7,6 +7,7 @@
 // End of Source File Header
 
 #include "Validators.h"
+#include <MG_Backend/BackendObjects.h>
 #include <MG_State/GLState/Core.h>
 #include <MG_State/GLState/ErrorState/Error.h>
 #include <MG_Util/Converters/GLToStr/GLEnumConverter.h>
@@ -60,6 +61,26 @@ namespace MobileGL::MG_Impl::GLImpl::FramebufferImpl {
         return true;
     }
 
+    Bool ValidateColorAttachmentInRange(FramebufferAttachmentType attachment, const char* caller) {
+        const auto first = static_cast<SizeT>(FramebufferAttachmentType::Color0);
+        const auto index = static_cast<SizeT>(attachment);
+        if (index < first) return true;
+        const auto colorIndex = index - first;
+        const auto limit = static_cast<SizeT>(
+            MG_Backend::pActiveBackendObject ? MG_Backend::pActiveBackendObject->GetDynamicParameters()
+                                                   .MaxColorAttachments
+                                             : static_cast<Int>(MG_State::GLState::FramebufferObject::MAX_DRAW_BUFFERS));
+        if (colorIndex >= limit) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>(
+                    "MG_Impl/GLImpl/FramebufferImpl", caller,
+                    std::format("Colour attachment {} is beyond GL_MAX_COLOR_ATTACHMENTS ({}).", colorIndex, limit)));
+            return false;
+        }
+        return true;
+    }
+
     Bool ValidateRenderbufferTarget(RenderbufferTarget target) {
         if (target == RenderbufferTarget::Unknown) {
             using namespace MG_Util;

MobileGL/MG_Impl/GLImpl/Framebuffer/Validators.h

@@ -14,6 +14,10 @@ namespace MobileGL::MG_Impl::GLImpl::FramebufferImpl {
     Bool ValidateFramebufferTarget(FramebufferTarget target);
     Bool ValidateFramebufferName(Uint index, Bool allowZero = true);
     Bool ValidateFramebufferAttachmentType(FramebufferAttachmentType attachment);
+    // GL_COLOR_ATTACHMENTn is a token per n up to 31, but only the first GL_MAX_COLOR_ATTACHMENTS of
+    // them name an attachment point of a framebuffer object; the rest are INVALID_OPERATION for the
+    // attaching entry points (GL 4.6 core 9.2.7). Non-colour attachments pass through unchanged.
+    Bool ValidateColorAttachmentInRange(FramebufferAttachmentType attachment, const char* caller);
     Bool ValidateRenderbufferTarget(RenderbufferTarget target);
     Bool ValidateRenderbufferName(Uint index, Bool allowZero = true);
 } // namespace MobileGL::MG_Impl::GLImpl::FramebufferImpl

MobileGL/MG_Impl/GLImpl/Getter/GL_Getter.cpp

@@ -213,8 +213,13 @@ namespace MobileGL::MG_Impl::GLImpl {
 
             GLint maxSamples = 0;
             for (const auto& attachment : drawFbo->GetAllAttachmentObjects()) {
-                if (!attachment.IsRenderbuffer() || !attachment.GetRenderbuffer()) continue;
-                maxSamples = std::max(maxSamples, static_cast<GLint>(attachment.GetRenderbuffer()->GetSamples()));
+                if (attachment.IsRenderbuffer() && attachment.GetRenderbuffer()) {
+                    maxSamples = std::max(maxSamples, static_cast<GLint>(attachment.GetRenderbuffer()->GetSamples()));
+                } else if (attachment.IsTexture() && attachment.GetTexture()) {
+                    // Multisample texture attachments count too (GL_SAMPLE_BUFFERS must
+                    // report 1 for any multisampled draw framebuffer).
+                    maxSamples = std::max(maxSamples, static_cast<GLint>(attachment.GetTexture()->GetSamples()));
+                }
             }
             return maxSamples;
         }
@@ -465,6 +470,14 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_STENCIL_TEST:
             *params = MG_State::pGLContext->IsCapabilityEnabled(CapabilityInput::StencilTest) ? GL_TRUE : GL_FALSE;
             return;
+        case GL_MIN_FRAGMENT_INTERPOLATION_OFFSET:
+        case GL_MAX_FRAGMENT_INTERPOLATION_OFFSET:
+        case GL_FRAGMENT_INTERPOLATION_OFFSET_BITS: {
+            GLfloat value = 0.0f;
+            GetFloatv(pname, &value);
+            *params = value != 0.0f ? GL_TRUE : GL_FALSE;
+            return;
+        }
         default:
             break;
         }
@@ -520,6 +533,19 @@ namespace MobileGL::MG_Impl::GLImpl {
             params[1] = dynamicParameters.ViewportBoundsRangeMax;
             return;
         }
+        case GL_MIN_FRAGMENT_INTERPOLATION_OFFSET:
+        case GL_MAX_FRAGMENT_INTERPOLATION_OFFSET:
+        case GL_FRAGMENT_INTERPOLATION_OFFSET_BITS: {
+            const auto& dynamicParameters = MG_Backend::pActiveBackendObject->GetDynamicParameters();
+            if (pname == GL_MIN_FRAGMENT_INTERPOLATION_OFFSET) {
+                params[0] = dynamicParameters.MinFragmentInterpolationOffset;
+            } else if (pname == GL_MAX_FRAGMENT_INTERPOLATION_OFFSET) {
+                params[0] = dynamicParameters.MaxFragmentInterpolationOffset;
+            } else {
+                params[0] = static_cast<GLfloat>(dynamicParameters.FragmentInterpolationOffsetBits);
+            }
+            return;
+        }
         case GL_DEPTH_CLEAR_VALUE:
             params[0] = MG_State::pGLContext->GetClearDepth();
             return;
@@ -645,6 +671,40 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
 
         switch (target) {
+        // The vertex buffer binding points of the vertex array object that is bound. Indexed by
+        // binding point, not by attribute (GL 4.6 core 10.3.1).
+        case GL_VERTEX_BINDING_BUFFER:
+        case GL_VERTEX_BINDING_DIVISOR:
+        case GL_VERTEX_BINDING_OFFSET:
+        case GL_VERTEX_BINDING_STRIDE: {
+            if (index >= VertexArrayImpl::GetMaxVertexAttribBindings()) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidValue,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                                 "Vertex buffer binding index is out of range."));
+                return;
+            }
+            const auto& vao = MG_State::pGLContext->GetBoundVertexArray();
+            if (!vao) {
+                *data = 0;
+                return;
+            }
+            const auto& binding = vao->GetBindingPoint(index);
+            switch (target) {
+            case GL_VERTEX_BINDING_BUFFER:
+                *data = binding.Buffer ? static_cast<GLint>(binding.Buffer->GetExternalIndex()) : 0;
+                return;
+            case GL_VERTEX_BINDING_DIVISOR:
+                *data = static_cast<GLint>(binding.Divisor);
+                return;
+            case GL_VERTEX_BINDING_OFFSET:
+                *data = static_cast<GLint>(binding.Offset);
+                return;
+            default:
+                *data = static_cast<GLint>(binding.Stride);
+                return;
+            }
+        }
         case GL_IMAGE_BINDING_NAME:
         case GL_IMAGE_BINDING_LEVEL:
         case GL_IMAGE_BINDING_LAYERED:
@@ -1004,6 +1064,11 @@ namespace MobileGL::MG_Impl::GLImpl {
             *params = obj ? static_cast<GLint>(obj->GetExternalIndex()) : 0;
             return;
         }
+        case GL_DRAW_INDIRECT_BUFFER_BINDING: {
+            auto& obj = MG_State::pGLContext->GetBufferBindingSlot(BufferTarget::DrawIndirect).GetBoundObject();
+            *params = obj ? static_cast<GLint>(obj->GetExternalIndex()) : 0;
+            return;
+        }
         case GL_MAX_DEBUG_GROUP_STACK_DEPTH:
             *params = 0; // debug-group entrypoints are stubbed
             return;
@@ -1638,7 +1703,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             *params = static_cast<GLint>(MG_State::pGLContext->GetHint(pname));
             return;
         case GL_TEXTURE_BUFFER_OFFSET_ALIGNMENT:
-            *params = 0; // texture-buffer range entrypoints are stubbed
+            *params = MG_Backend::pActiveBackendObject->GetDynamicParameters().TextureBufferOffsetAlignment;
             return;
         case GL_TIMESTAMP: {
             Int64 timestamp = 0;
@@ -1707,20 +1772,22 @@ namespace MobileGL::MG_Impl::GLImpl {
             *params = vao ? static_cast<GLint>(vao->GetExternalIndex()) : 0;
             return;
         }
+        // The vertex buffer binding points are per-binding-index state, so the non-indexed getter
+        // has nothing to answer with (GL 4.6 core table 23.4).
+        case GL_VERTEX_BINDING_BUFFER:
         case GL_VERTEX_BINDING_DIVISOR:
-            *params = 0; // vertex-binding entrypoints are stubbed
-            return;
         case GL_VERTEX_BINDING_OFFSET:
-            *params = 0; // vertex-binding entrypoints are stubbed
-            return;
         case GL_VERTEX_BINDING_STRIDE:
-            *params = 0; // vertex-binding entrypoints are stubbed
+            RecordIndexedOnlyGetterError(__func__, pname);
             return;
         case GL_MAX_VERTEX_ATTRIB_RELATIVE_OFFSET:
-            *params = 0; // vertex-binding entrypoints are stubbed
+            *params = static_cast<GLint>(VertexArrayImpl::GetMaxVertexAttribRelativeOffset());
             return;
         case GL_MAX_VERTEX_ATTRIB_BINDINGS:
-            *params = 0; // vertex-binding entrypoints are stubbed
+            *params = static_cast<GLint>(VertexArrayImpl::GetMaxVertexAttribBindings());
+            return;
+        case GL_MAX_VERTEX_ATTRIB_STRIDE:
+            *params = static_cast<GLint>(VertexArrayImpl::GetMaxVertexAttribStride());
             return;
         case GL_VIEWPORT: {
             const auto& vp = MG_State::pGLContext->GetViewport();
@@ -1886,6 +1953,21 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_MAX_SAMPLE_MASK_WORDS:
             *params = dynamicParameters.MaxSampleMaskWords;
             break;
+        case GL_PATCH_VERTICES:
+            *params = static_cast<GLint>(MG_State::pGLContext->GetPatchVertices());
+            break;
+        case GL_MAX_PATCH_VERTICES:
+            *params = dynamicParameters.MaxPatchVertices;
+            break;
+        case GL_MAX_TESS_GEN_LEVEL:
+            *params = dynamicParameters.MaxTessGenLevel;
+            break;
+        case GL_MIN_PROGRAM_TEXTURE_GATHER_OFFSET:
+            *params = dynamicParameters.MinProgramTextureGatherOffset;
+            break;
+        case GL_MAX_PROGRAM_TEXTURE_GATHER_OFFSET:
+            *params = dynamicParameters.MaxProgramTextureGatherOffset;
+            break;
         case GL_MAX_SHADER_STORAGE_BUFFER_BINDINGS:
             *params = static_cast<GLint>(GetIndexedBufferQueryPointCount(BufferTarget::ShaderStorage));
             break;
@@ -1901,6 +1983,25 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_COMPONENTS:
             *params = kFrontendMaxTransformFeedbackSeparateComponents;
             break;
+        // ARB_transform_feedback3 limits. The GL CTS queries these before checking
+        // whether the extension is advertised and requires no GL error; desktop
+        // drivers all accept them, so answer with the separate-attrib capacity and
+        // the single vertex stream the backends provide.
+        case GL_MAX_TRANSFORM_FEEDBACK_BUFFERS:
+            *params = kFrontendMaxTransformFeedbackSeparateAttribs;
+            break;
+        case GL_MAX_VERTEX_STREAMS:
+            *params = 1;
+            break;
+        case GL_TRANSFORM_FEEDBACK_ACTIVE:
+            *params = MG_State::pGLContext->IsTransformFeedbackActive() ? 1 : 0;
+            break;
+        case GL_TRANSFORM_FEEDBACK_PAUSED:
+            *params = MG_State::pGLContext->IsTransformFeedbackPaused() ? 1 : 0;
+            break;
+        case GL_TRANSFORM_FEEDBACK_BINDING:
+            *params = static_cast<GLint>(MG_State::pGLContext->GetBoundTransformFeedbackName());
+            break;
         case GL_MAX_TEXTURE_IMAGE_UNITS:
             *params = dynamicParameters.MaxTextureImageUnits;
             break;
@@ -1957,6 +2058,15 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_SUBPIXEL_BITS:
             *params = std::max(dynamicParameters.ViewportSubpixelBits, kFrontendSubpixelBits);
             break;
+        case GL_MIN_FRAGMENT_INTERPOLATION_OFFSET:
+            *params = static_cast<GLint>(std::lround(dynamicParameters.MinFragmentInterpolationOffset));
+            break;
+        case GL_MAX_FRAGMENT_INTERPOLATION_OFFSET:
+            *params = static_cast<GLint>(std::lround(dynamicParameters.MaxFragmentInterpolationOffset));
+            break;
+        case GL_FRAGMENT_INTERPOLATION_OFFSET_BITS:
+            *params = dynamicParameters.FragmentInterpolationOffsetBits;
+            break;
         case GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT:
             *params = static_cast<Int>(dynamicParameters.UniformBufferOffsetAlignment);
             break;

MobileGL/MG_Impl/GLImpl/Program/GL_Program.cpp

@@ -8,6 +8,8 @@
 
 #include "GL_Program.h"
 #include "Config.h"
+#include <cmath>
+#include <limits>
 #include <MG_Impl/GLImpl/VertexArray/Validators.h>
 #include <MG_State/GLState/Core.h>
 #include <MG_Util/Converters/GLToStr/GLEnumConverter.h>
@@ -49,10 +51,18 @@ namespace MobileGL::MG_Impl::GLImpl {
 
     static bool CheckProgramNameValidity(GLuint program) {
         if (!MG_State::pGLContext->ValidateProgramName(program)) {
+            // Programs and shaders share one name space: a name that exists but
+            // belongs to a shader is INVALID_OPERATION, a name GL never handed
+            // out is INVALID_VALUE (GL 3.3 core 2.11.x).
+            const ErrorCode error = MG_State::pGLContext->ValidateShaderName(program)
+                ? ErrorCode::InvalidOperation
+                : ErrorCode::InvalidValue;
             MG_State::pGLContext->RecordError(
-                ErrorCode::InvalidValue,
+                error,
                 MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
-                                             std::to_string(program) + " is not a valid name."));
+                                             std::to_string(program) +
+                                                 (error == ErrorCode::InvalidOperation ? " is not a program object."
+                                                                                       : " is not a valid name.")));
             return false;
         }
         return true;
@@ -195,6 +205,50 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
     }
 
+    // The GL_UNIFORM interface and glGetActiveUniform(s)iv are the same query in two
+    // spellings, so they answer from the same place - the frontend reflection. The backend
+    // program is not that place: it does not exist at all for a program whose types its
+    // shading language cannot express (a double-precision uniform has no ESSL form), and
+    // the interface queries would then describe a program with no uniforms.
+    //
+    // Writes the GL_UNIFORM value of `prop` for active uniform `index`; false for a prop
+    // the reflection does not model, which the caller forwards to the backend instead.
+    Bool GetUniformResourceProp(const SharedPtr<MG_State::GLState::ProgramObject>& programObject, Uint index,
+                                GLenum prop, GLint* out) {
+        switch (prop) {
+        case GL_TYPE:
+            *out = static_cast<GLint>(programObject->GetActiveUniformType(index));
+            return true;
+        case GL_ARRAY_SIZE:
+            *out = programObject->GetActiveUniformArraySize(index);
+            return true;
+        case GL_NAME_LENGTH:
+            *out = static_cast<GLint>(programObject->GetActiveUniformName(index).length() + 1);
+            return true;
+        case GL_BLOCK_INDEX:
+            *out = programObject->GetActiveUniformBlockIndex(index);
+            return true;
+        case GL_OFFSET:
+            *out = programObject->GetActiveUniformOffset(index);
+            return true;
+        case GL_ARRAY_STRIDE:
+            *out = programObject->GetActiveUniformArrayStride(index);
+            return true;
+        case GL_MATRIX_STRIDE:
+            *out = programObject->GetActiveUniformMatrixStride(index);
+            return true;
+        case GL_IS_ROW_MAJOR:
+            *out = programObject->GetActiveUniformIsRowMajor(index);
+            return true;
+        case GL_LOCATION:
+            // A block member has no location; GetUniformLocation already reports -1 for one.
+            *out = programObject->GetUniformLocation(programObject->GetActiveUniformName(index));
+            return true;
+        default:
+            return false;
+        }
+    }
+
     void CopyStr(GLsizei bufSize, GLsizei* length, GLchar* dst, const char* src, GLsizei srcLength) {
         if (bufSize <= 0) {
             if (length) *length = 0;
@@ -317,11 +371,16 @@ namespace MobileGL::MG_Impl::GLImpl {
     }
 
     void DeleteProgram_State(GLuint program) {
+        // "If program is zero, it is silently ignored" (GL 4.6 core 7.3) - unlike every
+        // other program entry point, where 0 is a name GL never handed out.
+        if (program == 0) return;
         if (!CheckProgramNameValidity(program)) return;
         MG_State::pGLContext->MarkProgramForDeletion(program);
     }
 
     void DeleteShader_State(GLuint shader) {
+        // Same silent-zero rule as glDeleteProgram (GL 4.6 core 7.1).
+        if (shader == 0) return;
         if (!CheckShaderNameValidity(shader)) return;
         MG_State::pGLContext->MarkShaderForDeletion(shader);
     }
@@ -605,6 +664,18 @@ namespace MobileGL::MG_Impl::GLImpl {
             *params = programObject->GetActiveUniformBlocksMaxNameLength() + 1;
             MGLOG_D("%s: %s = %d", __func__, MG_Util::ConvertGLEnumToString(pname).c_str(), *params);
             break;
+        case GL_TRANSFORM_FEEDBACK_VARYINGS:
+            *params = static_cast<GLint>(programObject->GetTransformFeedbackVaryingCount());
+            MGLOG_D("%s: %s = %d", __func__, MG_Util::ConvertGLEnumToString(pname).c_str(), *params);
+            break;
+        case GL_TRANSFORM_FEEDBACK_BUFFER_MODE:
+            *params = static_cast<GLint>(programObject->GetTransformFeedbackBufferMode());
+            MGLOG_D("%s: %s = %d", __func__, MG_Util::ConvertGLEnumToString(pname).c_str(), *params);
+            break;
+        case GL_TRANSFORM_FEEDBACK_VARYING_MAX_LENGTH:
+            *params = programObject->GetTransformFeedbackVaryingMaxLength();
+            MGLOG_D("%s: %s = %d", __func__, MG_Util::ConvertGLEnumToString(pname).c_str(), *params);
+            break;
         case GL_COMPUTE_WORK_GROUP_SIZE: { // GL >= 4.3
             if (!programObject->GetLinkStatus() || programObject->GetShaderIndexByStage(ShaderStage::Compute) < 0) {
                 MG_State::pGLContext->RecordError(
@@ -623,10 +694,14 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
 
         case GL_PROGRAM_BINARY_LENGTH:
+            // No program binary format is exposed, so a program never has a retrievable
+            // binary and its length is zero (ARB_get_program_binary).
+            *params = 0;
+            break;
+        case GL_PROGRAM_BINARY_RETRIEVABLE_HINT:
+            *params = programObject->GetBinaryRetrievableHint() ? GL_TRUE : GL_FALSE;
+            break;
 
-        case GL_TRANSFORM_FEEDBACK_BUFFER_MODE:
-        case GL_TRANSFORM_FEEDBACK_VARYINGS:
-        case GL_TRANSFORM_FEEDBACK_VARYING_MAX_LENGTH:
         case GL_GEOMETRY_VERTICES_OUT:
         case GL_GEOMETRY_INPUT_TYPE:
         case GL_GEOMETRY_OUTPUT_TYPE:
@@ -801,7 +876,8 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
 
         if constexpr (std::is_same_v<T, GLfloat>) {
-            if (ttype->isMatrix() && ttype->getMatrixCols() == 3) {
+            if (ttype->getBasicType() != glslang::EbtDouble && ttype->isMatrix() &&
+                ttype->getMatrixCols() == 3) {
                 auto* pBase = pUBO + offset;
                 for (int i = 0; i < ttype->getMatrixRows(); i++) {
                     Memcpy(reinterpret_cast<char*>(params) + ttype->getMatrixCols() * sizeof(GLfloat) * i,
@@ -811,9 +887,46 @@ namespace MobileGL::MG_Impl::GLImpl {
             }
         }
 
+        // A double-precision uniform is the one case where the stored component type can
+        // differ from the queried one for a non-opaque uniform, and the difference is not
+        // just a reinterpretation: it is twice as wide, so a raw copy would overrun the
+        // caller's buffer as well as return nonsense. Read component by component and let
+        // GL's conversion rules (7.6: round to nearest for the integer queries) apply.
+        if (ttype->getBasicType() == glslang::EbtDouble) {
+            const Int columns = ttype->isMatrix() ? ttype->getMatrixCols() : 1;
+            const Int rows = ttype->isMatrix() ? ttype->getMatrixRows()
+                                               : (ttype->isVector() ? ttype->getVectorSize() : 1);
+            // The slot the linker handed out is exactly `columns` columns wide, so it also
+            // states the column stride - which for a double matrix is not a float's 16 bytes.
+            const SizeT columnStride = columns > 0 ? size / static_cast<SizeT>(columns) : size;
+            for (Int column = 0; column < columns; ++column) {
+                for (Int row = 0; row < rows; ++row) {
+                    GLdouble component = 0.0;
+                    Memcpy(&component, pUBO + offset + column * columnStride + row * sizeof(GLdouble),
+                           sizeof(component));
+                    if constexpr (std::is_integral_v<T>) {
+                        // Rounded to the nearest integer and clamped into the queried type's
+                        // range, so a negative double read through glGetUniformuiv is 0
+                        // rather than its two's complement.
+                        const GLdouble rounded = std::nearbyint(component);
+                        const GLdouble lowest = static_cast<GLdouble>(std::numeric_limits<T>::lowest());
+                        const GLdouble highest = static_cast<GLdouble>(std::numeric_limits<T>::max());
+                        params[column * rows + row] = static_cast<T>(std::clamp(rounded, lowest, highest));
+                    } else {
+                        params[column * rows + row] = static_cast<T>(component);
+                    }
+                }
+            }
+            return;
+        }
+
         Memcpy(params, pUBO + offset, size);
     }
 
+    void GetUniformdv_State(GLuint program, GLint location, GLdouble* params) {
+        GetUniformScalar_State(program, location, params);
+    }
+
     void GetUniformfv_State(GLuint program, GLint location, GLfloat* params) {
         GetUniformScalar_State(program, location, params);
     }
@@ -827,19 +940,13 @@ namespace MobileGL::MG_Impl::GLImpl {
     }
 
     GLboolean IsProgram_State(GLuint program) {
-        /* FIXME: Handle situations that:
-         * A program object marked for deletion with glDeleteProgram but still in use as part of current
-         * rendering state is still considered a program object and glIsProgram will return GL_TRUE.
-         */
+        // Deletion-flagged names stay valid while the object is still GL-visible (program in
+        // use, shader attached), so name validity is exactly the Is* answer.
         if (program == 0) return GL_FALSE;
         return MG_State::pGLContext->ValidateProgramName(program) ? GL_TRUE : GL_FALSE;
     }
 
     GLboolean IsShader_State(GLuint shader) {
-        /* FIXME: Handle situations that:
-         * A shader object marked for deletion with glDeleteShader but still attached to a program object is still
-         * considered a shader object and glIsShader will return GL_TRUE.
-         */
         if (shader == 0) return GL_FALSE;
         return MG_State::pGLContext->ValidateShaderName(shader) ? GL_TRUE : GL_FALSE;
     }
@@ -849,6 +956,18 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (!programObject) return;
         MGLOG_D("%s: linking program %d", __func__, program);
 
+        // Relinking the program an active transform feedback captures from would
+        // invalidate its varyings mid-capture (GL 3.3 core 2.11.3).
+        if (MG_State::pGLContext->IsTransformFeedbackActive() &&
+            MG_State::pGLContext->GetTransformFeedbackProgram().get() == programObject.get()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>(
+                    "MG_Impl/GLImpl", __func__,
+                    "The program used by active transform feedback cannot be relinked."));
+            return;
+        }
+
         static Bool allowVSOnlyPrograms;
         static Bool initialized = false;
         if (!initialized) {
@@ -892,6 +1011,18 @@ namespace MobileGL::MG_Impl::GLImpl {
     void UseProgram_State(GLuint program) {
         MGLOG_D("UseProgram_State: program=%u", program);
 
+        // The program in use may not change while transform feedback is active - unless
+        // the capture is paused, which is exactly what ARB_transform_feedback2 added the
+        // pause for (GL 4.6 core 7.3).
+        if (MG_State::pGLContext->IsTransformFeedbackActive() &&
+            !MG_State::pGLContext->IsTransformFeedbackPaused()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "The current program cannot change while transform feedback is active."));
+            return;
+        }
+
         if (program == 0) {
             MG_State::pGLContext->UseProgram(0);
             return;
@@ -1014,6 +1145,38 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
     }
 
+    // glUniform*d / glUniformMatrix*dv. The vector forms need nothing beyond the shared
+    // upload template - it is already typed on the component - but a matrix does: the
+    // column stride the linker used for a double matrix is not the 16 bytes a float one
+    // gets. It is not guessed here; the slot the uniform was given is exactly `columns`
+    // columns wide, so dividing states the stride the rest of the pipeline agreed on.
+    template <typename Program>
+    void UniformMatrixdv_Object(Program& programObject, GLint location, GLsizei count, GLboolean transpose,
+                                const GLdouble* value, Int columns, Int rows) {
+        const SizeT slotSize = programObject.GetUniformSizesInBytes(location);
+        const SizeT columnStride = columns > 0 ? slotSize / static_cast<SizeT>(columns) : slotSize;
+        const SizeT componentCount = static_cast<SizeT>(columns) * static_cast<SizeT>(rows);
+        Vector<GLdouble> column(static_cast<SizeT>(rows));
+        for (GLint matrix = 0; matrix < count; ++matrix) {
+            if (matrix > 0 && !programObject.UniformLocationsAliasSameUniform(location, location + matrix)) break;
+            if (!programObject.IsValidUniformLocation(location + matrix)) {
+                RecordInvalidUniformLocationError(__func__, location + matrix, "the current program object");
+                return;
+            }
+            const GLdouble* source = value + matrix * componentCount;
+            for (Int c = 0; c < columns; ++c) {
+                for (Int r = 0; r < rows; ++r) {
+                    column[r] = transpose == GL_TRUE ? source[r * columns + c] : source[c * rows + r];
+                }
+                Uniform_State<1>(programObject, location + matrix, column.data(), c * columnStride);
+                for (Int r = 1; r < rows; ++r) {
+                    Uniform_State<1>(programObject, location + matrix, column.data() + r,
+                                     c * columnStride + r * sizeof(GLdouble));
+                }
+            }
+        }
+    }
+
     // Helper function to transpose a 2x2 matrix
     void TransposeMatrix2x2(const GLfloat* input, GLfloat* output) {
         // Input matrix is in column-major order (OpenGL default)
@@ -1822,6 +1985,312 @@ namespace MobileGL::MG_Impl::GLImpl {
         GLuint v[] = {v0, v1, v2, v3};
         Uniform4uiv(location, 1, v);
     }
+    void Uniform1d(GLint location, GLdouble v0) {
+        const GLdouble v[] = {v0};
+        Uniformv_State<1>(location, 1, v);
+    }
+
+    void Uniform1dv(GLint location, GLsizei count, const GLdouble* value) {
+        Uniformv_State<1>(location, count, value);
+    }
+
+    void ProgramUniform1d(GLuint program, GLint location, GLdouble v0) {
+        const GLdouble v[] = {v0};
+        ProgramUniformv_State<1>(program, location, 1, v);
+    }
+
+    void ProgramUniform1dv(GLuint program, GLint location, GLsizei count, const GLdouble* value) {
+        ProgramUniformv_State<1>(program, location, count, value);
+    }
+    void Uniform2d(GLint location, GLdouble v0, GLdouble v1) {
+        const GLdouble v[] = {v0, v1};
+        Uniformv_State<2>(location, 1, v);
+    }
+
+    void Uniform2dv(GLint location, GLsizei count, const GLdouble* value) {
+        Uniformv_State<2>(location, count, value);
+    }
+
+    void ProgramUniform2d(GLuint program, GLint location, GLdouble v0, GLdouble v1) {
+        const GLdouble v[] = {v0, v1};
+        ProgramUniformv_State<2>(program, location, 1, v);
+    }
+
+    void ProgramUniform2dv(GLuint program, GLint location, GLsizei count, const GLdouble* value) {
+        ProgramUniformv_State<2>(program, location, count, value);
+    }
+    void Uniform3d(GLint location, GLdouble v0, GLdouble v1, GLdouble v2) {
+        const GLdouble v[] = {v0, v1, v2};
+        Uniformv_State<3>(location, 1, v);
+    }
+
+    void Uniform3dv(GLint location, GLsizei count, const GLdouble* value) {
+        Uniformv_State<3>(location, count, value);
+    }
+
+    void ProgramUniform3d(GLuint program, GLint location, GLdouble v0, GLdouble v1, GLdouble v2) {
+        const GLdouble v[] = {v0, v1, v2};
+        ProgramUniformv_State<3>(program, location, 1, v);
+    }
+
+    void ProgramUniform3dv(GLuint program, GLint location, GLsizei count, const GLdouble* value) {
+        ProgramUniformv_State<3>(program, location, count, value);
+    }
+    void Uniform4d(GLint location, GLdouble v0, GLdouble v1, GLdouble v2, GLdouble v3) {
+        const GLdouble v[] = {v0, v1, v2, v3};
+        Uniformv_State<4>(location, 1, v);
+    }
+
+    void Uniform4dv(GLint location, GLsizei count, const GLdouble* value) {
+        Uniformv_State<4>(location, count, value);
+    }
+
+    void ProgramUniform4d(GLuint program, GLint location, GLdouble v0, GLdouble v1, GLdouble v2, GLdouble v3) {
+        const GLdouble v[] = {v0, v1, v2, v3};
+        ProgramUniformv_State<4>(program, location, 1, v);
+    }
+
+    void ProgramUniform4dv(GLuint program, GLint location, GLsizei count, const GLdouble* value) {
+        ProgramUniformv_State<4>(program, location, count, value);
+    }
+    void UniformMatrix2dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = MG_State::pGLContext->GetCurrentProgram();
+        if (programObject == nullptr) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "There is no current program object."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 2, 2);
+    }
+
+    void ProgramUniformMatrix2dv(GLuint program, GLint location, GLsizei count, GLboolean transpose,
+                                      const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = TryToGetProgramObject(program);
+        if (!programObject) return;
+        if (!programObject->GetLinkStatus()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "program " + std::to_string(program) + " is not linked."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 2, 2);
+    }
+    void UniformMatrix3dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = MG_State::pGLContext->GetCurrentProgram();
+        if (programObject == nullptr) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "There is no current program object."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 3, 3);
+    }
+
+    void ProgramUniformMatrix3dv(GLuint program, GLint location, GLsizei count, GLboolean transpose,
+                                      const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = TryToGetProgramObject(program);
+        if (!programObject) return;
+        if (!programObject->GetLinkStatus()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "program " + std::to_string(program) + " is not linked."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 3, 3);
+    }
+    void UniformMatrix4dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = MG_State::pGLContext->GetCurrentProgram();
+        if (programObject == nullptr) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "There is no current program object."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 4, 4);
+    }
+
+    void ProgramUniformMatrix4dv(GLuint program, GLint location, GLsizei count, GLboolean transpose,
+                                      const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = TryToGetProgramObject(program);
+        if (!programObject) return;
+        if (!programObject->GetLinkStatus()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "program " + std::to_string(program) + " is not linked."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 4, 4);
+    }
+    void UniformMatrix2x3dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = MG_State::pGLContext->GetCurrentProgram();
+        if (programObject == nullptr) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "There is no current program object."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 2, 3);
+    }
+
+    void ProgramUniformMatrix2x3dv(GLuint program, GLint location, GLsizei count, GLboolean transpose,
+                                      const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = TryToGetProgramObject(program);
+        if (!programObject) return;
+        if (!programObject->GetLinkStatus()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "program " + std::to_string(program) + " is not linked."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 2, 3);
+    }
+    void UniformMatrix2x4dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = MG_State::pGLContext->GetCurrentProgram();
+        if (programObject == nullptr) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "There is no current program object."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 2, 4);
+    }
+
+    void ProgramUniformMatrix2x4dv(GLuint program, GLint location, GLsizei count, GLboolean transpose,
+                                      const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = TryToGetProgramObject(program);
+        if (!programObject) return;
+        if (!programObject->GetLinkStatus()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "program " + std::to_string(program) + " is not linked."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 2, 4);
+    }
+    void UniformMatrix3x2dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = MG_State::pGLContext->GetCurrentProgram();
+        if (programObject == nullptr) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "There is no current program object."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 3, 2);
+    }
+
+    void ProgramUniformMatrix3x2dv(GLuint program, GLint location, GLsizei count, GLboolean transpose,
+                                      const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = TryToGetProgramObject(program);
+        if (!programObject) return;
+        if (!programObject->GetLinkStatus()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "program " + std::to_string(program) + " is not linked."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 3, 2);
+    }
+    void UniformMatrix3x4dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = MG_State::pGLContext->GetCurrentProgram();
+        if (programObject == nullptr) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "There is no current program object."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 3, 4);
+    }
+
+    void ProgramUniformMatrix3x4dv(GLuint program, GLint location, GLsizei count, GLboolean transpose,
+                                      const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = TryToGetProgramObject(program);
+        if (!programObject) return;
+        if (!programObject->GetLinkStatus()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "program " + std::to_string(program) + " is not linked."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 3, 4);
+    }
+    void UniformMatrix4x2dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = MG_State::pGLContext->GetCurrentProgram();
+        if (programObject == nullptr) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "There is no current program object."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 4, 2);
+    }
+
+    void ProgramUniformMatrix4x2dv(GLuint program, GLint location, GLsizei count, GLboolean transpose,
+                                      const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = TryToGetProgramObject(program);
+        if (!programObject) return;
+        if (!programObject->GetLinkStatus()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "program " + std::to_string(program) + " is not linked."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 4, 2);
+    }
+    void UniformMatrix4x3dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = MG_State::pGLContext->GetCurrentProgram();
+        if (programObject == nullptr) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "There is no current program object."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 4, 3);
+    }
+
+    void ProgramUniformMatrix4x3dv(GLuint program, GLint location, GLsizei count, GLboolean transpose,
+                                      const GLdouble* value) {
+        if (location == -1) return;
+        auto& programObject = TryToGetProgramObject(program);
+        if (!programObject) return;
+        if (!programObject->GetLinkStatus()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "program " + std::to_string(program) + " is not linked."));
+            return;
+        }
+        UniformMatrixdv_Object(*programObject, location, count, transpose, value, 4, 3);
+    }
+    void GetUniformdv(GLuint program, GLint location, GLdouble* params) {
+        GetUniformdv_State(program, location, params);
+    }
+
     void Uniform1fv(GLint location, GLsizei count, const GLfloat* value) {
         Uniform1fv_State(location, count, value);
     }
@@ -2101,6 +2570,17 @@ namespace MobileGL::MG_Impl::GLImpl {
                                              "Backend does not support program interface queries."));
             return;
         }
+        if (programInterface == GL_UNIFORM) {
+            if (pname == GL_ACTIVE_RESOURCES) {
+                *params = static_cast<GLint>(programObject->GetUniformCount());
+                return;
+            }
+            if (pname == GL_MAX_NAME_LENGTH) {
+                // Stored as the bare length; GL_MAX_NAME_LENGTH counts the terminator.
+                *params = programObject->GetUniformMaxLength() + 1;
+                return;
+            }
+        }
         getProgramInterfaceiv(program, programInterface, pname, params);
     }
 
@@ -2109,6 +2589,10 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (!programObject) return GL_INVALID_INDEX;
         if (!ValidateNamedProgramResourceInterface(programInterface, __func__)) return GL_INVALID_INDEX;
         if (!name) return GL_INVALID_INDEX;
+        if (programInterface == GL_UNIFORM) {
+            const Int uniformIndex = programObject->GetActiveUniformIndex(name);
+            return uniformIndex < 0 ? GL_INVALID_INDEX : static_cast<GLuint>(uniformIndex);
+        }
         auto getProgramResourceIndex = MG_Backend::gBackendFunctionsTable.GL.GetProgramResourceIndex;
         if (!getProgramResourceIndex) {
             MG_State::pGLContext->RecordError(
@@ -2145,6 +2629,13 @@ namespace MobileGL::MG_Impl::GLImpl {
                 MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "bufSize must be non-negative."));
             return;
         }
+        if (programInterface == GL_UNIFORM) {
+            // Same index space GetProgramResourceIndex answers in, and the range check above
+            // already used it.
+            const String& uniformName = programObject->GetActiveUniformName(index);
+            CopyStr(bufSize, length, name, uniformName.c_str(), static_cast<GLsizei>(uniformName.length()));
+            return;
+        }
         auto getProgramResourceName = MG_Backend::gBackendFunctionsTable.GL.GetProgramResourceName;
         if (!getProgramResourceName) {
             MG_State::pGLContext->RecordError(
@@ -2166,6 +2657,36 @@ namespace MobileGL::MG_Impl::GLImpl {
                                                                       "propCount and bufSize must be non-negative."));
             return;
         }
+        if (programInterface == GL_UNIFORM) {
+            if (index >= programObject->GetUniformCount()) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidValue,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "index is out of range."));
+                return;
+            }
+            if (props == nullptr || params == nullptr) return;
+            GLsizei written = 0;
+            for (GLsizei i = 0; i < propCount && written < bufSize; ++i) {
+                GLint value = 0;
+                if (!GetUniformResourceProp(programObject, index, props[i], &value)) {
+                    // GL_ATOMIC_COUNTER_BUFFER_INDEX and the GL_REFERENCED_BY_* stage props are
+                    // not modelled here; ask the backend, which indexes resources by name.
+                    auto backendGetIndex = MG_Backend::gBackendFunctionsTable.GL.GetProgramResourceIndex;
+                    auto backendGetiv = MG_Backend::gBackendFunctionsTable.GL.GetProgramResourceiv;
+                    if (backendGetIndex && backendGetiv) {
+                        const GLuint backendIndex = backendGetIndex(program, GL_UNIFORM,
+                                                                   programObject->GetActiveUniformName(index).c_str());
+                        if (backendIndex != GL_INVALID_INDEX) {
+                            GLsizei one = 0;
+                            backendGetiv(program, GL_UNIFORM, backendIndex, 1, &props[i], 1, &one, &value);
+                        }
+                    }
+                }
+                params[written++] = value;
+            }
+            if (length) *length = written;
+            return;
+        }
         auto getProgramResourceiv = MG_Backend::gBackendFunctionsTable.GL.GetProgramResourceiv;
         if (!getProgramResourceiv) {
             MG_State::pGLContext->RecordError(
@@ -2217,4 +2738,150 @@ namespace MobileGL::MG_Impl::GLImpl {
     void ValidateProgram(GLuint program) {
         ValidateProgram_State(program);
     }
+
+    // ARB_get_program_binary with no supported binary format (GL_NUM_PROGRAM_BINARY_FORMATS
+    // is 0, which the extension explicitly allows). The three entry points below are what an
+    // application - and dEQP's function loader - reach through the extension; without it
+    // glProgramParameteri is not exposed in a 4.0 context at all.
+    void ProgramParameteri(GLuint program, GLenum pname, GLint value) {
+        auto& programObject = TryToGetProgramObject(program);
+        if (!programObject) return;
+        if (pname != GL_PROGRAM_BINARY_RETRIEVABLE_HINT) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "pname is not an accepted value."));
+            return;
+        }
+        if (value != GL_TRUE && value != GL_FALSE) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "value must be GL_TRUE or GL_FALSE."));
+            return;
+        }
+        programObject->SetBinaryRetrievableHint(value == GL_TRUE);
+    }
+
+    void GetProgramBinary(GLuint program, GLsizei bufSize, GLsizei* length, GLenum* binaryFormat, void* binary) {
+        (void)binaryFormat;
+        (void)binary;
+        auto& programObject = TryToGetProgramObject(program);
+        if (!programObject) return;
+        if (bufSize < 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "bufSize must be non-negative."));
+            return;
+        }
+        if (length) *length = 0;
+        // GL_PROGRAM_BINARY_LENGTH is always zero here, which the spec makes an error to ask for.
+        MG_State::pGLContext->RecordError(
+            ErrorCode::InvalidOperation,
+            MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "The program has no retrievable binary."));
+    }
+
+    void ProgramBinary(GLuint program, GLenum binaryFormat, const void* binary, GLsizei length) {
+        (void)binaryFormat;
+        (void)binary;
+        auto& programObject = TryToGetProgramObject(program);
+        if (!programObject) return;
+        if (length < 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "length must be non-negative."));
+            return;
+        }
+        // No format is supported, so every binary is rejected - and the program's link status
+        // has to read FALSE afterwards.
+        programObject->MarkLinkFailedByProgramBinary();
+        MG_State::pGLContext->RecordError(
+            ErrorCode::InvalidEnum,
+            MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "binaryFormat is not a supported format."));
+    }
+
+    void TransformFeedbackVaryings(GLuint program, GLsizei count, const GLchar* const* varyings, GLenum bufferMode) {
+        auto& programObject = TryToGetProgramObject(program);
+        if (!programObject) return;
+        if (bufferMode != GL_INTERLEAVED_ATTRIBS && bufferMode != GL_SEPARATE_ATTRIBS) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "bufferMode is not a valid capture mode."));
+            return;
+        }
+        if (count < 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "count must be non-negative."));
+            return;
+        }
+        // GL 3.3 core: SEPARATE_ATTRIBS count may not exceed the separate-attrib limit.
+        if (bufferMode == GL_SEPARATE_ATTRIBS && count > 4) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "count exceeds GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_ATTRIBS."));
+            return;
+        }
+        Vector<String> names;
+        names.reserve(static_cast<SizeT>(count));
+        for (GLsizei i = 0; i < count; ++i) {
+            names.emplace_back(varyings != nullptr && varyings[i] != nullptr ? varyings[i] : "");
+        }
+        // ARB_transform_feedback3's special names only mean anything in an interleaved
+        // capture, and gl_NextBuffer cannot advance past the last capture buffer.
+        constexpr Uint maxTransformFeedbackBuffers = 4;
+        Uint nextBufferCount = 0;
+        for (const String& name : names) {
+            const Bool isNextBuffer = name == "gl_NextBuffer";
+            const Bool isSkipComponents = name.size() == 18 && name.compare(0, 17, "gl_SkipComponents") == 0 &&
+                                          name[17] >= '1' && name[17] <= '4';
+            if (!isNextBuffer && !isSkipComponents) continue;
+            if (bufferMode != GL_INTERLEAVED_ATTRIBS) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                                 "'" + name + "' requires GL_INTERLEAVED_ATTRIBS."));
+                return;
+            }
+            if (isNextBuffer && ++nextBufferCount >= maxTransformFeedbackBuffers) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                                 "More gl_NextBuffer entries than "
+                                                 "GL_MAX_TRANSFORM_FEEDBACK_BUFFERS allows."));
+                return;
+            }
+        }
+        programObject->SetTransformFeedbackVaryings(Move(names), bufferMode);
+    }
+
+    void GetTransformFeedbackVarying(GLuint program, GLuint index, GLsizei bufSize, GLsizei* length, GLsizei* size,
+                                     GLenum* type, GLchar* name) {
+        auto& programObject = TryToGetProgramObject(program);
+        if (!programObject) return;
+        if (!programObject->GetLinkStatus()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             std::to_string(program) + " has not been successfully linked."));
+            return;
+        }
+        const auto* varying = programObject->GetTransformFeedbackVarying(index);
+        if (varying == nullptr) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>(
+                    "MG_Impl/GLImpl", __func__,
+                    "index is not an active transform feedback varying of the program."));
+            return;
+        }
+        if (size != nullptr) *size = varying->size;
+        if (type != nullptr) *type = varying->type;
+        GLsizei written = 0;
+        if (name != nullptr && bufSize > 0) {
+            written = std::min<GLsizei>(bufSize - 1, static_cast<GLsizei>(varying->name.size()));
+            Memcpy(name, varying->name.data(), static_cast<SizeT>(written));
+            name[written] = '\0';
+        }
+        if (length != nullptr) *length = written;
+    }
 } // namespace MobileGL::MG_Impl::GLImpl

MobileGL/MG_Impl/GLImpl/Program/GL_Program.h

@@ -137,5 +137,46 @@ namespace MobileGL::MG_Impl::GLImpl {
     GLint GetProgramResourceLocation(GLuint program, GLenum programInterface, const GLchar* name);
     GLint GetProgramResourceLocationIndex(GLuint program, GLenum programInterface, const GLchar* name);
     void ShaderStorageBlockBinding(GLuint program, GLuint storageBlockIndex, GLuint storageBlockBinding);
+    void Uniform1d(GLint location, GLdouble v0);
+    void Uniform1dv(GLint location, GLsizei count, const GLdouble* value);
+    void ProgramUniform1d(GLuint program, GLint location, GLdouble v0);
+    void ProgramUniform1dv(GLuint program, GLint location, GLsizei count, const GLdouble* value);
+    void Uniform2d(GLint location, GLdouble v0, GLdouble v1);
+    void Uniform2dv(GLint location, GLsizei count, const GLdouble* value);
+    void ProgramUniform2d(GLuint program, GLint location, GLdouble v0, GLdouble v1);
+    void ProgramUniform2dv(GLuint program, GLint location, GLsizei count, const GLdouble* value);
+    void Uniform3d(GLint location, GLdouble v0, GLdouble v1, GLdouble v2);
+    void Uniform3dv(GLint location, GLsizei count, const GLdouble* value);
+    void ProgramUniform3d(GLuint program, GLint location, GLdouble v0, GLdouble v1, GLdouble v2);
+    void ProgramUniform3dv(GLuint program, GLint location, GLsizei count, const GLdouble* value);
+    void Uniform4d(GLint location, GLdouble v0, GLdouble v1, GLdouble v2, GLdouble v3);
+    void Uniform4dv(GLint location, GLsizei count, const GLdouble* value);
+    void ProgramUniform4d(GLuint program, GLint location, GLdouble v0, GLdouble v1, GLdouble v2, GLdouble v3);
+    void ProgramUniform4dv(GLuint program, GLint location, GLsizei count, const GLdouble* value);
+    void UniformMatrix2dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void ProgramUniformMatrix2dv(GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void UniformMatrix3dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void ProgramUniformMatrix3dv(GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void UniformMatrix4dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void ProgramUniformMatrix4dv(GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void UniformMatrix2x3dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void ProgramUniformMatrix2x3dv(GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void UniformMatrix2x4dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void ProgramUniformMatrix2x4dv(GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void UniformMatrix3x2dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void ProgramUniformMatrix3x2dv(GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void UniformMatrix3x4dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void ProgramUniformMatrix3x4dv(GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void UniformMatrix4x2dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void ProgramUniformMatrix4x2dv(GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void UniformMatrix4x3dv(GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void ProgramUniformMatrix4x3dv(GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLdouble* value);
+    void GetUniformdv(GLuint program, GLint location, GLdouble* params);
     void ValidateProgram(GLuint program);
+    void ProgramParameteri(GLuint program, GLenum pname, GLint value);
+    void GetProgramBinary(GLuint program, GLsizei bufSize, GLsizei* length, GLenum* binaryFormat, void* binary);
+    void ProgramBinary(GLuint program, GLenum binaryFormat, const void* binary, GLsizei length);
+    void TransformFeedbackVaryings(GLuint program, GLsizei count, const GLchar* const* varyings, GLenum bufferMode);
+    void GetTransformFeedbackVarying(GLuint program, GLuint index, GLsizei bufSize, GLsizei* length, GLsizei* size,
+                                     GLenum* type, GLchar* name);
 } // namespace MobileGL::MG_Impl::GLImpl

MobileGL/MG_Impl/GLImpl/Query/GL_Query.cpp

@@ -7,6 +7,7 @@
 // End of Source File Header
 
 #include "GL_Query.h"
+#include "../Getter/GL_Getter.h"
 #include <Config.h>
 #include <MG_Backend/BackendObjects.h>
 #include <MG_State/GLState/Core.h>
@@ -22,11 +23,16 @@ namespace MobileGL::MG_Impl::GLImpl {
         struct QueryObject {
             GLuint id = 0;
             GLenum target = 0; // 0 = gen'd but never used with BeginQuery/QueryCounter
+            // glCreateQueries makes the object outright; glGenQueries only reserves the name,
+            // and the object appears when the name is first used (GL 4.6 core 4.2.1).
+            Bool created = false;
             MG_Backend::BackendQueryHandle backendHandle = nullptr;
             Bool active = false;
             Bool ended = false;
             Bool resultCached = false;
             Uint64 cachedResult = 0;
+            // Transform feedback primitive counter at BeginQuery time.
+            Uint64 counterSnapshot = 0;
         };
 
         // Query calls may arrive from any thread (launchers migrate the context
@@ -41,6 +47,11 @@ namespace MobileGL::MG_Impl::GLImpl {
         GLuint g_nextQueryId = 1;
         // Id of the query currently active on GL_TIME_ELAPSED (0 = none).
         GLuint g_activeTimeElapsedQueryId = 0;
+        // Ids of the queries active on the transform feedback targets (0 = none).
+        GLuint g_activePrimitivesWrittenQueryId = 0;
+        GLuint g_activePrimitivesGeneratedQueryId = 0;
+        // Id of the query active on GL_SAMPLES_PASSED (0 = none).
+        GLuint g_activeSamplesPassedQueryId = 0;
 
         Bool TimerQueryDisabled() {
             return MG_Config::Features.DisableTimerQuery;
@@ -126,6 +137,11 @@ namespace MobileGL::MG_Impl::GLImpl {
                         outValue = 0;
                         return true;
                     }
+                    // ANY_SAMPLES_PASSED* report a boolean.
+                    if (queryObject->target == GL_ANY_SAMPLES_PASSED ||
+                        queryObject->target == GL_ANY_SAMPLES_PASSED_CONSERVATIVE) {
+                        result = result != 0 ? 1 : 0;
+                    }
                     // Final value produced (or no GetQueryResult64 hook: the
                     // query degrades to a zero result); the backend handle is
                     // consumed and the value cached for later reads.
@@ -164,6 +180,41 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
     }
 
+    // glCreateQueries differs from glGenQueries in creating the objects outright, with their
+    // target already fixed and the rest of their state at the defaults (GL 4.6 core 4.2.1).
+    void CreateQueries(GLenum target, GLsizei n, GLuint* ids) {
+        switch (target) {
+        case GL_SAMPLES_PASSED:
+        case GL_ANY_SAMPLES_PASSED:
+        case GL_ANY_SAMPLES_PASSED_CONSERVATIVE:
+        case GL_TIME_ELAPSED:
+        case GL_TIMESTAMP:
+        case GL_PRIMITIVES_GENERATED:
+        case GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN:
+            break;
+        default:
+            RecordQueryError(ErrorCode::InvalidEnum, __FUNCTION__, "Query target is not accepted.");
+            return;
+        }
+        if (n < 0) {
+            RecordQueryError(ErrorCode::InvalidValue, __FUNCTION__, "n cannot be negative.");
+            return;
+        }
+        if (!ids) {
+            return;
+        }
+        const std::lock_guard<std::mutex> lock(g_queryObjectsMutex);
+        for (GLsizei i = 0; i < n; ++i) {
+            const GLuint id = g_nextQueryId++;
+            auto* queryObject = new QueryObject;
+            queryObject->id = id;
+            queryObject->target = target;
+            queryObject->created = true;
+            g_liveQueryObjects[id] = queryObject;
+            ids[i] = id;
+        }
+    }
+
     void DeleteQueries(GLsizei n, const GLuint* ids) {
         if (n < 0) {
             RecordQueryError(ErrorCode::InvalidValue, __FUNCTION__, "n cannot be negative.");
@@ -180,7 +231,24 @@ namespace MobileGL::MG_Impl::GLImpl {
             }
             QueryObject* queryObject = it->second;
             if (queryObject->active) {
-                EndTimeElapsedQueryLocked(queryObject); // implicitly end before deletion
+                // Implicitly end before deletion, releasing the matching active slot.
+                if (queryObject->target == GL_SAMPLES_PASSED || queryObject->target == GL_ANY_SAMPLES_PASSED ||
+                    queryObject->target == GL_ANY_SAMPLES_PASSED_CONSERVATIVE) {
+                    if (const auto endOcclusionQuery = MG_Backend::gBackendFunctionsTable.GL.EndOcclusionQuery;
+                        endOcclusionQuery && queryObject->backendHandle) {
+                        endOcclusionQuery(queryObject->backendHandle);
+                    }
+                    queryObject->active = false;
+                    g_activeSamplesPassedQueryId = 0;
+                } else if (queryObject->target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN ||
+                           queryObject->target == GL_PRIMITIVES_GENERATED) {
+                    queryObject->active = false;
+                    (queryObject->target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN
+                         ? g_activePrimitivesWrittenQueryId
+                         : g_activePrimitivesGeneratedQueryId) = 0;
+                } else {
+                    EndTimeElapsedQueryLocked(queryObject);
+                }
             }
             if (queryObject->backendHandle) {
                 if (const auto deleteBackendQuery = MG_Backend::gBackendFunctionsTable.GL.DeleteBackendQuery) {
@@ -198,16 +266,23 @@ namespace MobileGL::MG_Impl::GLImpl {
             return GL_FALSE;
         }
         const std::lock_guard<std::mutex> lock(g_queryObjectsMutex);
-        // Gen'd ids count as query objects here: the registry creates live
-        // objects at GenQueries time.
-        return FindQueryObjectLocked(id) != nullptr ? GL_TRUE : GL_FALSE;
+        // A name from glGenQueries is not yet a query object: it becomes one when it is first
+        // used with BeginQuery/QueryCounter (which is what a non-zero target records), or
+        // immediately if it came from glCreateQueries.
+        const auto* queryObject = FindQueryObjectLocked(id);
+        return (queryObject != nullptr && (queryObject->created || queryObject->target != 0)) ? GL_TRUE : GL_FALSE;
     }
 
     void BeginQuery(GLenum target, GLuint id) {
-        if (target != GL_TIME_ELAPSED) {
-            // Only GL_TIME_ELAPSED timer queries are implemented (occlusion and
-            // primitive queries remain stubs); GL_TIMESTAMP is not a valid
-            // BeginQuery target either.
+        const Bool isTransformFeedbackQuery =
+            target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN || target == GL_PRIMITIVES_GENERATED;
+        const Bool isOcclusionQuery =
+            (target == GL_SAMPLES_PASSED || target == GL_ANY_SAMPLES_PASSED ||
+             target == GL_ANY_SAMPLES_PASSED_CONSERVATIVE) &&
+            MG_Backend::gBackendFunctionsTable.GL.BeginOcclusionQuery != nullptr;
+        if (target != GL_TIME_ELAPSED && !isTransformFeedbackQuery && !isOcclusionQuery) {
+            // GL_TIMESTAMP is not a valid BeginQuery target; the occlusion targets
+            // need backend support.
             RecordQueryError(ErrorCode::InvalidEnum, __FUNCTION__, "Query target is not supported.");
             return;
         }
@@ -221,9 +296,13 @@ namespace MobileGL::MG_Impl::GLImpl {
             RecordQueryError(ErrorCode::InvalidOperation, __FUNCTION__, "Query object does not exist.");
             return;
         }
-        if (g_activeTimeElapsedQueryId != 0) {
+        GLuint& activeQueryId = isTransformFeedbackQuery
+            ? (target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN ? g_activePrimitivesWrittenQueryId
+                                                                  : g_activePrimitivesGeneratedQueryId)
+            : (isOcclusionQuery ? g_activeSamplesPassedQueryId : g_activeTimeElapsedQueryId);
+        if (activeQueryId != 0) {
             RecordQueryError(ErrorCode::InvalidOperation, __FUNCTION__,
-                             "A query is already active on GL_TIME_ELAPSED.");
+                             "A query is already active on this target.");
             return;
         }
         if (queryObject->active) {
@@ -239,25 +318,72 @@ namespace MobileGL::MG_Impl::GLImpl {
         ResetQueryObjectLocked(queryObject); // discard any previous result
         queryObject->target = target;
         queryObject->active = true;
-        const auto beginTimeElapsedQuery = MG_Backend::gBackendFunctionsTable.GL.BeginTimeElapsedQuery;
-        queryObject->backendHandle =
-            (!TimerQueryDisabled() && beginTimeElapsedQuery) ? beginTimeElapsedQuery() : nullptr;
-        g_activeTimeElapsedQueryId = id;
+        if (isTransformFeedbackQuery) {
+            // Prefer real GPU transform-feedback queries (exact with geometry shaders);
+            // the CPU accounting delta stays as the fallback when the backend lacks them.
+            const auto beginXfbPrimitivesQuery = MG_Backend::gBackendFunctionsTable.GL.BeginXfbPrimitivesQuery;
+            queryObject->backendHandle =
+                beginXfbPrimitivesQuery ? beginXfbPrimitivesQuery(target == GL_PRIMITIVES_GENERATED) : nullptr;
+            queryObject->counterSnapshot = MG_State::pGLContext->GetTransformFeedbackPrimitiveCounter();
+        } else if (isOcclusionQuery) {
+            queryObject->backendHandle = MG_Backend::gBackendFunctionsTable.GL.BeginOcclusionQuery();
+        } else {
+            const auto beginTimeElapsedQuery = MG_Backend::gBackendFunctionsTable.GL.BeginTimeElapsedQuery;
+            queryObject->backendHandle =
+                (!TimerQueryDisabled() && beginTimeElapsedQuery) ? beginTimeElapsedQuery() : nullptr;
+        }
+        activeQueryId = id;
     }
 
     void EndQuery(GLenum target) {
-        if (target != GL_TIME_ELAPSED) {
+        const Bool isTransformFeedbackQuery =
+            target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN || target == GL_PRIMITIVES_GENERATED;
+        const Bool isOcclusionQuery =
+            (target == GL_SAMPLES_PASSED || target == GL_ANY_SAMPLES_PASSED ||
+             target == GL_ANY_SAMPLES_PASSED_CONSERVATIVE) &&
+            MG_Backend::gBackendFunctionsTable.GL.BeginOcclusionQuery != nullptr;
+        if (target != GL_TIME_ELAPSED && !isTransformFeedbackQuery && !isOcclusionQuery) {
             RecordQueryError(ErrorCode::InvalidEnum, __FUNCTION__, "Query target is not supported.");
             return;
         }
         const std::lock_guard<std::mutex> lock(g_queryObjectsMutex);
-        if (g_activeTimeElapsedQueryId == 0) {
-            RecordQueryError(ErrorCode::InvalidOperation, __FUNCTION__, "No query is active on GL_TIME_ELAPSED.");
+        GLuint& activeQueryId = isTransformFeedbackQuery
+            ? (target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN ? g_activePrimitivesWrittenQueryId
+                                                                  : g_activePrimitivesGeneratedQueryId)
+            : (isOcclusionQuery ? g_activeSamplesPassedQueryId : g_activeTimeElapsedQueryId);
+        if (activeQueryId == 0) {
+            RecordQueryError(ErrorCode::InvalidOperation, __FUNCTION__, "No query is active on this target.");
             return;
         }
-        auto* queryObject = FindQueryObjectLocked(g_activeTimeElapsedQueryId);
+        auto* queryObject = FindQueryObjectLocked(activeQueryId);
         if (!queryObject) {
-            g_activeTimeElapsedQueryId = 0; // should not happen; keep state consistent
+            activeQueryId = 0; // should not happen; keep state consistent
+            return;
+        }
+        if (isTransformFeedbackQuery) {
+            if (queryObject->backendHandle) {
+                if (const auto endXfbPrimitivesQuery = MG_Backend::gBackendFunctionsTable.GL.EndXfbPrimitivesQuery) {
+                    endXfbPrimitivesQuery(queryObject->backendHandle);
+                }
+                // Result comes from the GPU query at read time.
+            } else {
+                queryObject->cachedResult =
+                    MG_State::pGLContext->GetTransformFeedbackPrimitiveCounter() - queryObject->counterSnapshot;
+                queryObject->resultCached = true;
+            }
+            queryObject->active = false;
+            queryObject->ended = true;
+            activeQueryId = 0;
+            return;
+        }
+        if (isOcclusionQuery) {
+            if (const auto endOcclusionQuery = MG_Backend::gBackendFunctionsTable.GL.EndOcclusionQuery;
+                endOcclusionQuery && queryObject->backendHandle) {
+                endOcclusionQuery(queryObject->backendHandle);
+            }
+            queryObject->active = false;
+            queryObject->ended = true;
+            activeQueryId = 0;
             return;
         }
         EndTimeElapsedQueryLocked(queryObject);
@@ -303,9 +429,25 @@ namespace MobileGL::MG_Impl::GLImpl {
         switch (pname) {
         case GL_CURRENT_QUERY: {
             const std::lock_guard<std::mutex> lock(g_queryObjectsMutex);
-            // Only GL_TIME_ELAPSED queries can be active; GL_TIMESTAMP queries
-            // never are, and other targets remain unimplemented.
-            *params = target == GL_TIME_ELAPSED ? static_cast<GLint>(g_activeTimeElapsedQueryId) : 0;
+            switch (target) {
+            case GL_TIME_ELAPSED:
+                *params = static_cast<GLint>(g_activeTimeElapsedQueryId);
+                break;
+            case GL_SAMPLES_PASSED:
+            case GL_ANY_SAMPLES_PASSED:
+            case GL_ANY_SAMPLES_PASSED_CONSERVATIVE:
+                *params = static_cast<GLint>(g_activeSamplesPassedQueryId);
+                break;
+            case GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN:
+                *params = static_cast<GLint>(g_activePrimitivesWrittenQueryId);
+                break;
+            case GL_PRIMITIVES_GENERATED:
+                *params = static_cast<GLint>(g_activePrimitivesGeneratedQueryId);
+                break;
+            default:
+                *params = 0;
+                break;
+            }
             return;
         }
         case GL_QUERY_COUNTER_BITS: {
@@ -313,7 +455,13 @@ namespace MobileGL::MG_Impl::GLImpl {
             // time: IsTimerQuerySupported is the dynamic truth (extension /
             // entry points / timestamp valid bits at call time, not at table
             // init), and the MOBILEGL_DISABLE_TIMERQUERY kill switch always
-            // wins. Non-timer targets remain unimplemented and report 0.
+            // wins.
+            if (target == GL_SAMPLES_PASSED || target == GL_ANY_SAMPLES_PASSED ||
+                target == GL_ANY_SAMPLES_PASSED_CONSERVATIVE) {
+                const Bool occlusionSupported = MG_Backend::gBackendFunctionsTable.GL.BeginOcclusionQuery != nullptr;
+                *params = occlusionSupported ? (target == GL_SAMPLES_PASSED ? 32 : 1) : 0;
+                return;
+            }
             const Bool timerTarget = target == GL_TIME_ELAPSED || target == GL_TIMESTAMP;
             const auto isTimerQuerySupported = MG_Backend::gBackendFunctionsTable.GL.IsTimerQuerySupported;
             const Bool supported =
@@ -359,4 +507,42 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
         *params = static_cast<GLuint64>(value);
     }
+
+    namespace {
+        // The indexed query entry points differ from the plain ones only in the vertex
+        // stream they address (GL 4.6 core 4.2.1): index must be below GL_MAX_VERTEX_STREAMS
+        // for the two transform feedback targets and zero for every other target. With a
+        // single vertex stream both bounds are 1, so a valid call is always index 0 and
+        // forwards to the unindexed implementation.
+        Bool ValidateQueryStreamIndex(const char* function, GLenum target, GLuint index) {
+            const Bool perStreamTarget =
+                target == GL_PRIMITIVES_GENERATED || target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN;
+            GLint maxVertexStreams = 1;
+            if (perStreamTarget) {
+                GetIntegerv(GL_MAX_VERTEX_STREAMS, &maxVertexStreams);
+            }
+            if (index < static_cast<GLuint>(std::max(maxVertexStreams, 1))) {
+                return true;
+            }
+            RecordQueryError(ErrorCode::InvalidValue, function,
+                             perStreamTarget ? "index is not less than GL_MAX_VERTEX_STREAMS."
+                                             : "index must be zero for this query target.");
+            return false;
+        }
+    } // namespace
+
+    void BeginQueryIndexed(GLenum target, GLuint index, GLuint id) {
+        if (!ValidateQueryStreamIndex(__FUNCTION__, target, index)) return;
+        BeginQuery(target, id);
+    }
+
+    void EndQueryIndexed(GLenum target, GLuint index) {
+        if (!ValidateQueryStreamIndex(__FUNCTION__, target, index)) return;
+        EndQuery(target);
+    }
+
+    void GetQueryIndexediv(GLenum target, GLuint index, GLenum pname, GLint* params) {
+        if (!ValidateQueryStreamIndex(__FUNCTION__, target, index)) return;
+        GetQueryiv(target, pname, params);
+    }
 } // namespace MobileGL::MG_Impl::GLImpl

MobileGL/MG_Impl/GLImpl/Query/GL_Query.h

@@ -11,11 +11,15 @@
 
 namespace MobileGL::MG_Impl::GLImpl {
     void GenQueries(GLsizei n, GLuint* ids);
+    void CreateQueries(GLenum target, GLsizei n, GLuint* ids);
     void DeleteQueries(GLsizei n, const GLuint* ids);
     GLboolean IsQuery(GLuint id);
     void BeginQuery(GLenum target, GLuint id);
     void EndQuery(GLenum target);
     void GetQueryiv(GLenum target, GLenum pname, GLint* params);
+    void BeginQueryIndexed(GLenum target, GLuint index, GLuint id);
+    void EndQueryIndexed(GLenum target, GLuint index);
+    void GetQueryIndexediv(GLenum target, GLuint index, GLenum pname, GLint* params);
     void GetQueryObjectiv(GLuint id, GLenum pname, GLint* params);
     void GetQueryObjectuiv(GLuint id, GLenum pname, GLuint* params);
     void GetQueryObjecti64v(GLuint id, GLenum pname, GLint64* params);

MobileGL/MG_Impl/GLImpl/Sampler/GL_Sampler.cpp

@@ -185,6 +185,11 @@ namespace MobileGL::MG_Impl::GLImpl {
         static thread_local Vector<GLuint> names;
         MG_State::pGLContext->GenSamplerNames(count, names);
         Memcpy(samplers, names.data(), count * sizeof(GLuint));
+        // Unlike textures/buffers, glGenSamplers CREATES the sampler objects: each name
+        // is immediately a sampler (glIsSampler == GL_TRUE before any bind).
+        for (GLsizei i = 0; i < count; ++i) {
+            MG_State::pGLContext->CreateSamplerObject(names[i]);
+        }
     }
 
     void DeleteSamplers_State(GLsizei count, const GLuint* samplers) {

MobileGL/MG_Impl/GLImpl/Sync/GL_Sync.cpp

@@ -133,4 +133,33 @@ namespace MobileGL::MG_Impl::GLImpl {
             values[0] = value;
         }
     }
+
+    void DestroyAllSyncObjects() {
+        // Detach the registry under the lock, release outside it. Entries the app
+        // already deleted were erased by DeleteSync, so nothing here double-frees;
+        // a DeleteSync racing this sweep finds an empty registry and returns. A
+        // thread still blocked inside ClientWaitSync/GetSynciv during teardown
+        // holds a raw SyncObject* these deletes invalidate - the same undefined
+        // race an app-driven DeleteSync already has.
+        UnorderedMap<GLsync, SyncObject*> orphans;
+        {
+            const std::lock_guard<std::mutex> lock(g_syncObjectsMutex);
+            orphans.swap(g_liveSyncObjects);
+        }
+        if (orphans.empty()) {
+            return;
+        }
+        // Both backends' DeleteSync only free the heap wrapper once their GL
+        // context/renderer is gone (generation/current-thread guards), so this is
+        // safe after the backend has released its EGL resources - but not after
+        // the function table itself is cleared.
+        const auto backendDeleteSync = MG_Backend::gBackendFunctionsTable.GL.DeleteSync;
+        for (const auto& [_, syncObject] : orphans) {
+            if (backendDeleteSync && syncObject->backendHandle) {
+                backendDeleteSync(syncObject->backendHandle);
+            }
+            delete syncObject;
+        }
+        MGLOG_D("DestroyAllSyncObjects: reclaimed %zu sync object(s) the app left undeleted", orphans.size());
+    }
 } // namespace MobileGL::MG_Impl::GLImpl

MobileGL/MG_Impl/GLImpl/Sync/GL_Sync.h

@@ -16,4 +16,12 @@ namespace MobileGL::MG_Impl::GLImpl {
     void WaitSync(GLsync sync, GLbitfield flags, GLuint64 timeout);
     void DeleteSync(GLsync sync);
     void GetSynciv(GLsync sync, GLenum pname, GLsizei bufSize, GLsizei* length, GLint* values);
+    // Destroys every still-registered sync object exactly as DeleteSync would.
+    // GL requires syncs to die with their context; called only from full library
+    // teardown (DestroyImpl), where no context survives on any thread, so the
+    // process-global registry can be drained wholesale. Must run while the
+    // backend function table is still populated: each backend handle has to be
+    // released by the backend that created it, never by a later re-initialized
+    // one.
+    void DestroyAllSyncObjects();
 } // namespace MobileGL::MG_Impl::GLImpl

MobileGL/MG_Impl/GLImpl/Texture/GL_Texture.cpp

@@ -2035,6 +2035,128 @@ namespace MobileGL::MG_Impl::GLImpl {
         MaybeAutoGenerateMipmap(target, textureObject, isProxy, level);
     }
 
+    // The work glTexBuffer[Range] and glTextureBuffer[Range] all share once the texture has been
+    // resolved - by binding for the target forms, by name for the DSA ones. `size` is
+    // kWholeBuffer for the non-Range entry points, which attach the buffer as it grows rather
+    // than freezing the size it happens to have now.
+    // The sized internal formats a buffer texture accepts (GL 4.6 core table 8.16). This is a much
+    // shorter list than the renderable or texturable formats, so it cannot be inferred from either.
+    static Bool IsBufferTextureInternalFormat(GLenum internalformat) {
+        switch (internalformat) {
+        case GL_R8:
+        case GL_R16:
+        case GL_R16F:
+        case GL_R32F:
+        case GL_R8I:
+        case GL_R16I:
+        case GL_R32I:
+        case GL_R8UI:
+        case GL_R16UI:
+        case GL_R32UI:
+        case GL_RG8:
+        case GL_RG16:
+        case GL_RG16F:
+        case GL_RG32F:
+        case GL_RG8I:
+        case GL_RG16I:
+        case GL_RG32I:
+        case GL_RG8UI:
+        case GL_RG16UI:
+        case GL_RG32UI:
+        case GL_RGB32F:
+        case GL_RGB32I:
+        case GL_RGB32UI:
+        case GL_RGBA8:
+        case GL_RGBA16:
+        case GL_RGBA16F:
+        case GL_RGBA32F:
+        case GL_RGBA8I:
+        case GL_RGBA16I:
+        case GL_RGBA32I:
+        case GL_RGBA8UI:
+        case GL_RGBA16UI:
+        case GL_RGBA32UI:
+            return true;
+        default:
+            return false;
+        }
+    }
+
+    static void AttachBufferToTexture(const SharedPtr<MG_State::GLState::ITextureObject>& textureObject,
+                                      GLenum internalformat, GLuint buffer, GLintptr offset, SizeT size,
+                                      const char* caller) {
+        using MG_State::GLState::TextureObjectBuffer;
+        TextureInternalFormat textureInternalFormat = MG_Util::ConvertGLEnumToTextureInternalFormat(internalformat);
+        if (!IsBufferTextureInternalFormat(internalformat)) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>(
+                    "MG_Impl/GLImpl", caller,
+                    std::format("internalformat 0x{:X} is not one of the sized formats a buffer texture accepts.",
+                                internalformat)));
+            return;
+        }
+        if (!TextureImpl::ValidateTextureInternalFormat(textureInternalFormat)) return;
+
+        auto& bufferObject = MG_State::pGLContext->GetBufferObject(buffer);
+        if (buffer != 0 && !bufferObject) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller,
+                                             "`buffer` is not zero and is not the name of an existing buffer object."));
+            return;
+        }
+        if (!TextureImpl::ValidateTextureObject(textureObject)) return;
+        if (textureObject->GetStorageType() != TextureStorageType::Buffer) {
+            // A texture whose target is something else is a wrong object, not a wrong token
+            // (GL 4.6 core 8.9).
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller,
+                                             "The effective target of `texture` is not `GL_TEXTURE_BUFFER`."));
+            return;
+        }
+        if (size != TextureObjectBuffer::kWholeBuffer) {
+            // GL 4.6 core 8.9: offset must be non-negative and aligned to
+            // GL_TEXTURE_BUFFER_OFFSET_ALIGNMENT, and size must be positive.
+            if (offset < 0) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidValue,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller, "offset must be non-negative."));
+                return;
+            }
+            if (size == 0) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidValue,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller, "size must be greater than zero."));
+                return;
+            }
+            const Int alignment = std::max(
+                1, MG_Backend::pActiveBackendObject->GetDynamicParameters().TextureBufferOffsetAlignment);
+            if (offset % alignment != 0) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidValue,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller,
+                                                 "offset is not a multiple of GL_TEXTURE_BUFFER_OFFSET_ALIGNMENT."));
+                return;
+            }
+            // The range has to lie inside the buffer that is being attached. Detaching (buffer
+            // zero) carries no range to check.
+            if (bufferObject && static_cast<SizeT>(offset) + size > bufferObject->GetSize()) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidValue,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller,
+                                                 "offset + size is greater than the buffer object's GL_BUFFER_SIZE."));
+                return;
+            }
+        }
+
+        auto* texBufferObject = static_cast<TextureObjectBuffer*>(textureObject.get());
+        texBufferObject->GetBufferBindingSlot().Bind(bufferObject);
+        texBufferObject->SetBufferRange(static_cast<SizeT>(offset < 0 ? 0 : offset), size);
+        texBufferObject->SetInternalFormat(textureInternalFormat);
+    }
+
     void TexBuffer_State(GLenum target, GLenum internalformat, GLuint buffer) {
         // ======================= Converting ================================
         TextureTarget textureTarget = MG_Util::ConvertGLEnumToTextureTarget(target);
@@ -2081,6 +2203,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         auto& bufferSlot = texBufferObject->GetBufferBindingSlot();
         bufferSlot.Bind(bufferObject);
 
+        texBufferObject->SetBufferRange(0, MG_State::GLState::TextureObjectBuffer::kWholeBuffer);
         texBufferObject->SetInternalFormat(textureInternalFormat);
     }
 
@@ -2820,7 +2943,22 @@ namespace MobileGL::MG_Impl::GLImpl {
                                          "The attachment specified by the read buffer is incomplete."));               \
         return false;                                                                                                  \
     }
-        if (isDepth) {
+        if (isDepth && isStencil) {
+            // A combined internalformat copies both halves, so the read framebuffer
+            // must populate both attachment points.
+            const auto& stencilAttachment = currentReadFBO->GetAttachment(FramebufferAttachmentType::Stencil);
+            const auto& depthAttachment = currentReadFBO->GetAttachment(FramebufferAttachmentType::Depth);
+            if (!depthAttachment.IsValid() || depthAttachment.IsEmpty() || !stencilAttachment.IsValid() ||
+                stencilAttachment.IsEmpty()) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>(
+                        "MG_Impl/GLImpl", "CopyTexImage2D_State",
+                        "DEPTH_STENCIL copy requires both depth and stencil attachments in the read framebuffer."));
+                return false;
+            }
+            GET_SRC_INTERNAL_FORMAT(FramebufferAttachmentType::Depth);
+        } else if (isDepth) {
             GET_SRC_INTERNAL_FORMAT(FramebufferAttachmentType::Depth);
         } else if (isStencil) {
             GET_SRC_INTERNAL_FORMAT(FramebufferAttachmentType::Stencil);
@@ -3655,23 +3793,43 @@ namespace MobileGL::MG_Impl::GLImpl {
         GetTextureImage(texture, level, format, type, bufSize, pixels);
     }
 
+    // A buffer texture carries none of the sampler or level state these queries report. Reached by
+    // name there is no target token to blame, so the wrong object is INVALID_OPERATION rather than
+    // the INVALID_ENUM the target forms report for an unaccepted target (GL 4.6 core 8.11).
+    static Bool ValidateNamedTextureHasParameters(const SharedPtr<MG_State::GLState::ITextureObject>& textureObject,
+                                                  const char* caller) {
+        if (!textureObject) return false;
+        if (textureObject->GetStorageType() == TextureStorageType::Buffer) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller,
+                                             "The effective target of `texture` has no texture parameters."));
+            return false;
+        }
+        return true;
+    }
+
     void GetTextureParameteriv(GLuint texture, GLenum pname, GLint* params) {
         auto textureObject = GetTextureObjectByName(texture, __func__);
+        if (!ValidateNamedTextureHasParameters(textureObject, __func__)) return;
         WithTemporarilyBoundNamedTexture(textureObject, [&](GLenum target) { GetTexParameteriv_State(target, pname, params); });
     }
 
     void GetTextureParameterfv(GLuint texture, GLenum pname, GLfloat* params) {
         auto textureObject = GetTextureObjectByName(texture, __func__);
+        if (!ValidateNamedTextureHasParameters(textureObject, __func__)) return;
         WithTemporarilyBoundNamedTexture(textureObject, [&](GLenum target) { GetTexParameterfv_State(target, pname, params); });
     }
 
     void GetTextureParameterIiv(GLuint texture, GLenum pname, GLint* params) {
         auto textureObject = GetTextureObjectByName(texture, __func__);
+        if (!ValidateNamedTextureHasParameters(textureObject, __func__)) return;
         WithTemporarilyBoundNamedTexture(textureObject, [&](GLenum target) { GetTexParameterIiv_State(target, pname, params); });
     }
 
     void GetTextureParameterIuiv(GLuint texture, GLenum pname, GLuint* params) {
         auto textureObject = GetTextureObjectByName(texture, __func__);
+        if (!ValidateNamedTextureHasParameters(textureObject, __func__)) return;
         WithTemporarilyBoundNamedTexture(textureObject, [&](GLenum target) { GetTexParameterIuiv_State(target, pname, params); });
     }
 
@@ -4073,6 +4231,33 @@ namespace MobileGL::MG_Impl::GLImpl {
         TexBuffer_State(target, internalformat, buffer);
     }
 
+    // The buffer texture bound to `target` on the active unit - what the non-DSA range form
+    // operates on. Kept separate from TexBuffer_State because that one resolves the target
+    // itself and attaches the whole buffer.
+    static const SharedPtr<MG_State::GLState::ITextureObject>& GetBoundBufferTexture(GLenum target,
+                                                                                     const char* caller) {
+        TextureUploadTarget uploadTarget = MG_Util::ConvertGLEnumToTextureUploadTarget(target);
+        if (!TextureImpl::ValidateTextureUploadTarget(uploadTarget)) return nullTextureObject;
+        (void)caller;
+        auto& activeUnit = MG_State::pGLContext->GetTextureUnitObject(MG_State::pGLContext->GetActiveTextureUnit());
+        return activeUnit.GetBindingSlot(MG_Util::ConvertGLEnumToTextureTarget(target)).GetBoundObject();
+    }
+
+    void TexBufferRange(GLenum target, GLenum internalformat, GLuint buffer, GLintptr offset, GLsizeiptr size) {
+        AttachBufferToTexture(GetBoundBufferTexture(target, __func__), internalformat, buffer, offset,
+                              static_cast<SizeT>(size < 0 ? 0 : size), __func__);
+    }
+
+    void TextureBuffer(GLuint texture, GLenum internalformat, GLuint buffer) {
+        AttachBufferToTexture(GetTextureObjectByName(texture, __func__), internalformat, buffer, 0,
+                              MG_State::GLState::TextureObjectBuffer::kWholeBuffer, __func__);
+    }
+
+    void TextureBufferRange(GLuint texture, GLenum internalformat, GLuint buffer, GLintptr offset, GLsizeiptr size) {
+        AttachBufferToTexture(GetTextureObjectByName(texture, __func__), internalformat, buffer, offset,
+                              static_cast<SizeT>(size < 0 ? 0 : size), __func__);
+    }
+
     GLboolean IsTexture(GLuint texture) {
         return IsTexture_State(texture);
     }

MobileGL/MG_Impl/GLImpl/Texture/GL_Texture.h

@@ -42,6 +42,9 @@ namespace MobileGL::MG_Impl::GLImpl {
     void GenerateTextureMipmap(GLuint texture);
     void BindTextureUnit(GLuint unit, GLuint texture);
     void GetTextureImage(GLuint texture, GLint level, GLenum format, GLenum type, GLsizei bufSize, void* pixels);
+    void TexBufferRange(GLenum target, GLenum internalformat, GLuint buffer, GLintptr offset, GLsizeiptr size);
+    void TextureBuffer(GLuint texture, GLenum internalformat, GLuint buffer);
+    void TextureBufferRange(GLuint texture, GLenum internalformat, GLuint buffer, GLintptr offset, GLsizeiptr size);
     void GetTextureSubImage(GLuint texture, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width,
                             GLsizei height, GLsizei depth, GLenum format, GLenum type, GLsizei bufSize, void* pixels);
     void GetTextureParameterfv(GLuint texture, GLenum pname, GLfloat* params);

MobileGL/MG_Impl/GLImpl/VertexArray/GL_VertexArray.cpp

@@ -105,12 +105,45 @@ namespace MobileGL::MG_Impl::GLImpl {
             return pname == GL_CURRENT_VERTEX_ATTRIB;
         }
 
+        // The stride a pointer-style call gives its binding point: the argument when it is non-zero,
+        // otherwise the tightly packed element size (GL 4.6 core 10.3.2). A packed 2_10_10_10 or
+        // 10F_11F_11F attribute is one 32-bit word regardless of its component count.
+        static int EffectiveVertexStride(GLsizei stride, GLint size, GLenum type) {
+            if (stride != 0) return static_cast<int>(stride);
+            switch (type) {
+            case GL_INT_2_10_10_10_REV:
+            case GL_UNSIGNED_INT_2_10_10_10_REV:
+            case GL_UNSIGNED_INT_10F_11F_11F_REV:
+                return 4;
+            default:
+                break;
+            }
+            return static_cast<int>(size * MG_Util::GetGLTypeSize(type));
+        }
+
+        // glBindVertexBuffers / glVertexArrayVertexBuffers take a range of binding points, and a
+        // range that runs past the last one is INVALID_OPERATION rather than the INVALID_VALUE a
+        // single out-of-range index gets (GL 4.6 core 10.3.1).
+        static bool ValidateVertexBindingRange(GLuint first, GLsizei count, const char* funcName) {
+            if (count < 0) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidValue,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", funcName, "count must be non-negative."));
+                return false;
+            }
+            if (static_cast<Uint64>(first) + static_cast<Uint64>(count) >
+                VertexArrayImpl::GetMaxVertexAttribBindings()) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", funcName,
+                                                 "first + count exceeds GL_MAX_VERTEX_ATTRIB_BINDINGS."));
+                return false;
+            }
+            return true;
+        }
+
         static bool ValidateVertexBindingIndex(GLuint bindingindex, const char* funcName) {
-            // Bound by the same dynamic limit as attribute indices: the default attribute -> binding
-            // mapping is the identity, so a binding point the backend cannot address as an attribute
-            // would resolve into an attribute the backend must then reject on every draw. Real drivers
-            // likewise report MAX_VERTEX_ATTRIB_BINDINGS == MAX_VERTEX_ATTRIBS.
-            if (bindingindex >= VertexArrayImpl::GetMaxVertexAttribs()) {
+            if (bindingindex >= VertexArrayImpl::GetMaxVertexAttribBindings()) {
                 MG_State::pGLContext->RecordError(
                     ErrorCode::InvalidValue,
                     MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", funcName,
@@ -155,6 +188,17 @@ namespace MobileGL::MG_Impl::GLImpl {
 
     SharedPtr<MG_State::GLState::VertexArrayObject> GetNamedVertexArrayObject_State(GLuint vaobj,
                                                                                    const char* caller) {
+        // Name zero is not a vertex array object in a core profile: it names the default vertex
+        // array, which the by-name (direct state access) entry points never accept. MobileGL keeps a
+        // real object at index 0 for the compatibility paths, so the generic name validation below
+        // would otherwise let it through (GL 4.6 core 10.3.1).
+        if (vaobj == 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller,
+                                             "Vertex array name 0 is not a vertex array object."));
+            return nullptr;
+        }
         if (!VertexArrayImpl::ValidateVertexArrayName(vaobj)) return nullptr;
         if (!VertexArrayImpl::ValidateVertexArrayObject(vaobj)) return nullptr;
         return MG_State::pGLContext->GetVertexArrayObject(vaobj);
@@ -210,7 +254,7 @@ namespace MobileGL::MG_Impl::GLImpl {
 
         DataType dataType = MG_Util::ConvertGLEnumToDataType(type);
         // Integer path: never normalized, never BGRA/packed (the validator rejects those).
-        if (!VertexArrayImpl::ValidateVertexAttribFormat(index, size, dataType, false, stride, true)) return;
+        if (!VertexArrayImpl::ValidateVertexAttribFormat(index, size, type, dataType, false, stride, true)) return;
 
         auto& vao = MG_State::pGLContext->GetBoundVertexArray();
         if (!vao) {
@@ -227,6 +271,7 @@ namespace MobileGL::MG_Impl::GLImpl {
 
         vao->SetAttributeFormat(index, size, dataType, false, stride, offset, true, false);
         vao->BindAttributeBuffer(index, vbo);
+        vao->MirrorPointerIntoBinding(index, vbo, offset, EffectiveVertexStride(stride, size, type));
     }
 
     void VertexAttribPointer_State(GLuint index, GLint size, GLenum type, GLboolean normalized, GLsizei stride,
@@ -234,7 +279,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (!VertexArrayImpl::ValidateVertexAttributeIndex(index)) return;
 
         DataType dataType = MG_Util::ConvertGLEnumToDataType(type);
-        if (!VertexArrayImpl::ValidateVertexAttribFormat(index, size, dataType, normalized == GL_TRUE, stride, false))
+        if (!VertexArrayImpl::ValidateVertexAttribFormat(index, size, type, dataType, normalized == GL_TRUE, stride, false))
             return;
 
         auto& vao = MG_State::pGLContext->GetBoundVertexArray();
@@ -256,6 +301,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         const int effectiveSize = isBgra ? 4 : size;
         vao->SetAttributeFormat(index, effectiveSize, dataType, normalized, stride, offset, false, isBgra);
         vao->BindAttributeBuffer(index, vbo);
+        vao->MirrorPointerIntoBinding(index, vbo, offset, EffectiveVertexStride(stride, effectiveSize, type));
     }
 
     void BindVertexArray_State(GLuint array) {
@@ -359,6 +405,13 @@ namespace MobileGL::MG_Impl::GLImpl {
                 MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller, "offset and stride must be non-negative."));
             return;
         }
+        if (static_cast<Uint>(stride) > VertexArrayImpl::GetMaxVertexAttribStride()) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller,
+                                             "stride exceeds GL_MAX_VERTEX_ATTRIB_STRIDE."));
+            return;
+        }
         auto bufferObject = GetVertexArrayBufferObject_State(buffer, caller);
         if (buffer != 0 && !bufferObject) return;
 
@@ -376,6 +429,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                                         const GLintptr* offsets, const GLsizei* strides) {
         auto vao = GetNamedVertexArrayObject_State(vaobj, "VertexArrayVertexBuffers_State");
         if (!vao) return;
+        if (!ValidateVertexBindingRange(first, count, "VertexArrayVertexBuffers_State")) return;
         for (GLsizei i = 0; i < count; ++i) {
             if (!buffers) {
                 VertexBufferBinding_State(vao, first + i, 0, 0, 16, "VertexArrayVertexBuffers_State");
@@ -389,12 +443,36 @@ namespace MobileGL::MG_Impl::GLImpl {
     static void VertexAttribFormatSeparate_State(const SharedPtr<MG_State::GLState::VertexArrayObject>& vao,
                                                  GLuint attribindex, GLint size, GLenum type, GLboolean normalized,
                                                  GLuint relativeoffset, Bool isInteger, const char* caller) {
+        static_cast<void>(caller);
         if (!VertexArrayImpl::ValidateVertexAttributeIndex(attribindex)) return;
 
         DataType dataType = MG_Util::ConvertGLEnumToDataType(type);
-        if (!VertexArrayImpl::ValidateVertexAttribPointerParams(attribindex, size, dataType, 0)) return;
+        // The separate-format entry points take the same size/type rules as the pointer ones,
+        // GL_BGRA included, so they need the full format validation rather than the pointer-only
+        // subset - that one reports GL_BGRA as an out-of-range size.
+        if (!VertexArrayImpl::ValidateVertexAttribFormat(attribindex, size, type, dataType, normalized == GL_TRUE, 0,
+                                                         isInteger))
+            return;
+        if (!VertexArrayImpl::ValidateVertexAttribRelativeOffset(relativeoffset)) return;
 
-        vao->SetAttributeFormatSeparate(attribindex, size, dataType, normalized, isInteger, relativeoffset);
+        const Bool isBgra = (size == static_cast<GLint>(GL_BGRA));
+        vao->SetAttributeFormatSeparate(attribindex, isBgra ? 4 : size, dataType, normalized, isInteger,
+                                        relativeoffset, isBgra);
+    }
+
+    // The long (64-bit) attribute format. MobileGL has no 64-bit vertex attributes, so nothing is
+    // recorded; what the entry point owes the application is the parameter validation, which is
+    // observable through glGetError regardless of whether the format could be used in a draw.
+    static void VertexAttribLFormatSeparate_State(GLuint attribindex, GLint size, GLenum type,
+                                                  GLuint relativeoffset) {
+        if (!VertexArrayImpl::ValidateVertexAttributeIndex(attribindex)) return;
+        if (!VertexArrayImpl::ValidateVertexAttribLFormat(attribindex, size, type)) return;
+        if (!VertexArrayImpl::ValidateVertexAttribRelativeOffset(relativeoffset)) return;
+
+        MG_State::pGLContext->RecordError(
+            ErrorCode::InvalidOperation,
+            MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", "VertexAttribLFormat",
+                                         "64-bit vertex attributes are not supported."));
     }
 
     void VertexArrayAttribFormat_State(GLuint vaobj, GLuint attribindex, GLint size, GLenum type,
@@ -1044,6 +1122,83 @@ namespace MobileGL::MG_Impl::GLImpl {
         VertexArrayVertexBuffer_State(vaobj, bindingindex, buffer, offset, stride);
     }
 
+    // glGetVertexArrayiv reports exactly one thing (GL 4.6 core table 23.4): which buffer the
+    // named vertex array takes its indices from. Everything else about a vertex array is
+    // per-attribute and belongs to the indexed queries below.
+    void GetVertexArrayiv(GLuint vaobj, GLenum pname, GLint* param) {
+        auto vao = GetNamedVertexArrayObject_State(vaobj, __func__);
+        if (!vao || !param) return;
+        if (pname != GL_ELEMENT_ARRAY_BUFFER_BINDING) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "pname must be GL_ELEMENT_ARRAY_BUFFER_BINDING."));
+            return;
+        }
+        const auto& indexBuffer = vao->GetIndexBufferBindingSlot().GetBoundObject();
+        *param = indexBuffer ? static_cast<GLint>(indexBuffer->GetExternalIndex()) : 0;
+    }
+
+    void GetVertexArrayIndexediv(GLuint vaobj, GLuint index, GLenum pname, GLint* param) {
+        auto vao = GetNamedVertexArrayObject_State(vaobj, __func__);
+        if (!vao || !param) return;
+        if (!VertexArrayImpl::ValidateVertexAttributeIndex(index)) return;
+        const auto& attr = vao->GetAttribute(index);
+        switch (pname) {
+        case GL_VERTEX_ATTRIB_ARRAY_ENABLED:
+            *param = attr.Enabled ? GL_TRUE : GL_FALSE;
+            return;
+        case GL_VERTEX_ATTRIB_ARRAY_SIZE:
+            *param = static_cast<GLint>(attr.Size);
+            return;
+        case GL_VERTEX_ATTRIB_ARRAY_STRIDE:
+            *param = static_cast<GLint>(attr.Stride);
+            return;
+        case GL_VERTEX_ATTRIB_ARRAY_TYPE:
+            *param = static_cast<GLint>(MG_Util::ConvertDataTypeToGLEnum(attr.Type));
+            return;
+        case GL_VERTEX_ATTRIB_ARRAY_NORMALIZED:
+            *param = attr.Normalized ? GL_TRUE : GL_FALSE;
+            return;
+        case GL_VERTEX_ATTRIB_ARRAY_INTEGER:
+            *param = attr.IsInteger ? GL_TRUE : GL_FALSE;
+            return;
+        case GL_VERTEX_ATTRIB_ARRAY_LONG:
+            // 64-bit attributes are not supported, so no attribute is ever a long one.
+            *param = GL_FALSE;
+            return;
+        case GL_VERTEX_ATTRIB_ARRAY_DIVISOR:
+            *param = static_cast<GLint>(attr.Divisor);
+            return;
+        case GL_VERTEX_ATTRIB_RELATIVE_OFFSET:
+            *param = static_cast<GLint>(vao->GetAttributeRelativeOffset(index));
+            return;
+        default:
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "pname is not an accepted indexed vertex array query."));
+            return;
+        }
+    }
+
+    // Only GL_VERTEX_BINDING_OFFSET needs 64 bits. Its `index` names a vertex buffer binding
+    // point directly (GL 4.6 core 10.3.1), not an attribute - unlike every pname the 32-bit
+    // indexed query above accepts, which is why this one does not go through an attribute's
+    // binding index.
+    void GetVertexArrayIndexed64iv(GLuint vaobj, GLuint index, GLenum pname, GLint64* param) {
+        auto vao = GetNamedVertexArrayObject_State(vaobj, __func__);
+        if (!vao || !param) return;
+        if (!VertexArrayImpl::ValidateVertexAttributeIndex(index)) return;
+        if (pname != GL_VERTEX_BINDING_OFFSET) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "pname must be GL_VERTEX_BINDING_OFFSET."));
+            return;
+        }
+        *param = static_cast<GLint64>(vao->GetBindingPoint(index).Offset);
+    }
+
     void VertexArrayAttribFormat(GLuint vaobj, GLuint attribindex, GLint size, GLenum type, GLboolean normalized,
                                  GLuint relativeoffset) {
         VertexArrayAttribFormat_State(vaobj, attribindex, size, type, normalized, relativeoffset);
@@ -1076,6 +1231,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                            const GLsizei* strides) {
         auto vao = GetBoundVertexArrayOrError("BindVertexBuffers");
         if (!vao) return;
+        if (!ValidateVertexBindingRange(first, count, "BindVertexBuffers")) return;
         for (GLsizei i = 0; i < count; ++i) {
             if (!buffers) {
                 VertexBufferBinding_State(vao, first + i, 0, 0, 16, "BindVertexBuffers");
@@ -1100,6 +1256,18 @@ namespace MobileGL::MG_Impl::GLImpl {
                                          "VertexAttribIFormat");
     }
 
+    void VertexAttribLFormat(GLuint attribindex, GLint size, GLenum type, GLuint relativeoffset) {
+        auto vao = GetBoundVertexArrayOrError("VertexAttribLFormat");
+        if (!vao) return;
+        VertexAttribLFormatSeparate_State(attribindex, size, type, relativeoffset);
+    }
+
+    void VertexArrayAttribLFormat(GLuint vaobj, GLuint attribindex, GLint size, GLenum type, GLuint relativeoffset) {
+        auto vao = GetNamedVertexArrayObject_State(vaobj, "VertexArrayAttribLFormat");
+        if (!vao) return;
+        VertexAttribLFormatSeparate_State(attribindex, size, type, relativeoffset);
+    }
+
     void VertexAttribBinding(GLuint attribindex, GLuint bindingindex) {
         auto vao = GetBoundVertexArrayOrError("VertexAttribBinding");
         if (!vao) return;

MobileGL/MG_Impl/GLImpl/VertexArray/GL_VertexArray.h

@@ -92,9 +92,13 @@ namespace MobileGL::MG_Impl::GLImpl {
     void EnableVertexArrayAttrib(GLuint vaobj, GLuint index);
     void VertexArrayElementBuffer(GLuint vaobj, GLuint buffer);
     void VertexArrayVertexBuffer(GLuint vaobj, GLuint bindingindex, GLuint buffer, GLintptr offset, GLsizei stride);
+    void GetVertexArrayiv(GLuint vaobj, GLenum pname, GLint* param);
+    void GetVertexArrayIndexediv(GLuint vaobj, GLuint index, GLenum pname, GLint* param);
+    void GetVertexArrayIndexed64iv(GLuint vaobj, GLuint index, GLenum pname, GLint64* param);
     void VertexArrayAttribFormat(GLuint vaobj, GLuint attribindex, GLint size, GLenum type, GLboolean normalized,
                                  GLuint relativeoffset);
     void VertexArrayAttribIFormat(GLuint vaobj, GLuint attribindex, GLint size, GLenum type, GLuint relativeoffset);
+    void VertexArrayAttribLFormat(GLuint vaobj, GLuint attribindex, GLint size, GLenum type, GLuint relativeoffset);
     void VertexArrayAttribBinding(GLuint vaobj, GLuint attribindex, GLuint bindingindex);
     void VertexArrayBindingDivisor(GLuint vaobj, GLuint bindingindex, GLuint divisor);
     void VertexArrayVertexBuffers(GLuint vaobj, GLuint first, GLsizei count, const GLuint* buffers,
@@ -104,6 +108,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                            const GLsizei* strides);
     void VertexAttribFormat(GLuint attribindex, GLint size, GLenum type, GLboolean normalized, GLuint relativeoffset);
     void VertexAttribIFormat(GLuint attribindex, GLint size, GLenum type, GLuint relativeoffset);
+    void VertexAttribLFormat(GLuint attribindex, GLint size, GLenum type, GLuint relativeoffset);
     void VertexAttribBinding(GLuint attribindex, GLuint bindingindex);
     void VertexBindingDivisor(GLuint bindingindex, GLuint divisor);
     void VertexAttribDivisor(GLuint index, GLuint divisor);

MobileGL/MG_Impl/GLImpl/VertexArray/Validators.cpp

@@ -23,6 +23,18 @@ namespace MobileGL::MG_Impl::GLImpl::VertexArrayImpl {
         return std::min(static_cast<Uint>(backendLimit), capacity);
     }
 
+    Uint GetMaxVertexAttribBindings() {
+        return GetMaxVertexAttribs();
+    }
+
+    Uint GetMaxVertexAttribRelativeOffset() {
+        return 2047;
+    }
+
+    Uint GetMaxVertexAttribStride() {
+        return 2048;
+    }
+
     Bool ValidateVertexArrayName(Uint index) {
         Bool isValid = MG_State::pGLContext->ValidateVertexArrayName(index);
         if (!isValid) {
@@ -90,9 +102,31 @@ namespace MobileGL::MG_Impl::GLImpl::VertexArrayImpl {
         return true;
     }
 
-    Bool ValidateVertexAttribFormat(Uint index, GLint sizeRaw, DataType type, Bool normalized, Int stride,
-                                    Bool integerPath) {
+    Bool ValidateVertexAttribFormat(Uint index, GLint sizeRaw, GLenum glType, DataType type, Bool normalized,
+                                    Int stride, Bool integerPath) {
         constexpr const char* fn = "ValidateVertexAttribFormat";
+        // GL_UNSIGNED_INT_10F_11F_11F_REV is a three-component float-path-only packing that has no
+        // DataType of its own, so it has to be recognised by name before the conversion below turns
+        // it into Unknown and reports the wrong error (GL 4.6 core 10.3.2).
+        if (glType == GL_UNSIGNED_INT_10F_11F_11F_REV) {
+            if (integerPath) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidEnum,
+                    MakeUnique<GenericErrorInfo>(
+                        "MG_Impl/GLImpl", fn,
+                        std::format("GL_UNSIGNED_INT_10F_11F_11F_REV is not an integer-path type (attribute {}).",
+                                    index)));
+                return false;
+            }
+            if (sizeRaw != 3) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>(
+                        "MG_Impl/GLImpl", fn,
+                        std::format("GL_UNSIGNED_INT_10F_11F_11F_REV requires size 3 (attribute {}).", index)));
+                return false;
+            }
+        }
         if (type == DataType::Unknown) {
             MG_State::pGLContext->RecordError(
                 ErrorCode::InvalidEnum,
@@ -170,4 +204,40 @@ namespace MobileGL::MG_Impl::GLImpl::VertexArrayImpl {
         }
         return true;
     }
+
+    Bool ValidateVertexAttribLFormat(Uint index, GLint size, GLenum type) {
+        constexpr const char* fn = "ValidateVertexAttribLFormat";
+        if (size < 1 || size > 4) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>(
+                    "MG_Impl/GLImpl", fn,
+                    std::format("Invalid size {} for attribute {}. Must be 1-4.", size, index)));
+            return false;
+        }
+        // GL 4.6 core 10.3.2: the long form takes GL_DOUBLE and nothing else.
+        if (type != GL_DOUBLE) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>(
+                    "MG_Impl/GLImpl", fn,
+                    std::format("Type 0x{:X} is not GL_DOUBLE (attribute {}).", type, index)));
+            return false;
+        }
+        return true;
+    }
+
+    Bool ValidateVertexAttribRelativeOffset(Uint relativeOffset) {
+        const Uint limit = GetMaxVertexAttribRelativeOffset();
+        if (relativeOffset > limit) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>(
+                    "MG_Impl/GLImpl", "ValidateVertexAttribRelativeOffset",
+                    std::format("relativeoffset {} exceeds GL_MAX_VERTEX_ATTRIB_RELATIVE_OFFSET ({}).", relativeOffset,
+                                limit)));
+            return false;
+        }
+        return true;
+    }
 } // namespace MobileGL::MG_Impl::GLImpl::VertexArrayImpl

MobileGL/MG_Impl/GLImpl/VertexArray/Validators.h

@@ -15,6 +15,20 @@ namespace MobileGL::MG_Impl::GLImpl::VertexArrayImpl {
     // capacity). Falls back to the capacity when no backend is active (unit tests).
     Uint GetMaxVertexAttribs();
 
+    // GL_MAX_VERTEX_ATTRIB_BINDINGS. The default attribute -> binding mapping is the identity, so a
+    // binding point that cannot also be an attribute index would resolve into an attribute the
+    // backend has to reject on every draw; real drivers report the two limits equal as well.
+    Uint GetMaxVertexAttribBindings();
+
+    // GL_MAX_VERTEX_ATTRIB_RELATIVE_OFFSET. The relative offset is folded into the resolved
+    // attribute offset in the frontend and never reaches a backend limit, so this is the value the
+    // spec requires an implementation to support at minimum (GL 4.6 core table 23.63).
+    Uint GetMaxVertexAttribRelativeOffset();
+
+    // GL_MAX_VERTEX_ATTRIB_STRIDE. Like the relative offset above, the stride never reaches a
+    // backend limit of its own, so this is the spec minimum (GL 4.6 core table 23.63).
+    Uint GetMaxVertexAttribStride();
+
     Bool ValidateVertexArrayName(Uint index);
     Bool ValidateVertexArrayObject(Uint index);
     Bool ValidateVertexAttributeIndex(Uint index);
@@ -22,6 +36,13 @@ namespace MobileGL::MG_Impl::GLImpl::VertexArrayImpl {
     // Full glVertexAttribPointer / glVertexAttribIPointer format validation, including the packed
     // 2_10_10_10 types and GL_BGRA size. sizeRaw is the untranslated GL size (possibly GL_BGRA);
     // integerPath selects the glVertexAttribIPointer rules.
-    Bool ValidateVertexAttribFormat(Uint index, GLint sizeRaw, DataType type, Bool normalized, Int stride,
-                                    Bool integerPath);
+    Bool ValidateVertexAttribFormat(Uint index, GLint sizeRaw, GLenum glType, DataType type, Bool normalized,
+                                    Int stride, Bool integerPath);
+    // glVertexAttribLFormat / glVertexArrayAttribLFormat: the only accepted type is GL_DOUBLE and
+    // the size range is 1-4 (GL_BGRA is a float-path size). Separate from the function above
+    // because the long path shares none of its type or size rules.
+    Bool ValidateVertexAttribLFormat(Uint index, GLint size, GLenum type);
+    // Shared by every *Format entry point: INVALID_VALUE once relativeoffset leaves the range the
+    // implementation advertises.
+    Bool ValidateVertexAttribRelativeOffset(Uint relativeOffset);
 } // namespace MobileGL::MG_Impl::GLImpl::VertexArrayImpl

MobileGL/MG_Impl/GetProcAddress.cpp

@@ -85,6 +85,8 @@ namespace MobileGL::MG_Impl {
         GETPROC(CGLGetPixelFormat, name);
         GETPROC(CGLSetCurrentContext, name);
         GETPROC(CGLGetCurrentContext, name);
+        GETPROC(CGLSetVirtualScreen, name);
+        GETPROC(CGLGetVirtualScreen, name);
         GETPROC(CGLSetParameter, name);
         GETPROC(CGLGetParameter, name);
         GETPROC(CGLUpdateContext, name);

MobileGL/MG_Impl/NSOpenGLImpl/NSOpenGLImpl.cpp

@@ -29,10 +29,19 @@ namespace MobileGL::MG_Impl::NSOpenGLImpl {
         char kContextViewKey;
         char kContextLayerKey;
 
-        std::once_flag g_installOnce;
         IMP g_pixelFormatDealloc = nullptr;
         IMP g_contextDealloc = nullptr;
 
+        std::mutex& HookInstallMutex() {
+            static auto* mutex = new std::mutex();
+            return *mutex;
+        }
+
+        Bool& HooksInstalled() {
+            static auto* installed = new Bool(false);
+            return *installed;
+        }
+
         template <typename Fn>
         Fn ObjcMsgSend() {
             return reinterpret_cast<Fn>(objc_msgSend);
@@ -431,12 +440,12 @@ namespace MobileGL::MG_Impl::NSOpenGLImpl {
             method_setImplementation(method, replacement);
         }
 
-        void InstallHooksOnce() {
+        Bool InstallHooksOnce() {
             Class pixelFormatClass = objc_getClass("NSOpenGLPixelFormat");
             Class contextClass = objc_getClass("NSOpenGLContext");
             if (!pixelFormatClass || !contextClass) {
                 MGLOG_W("NSOpenGLImpl: NSOpenGL classes are not loaded; hooks not installed");
-                return;
+                return false;
             }
 
             ReplaceInstanceMethod(pixelFormatClass, "initWithAttributes:",
@@ -471,11 +480,34 @@ namespace MobileGL::MG_Impl::NSOpenGLImpl {
             ReplaceInstanceMethod(contextClass, "dealloc", reinterpret_cast<IMP>(ContextDealloc), &g_contextDealloc);
 
             MGLOG_I("NSOpenGLImpl hooks installed");
+            return true;
         }
     } // namespace
 
     void InstallHooks() {
-        std::call_once(g_installOnce, InstallHooksOnce);
+        const std::lock_guard<std::mutex> lock(HookInstallMutex());
+        if (!HooksInstalled()) {
+            // Do not permanently consume the install attempt when the OpenGL
+            // framework has not registered its Objective-C classes yet. The
+            // dyld bootstrap normally runs after framework dependencies, but
+            // an explicitly loaded/static-linked MobileGL can arrive earlier.
+            HooksInstalled() = InstallHooksOnce();
+        }
     }
 } // namespace MobileGL::MG_Impl::NSOpenGLImpl
+
+namespace {
+    // SDL's Cocoa backend creates NSOpenGLPixelFormat/NSOpenGLContext before
+    // its first dlsym("glGetString") or other MobileGL host-API call. Install
+    // only the lightweight Objective-C dispatch hooks while the injected dylib
+    // is loading so those first Cocoa objects are routed through CGLImpl. The
+    // hooked context constructor reaches EGLImpl::GetDisplay(), which performs
+    // the full, thread-safe MobileGL initialization outside this bootstrap.
+    //
+    // There is intentionally no matching destructor: backend teardown remains
+    // owned by the EGL lifecycle and process-exit globals remain leak-at-exit.
+    __attribute__((constructor)) void BootstrapNSOpenGLHooks() {
+        MobileGL::MG_Impl::NSOpenGLImpl::InstallHooks();
+    }
+} // namespace
 #endif

MobileGL/MG_State/GLState/BufferState/BufferObject.cpp

@@ -174,6 +174,21 @@ namespace MobileGL::MG_State::GLState {
         ++m_changeSerial;
     }
 
+    void BufferObject::MarkGpuWritten() {
+        m_gpuWritePending = true;
+    }
+
+    void BufferObject::SyncGpuWrites() {
+        if (!m_gpuWritePending) return;
+        // Cleared unconditionally: without a readback op the shadow can never catch up,
+        // and retrying on every subsequent read would only repeat the same no-op.
+        m_gpuWritePending = false;
+        if (m_size == 0 || g_bufferBackendOps == nullptr || g_bufferBackendOps->ReadbackFromGpu == nullptr) {
+            return;
+        }
+        g_bufferBackendOps->ReadbackFromGpu(*this);
+    }
+
     void BufferObject::UploadSubData(DataPtr data, SizeT atOffset) {
         MOBILEGL_ASSERT(!m_isMapped || (m_mappingAccess & BufferMappingAccessBit::Persistent),
                         "Cannot upload sub data while buffer is non-persistently mapped.");
@@ -204,11 +219,13 @@ namespace MobileGL::MG_State::GLState {
                         "Destination buffer copy out of bounds: dstOffset (%zu) + size (%zu) > m_size (%zu)", dstOffset,
                         size, m_size);
 
+        src->SyncGpuWrites();
         Memcpy(m_resource.Bytes() + dstOffset, src->m_resource.Bytes() + srcOffset, size);
         NotifyContentWrite(dstOffset, size);
     }
 
     void* BufferObject::AcquireMemory(Bool markMapped, Bool read, Bool write) {
+        SyncGpuWrites();
         if (markMapped) {
             m_isMapped = true;
             m_mappingAccess = (read ? BufferMappingAccessBit::Read : BufferMappingAccessBit::Null) |
@@ -231,10 +248,29 @@ namespace MobileGL::MG_State::GLState {
         return m_resource.Bytes();
     }
 
+    Bool BufferObject::EnsureGpuResidentStorage() {
+        if (m_resource.IsGpuResident()) {
+            return true;
+        }
+        if (m_size == 0 || g_bufferBackendOps == nullptr || g_bufferBackendOps->AcquirePersistentMap == nullptr) {
+            return false;
+        }
+        void* base = g_bufferBackendOps->AcquirePersistentMap(*this);
+        if (base == nullptr) {
+            return false;
+        }
+        m_resource.AdoptPersistentMap(base);
+        return true;
+    }
+
     void* BufferObject::AcquireMemoryRange(Range1D range, Flags<BufferMappingAccessBit> access) {
         MOBILEGL_ASSERT(range.end <= m_size && range.start <= range.end,
                         "AcquireMemoryRange out of bounds: range (%zu, %zu) exceeds m_size (%zu)", range.start,
                         range.end, m_size);
+        // The app is about to look at the bytes; a shader may have rewritten them since
+        // the shadow was last authoritative. Also needed for a write map without an
+        // invalidate bit, whose staging copy is seeded from the shadow.
+        SyncGpuWrites();
         m_isMapped = true;
         m_mappingAccess = access;
         m_mappedRange = range;

MobileGL/MG_State/GLState/BufferState/BufferObject.h

@@ -99,6 +99,13 @@ namespace MobileGL {
             // Must be idempotent: a second call for an already-backed buffer returns the
             // same base pointer.
             void* (*AcquirePersistentMap)(BufferObject& bufferObject) = nullptr;
+            // Pulls the backend's current contents for the whole buffer into the shadow
+            // (through WritebackFromBackend). Only ever called for a buffer the GPU may
+            // have written behind the frontend's back - a shader storage or atomic counter
+            // binding of a draw or dispatch - because nothing else can desynchronise the
+            // shadow. Backends that cannot read their storage back leave this null; the
+            // shadow then keeps its pre-dispatch bytes, which is the old behaviour.
+            void (*ReadbackFromGpu)(BufferObject& bufferObject) = nullptr;
         };
 
         // Registered by the active backend at init, cleared at shutdown.
@@ -133,6 +140,11 @@ namespace MobileGL {
 
             void* AcquireMemory(Bool markMapped, Bool read, Bool write);
             void* AcquireMemoryRange(Range1D range, Flags<BufferMappingAccessBit> access);
+            // Adopt backend host-visible coherent GPU storage as the source of truth
+            // (used for GPU-written targets like transform feedback capture, so
+            // MapBuffer/GetBufferSubData read real GPU results). No-op when already
+            // resident or when the backend declines.
+            Bool EnsureGpuResidentStorage();
             void ReleaseMemory();
             void FlushMemoryRange(SizeT offset, SizeT length);
 
@@ -144,6 +156,16 @@ namespace MobileGL {
             // backend op: the backend storage already holds these bytes.
             void WritebackFromBackend(DataPtr data, SizeT atOffset);
 
+            // A draw or dispatch just ran with this buffer bound where a shader can write
+            // it (shader storage / atomic counter). The next read has to reconcile with
+            // that: pull the bytes back, or - when the shadow already IS coherent GPU
+            // memory - wait for the work that wrote them to retire. Which of the two is
+            // the backend's business; the flag only says a GPU write is outstanding.
+            void MarkGpuWritten();
+            // Refreshes the shadow from the backend when a GPU write is outstanding. Called
+            // from every path that reads the shadow on the app's behalf.
+            void SyncGpuWrites();
+
             Bool IsMapped() const;
             Bool IsImmutableStorage() const;
             SizeT GetSize() const;
@@ -191,6 +213,8 @@ namespace MobileGL {
             Bool m_isImmutableStorage = false;
             GLbitfield m_storageFlags = 0;
             Uint64 m_changeSerial = 0;
+            // Set by MarkGpuWritten, cleared by SyncGpuWrites once the shadow is refreshed.
+            Bool m_gpuWritePending = false;
             Range1D m_mappedRange;
             Vector<Uint8> m_stagingData;
             Bool m_ownsStagingData;

MobileGL/MG_State/GLState/BufferState/BufferState.h

@@ -31,6 +31,9 @@ namespace MobileGL::MG_State::GLState {
         BindingSlot<BufferObject>& GetBindingSlot(BufferTarget target);
         // For glBindBufferBase / glBindBufferRange
         BindingSlotRange1D<BufferObject>& GetBindingPoint(BufferTarget target, Uint index);
+        const BindingSlotRange1D<BufferObject>& GetBindingPoint(BufferTarget target, Uint index) const {
+            return const_cast<BufferState*>(this)->GetBindingPoint(target, index);
+        }
         constexpr SizeT GetBindingPointCount(const BufferTarget target) const {
             auto it = std::find(BufferBindPointTargets.begin(), BufferBindPointTargets.end(), target);
             auto index = std::distance(BufferBindPointTargets.begin(), it);

MobileGL/MG_State/GLState/Core.cpp

@@ -241,6 +241,32 @@ namespace MobileGL::MG_State {
         }
 
         void GLContext::MarkTextureObjectForDeletion(Uint index) {
+            // GL 3.3 core 4.4.2: deleting a texture whose image is attached to the framebuffer
+            // that is currently bound acts as if FramebufferTexture* had been called with texture
+            // zero for every attachment point it occupied there. Framebuffers that are NOT bound
+            // keep the orphaned attachment, so only the bound ones are touched.
+            //
+            // Without this the framebuffer object goes on holding the deleted texture alive as its
+            // attachment, and a later read through that framebuffer returns the dead texture's
+            // contents rather than those of whatever the application put in its place - the name
+            // it deleted usually comes straight back from the next glGenTextures, so the two are
+            // indistinguishable from the outside (KHR-GL32.packed_pixels read a stale gradient).
+            if (const auto& textureObject = m_textureState.GetTextureObject(index)) {
+                for (SizeT targetIndex = 0; targetIndex < SizeT(FramebufferTarget::FramebufferTargetCount);
+                     ++targetIndex) {
+                    const auto& framebuffer =
+                        GetFramebufferBindingSlot(static_cast<FramebufferTarget>(targetIndex)).GetBoundObject();
+                    if (!framebuffer || framebuffer->IsDefaultFramebuffer()) {
+                        continue;
+                    }
+                    const auto& attachments = framebuffer->GetAllAttachmentObjects();
+                    for (SizeT i = 0; i < attachments.size(); ++i) {
+                        if (attachments[i].IsTexture() && attachments[i].GetTexture() == textureObject) {
+                            framebuffer->Detach(static_cast<FramebufferAttachmentType>(i));
+                        }
+                    }
+                }
+            }
             m_textureState.MarkTextureObjectForDeletion(index, IsRelaxedSemanticsActive());
         }
 
@@ -398,6 +424,14 @@ namespace MobileGL::MG_State {
             m_renderState.SetPointSize(size);
         }
 
+        void GLContext::SetPatchVertices(Uint vertices) {
+            m_renderState.SetPatchVertices(vertices);
+        }
+
+        Uint GLContext::GetPatchVertices() const {
+            return m_renderState.GetPatchVertices();
+        }
+
         Float GLContext::GetPointSize() const {
             return m_renderState.GetPointSize();
         }
@@ -719,6 +753,153 @@ namespace MobileGL::MG_State {
         Bool GLContext::ValidateRenderbufferObject(Uint index) const {
             return m_renderbufferState.ValidateRenderbufferObject(index);
         }
+
+        void GLContext::SaveBoundTransformFeedbackState() {
+            auto& object = m_transformFeedbackObjects[m_boundTransformFeedback];
+            for (Uint i = 0; i < MAX_TRANSFORM_FEEDBACK_BUFFERS; ++i) {
+                const auto& point = m_bufferState.GetBindingPoint(BufferTarget::TransformFeedback, i);
+                object.bindings[i] = {point.GetBoundObject(), point.GetRange(), point.HasExplicitRange()};
+            }
+            object.active = m_transformFeedbackActive;
+            object.paused = m_transformFeedbackPaused;
+            object.primitiveMode = m_transformFeedbackPrimitiveMode;
+            object.program = m_transformFeedbackProgram;
+            object.generation = m_transformFeedbackGeneration;
+            object.capturedVertices = m_transformFeedbackCapturedVertices;
+            object.inputPrimitives = m_transformFeedbackInputPrimitives;
+        }
+
+        void GLContext::RestoreBoundTransformFeedbackState() {
+            const auto& object = m_transformFeedbackObjects[m_boundTransformFeedback];
+            for (Uint i = 0; i < MAX_TRANSFORM_FEEDBACK_BUFFERS; ++i) {
+                auto& point = m_bufferState.GetBindingPoint(BufferTarget::TransformFeedback, i);
+                point.Bind(object.bindings[i].buffer);
+                if (object.bindings[i].buffer) {
+                    point.SetRange(object.bindings[i].range, object.bindings[i].hasExplicitRange);
+                } else {
+                    point.ClearRange();
+                }
+            }
+            m_transformFeedbackActive = object.active;
+            m_transformFeedbackPaused = object.paused;
+            m_transformFeedbackPrimitiveMode = object.primitiveMode;
+            m_transformFeedbackProgram = object.program;
+            // The generation identifies one capture span, and a span belongs to the object
+            // that opened it - a backend keys its append state on it, so switching objects
+            // has to bring the right one back.
+            m_transformFeedbackGeneration = object.generation;
+            m_transformFeedbackCapturedVertices = object.capturedVertices;
+            m_transformFeedbackInputPrimitives = object.inputPrimitives;
+        }
+
+        void GLContext::GenTransformFeedbackNames(Uint number, Vector<Uint>& ids) {
+            ids.resize(number);
+            if (number == 0) return;
+            m_transformFeedbackNames.Generate(number, ids.data());
+            // A generated name already denotes an object with the default state, so that a
+            // bind never has to distinguish "first use" from any later one.
+            for (const Uint id : ids) {
+                m_transformFeedbackObjects[id] = {};
+            }
+        }
+
+        Bool GLContext::ValidateTransformFeedbackName(Uint index) const {
+            return index == 0 || m_transformFeedbackNames.IsValid(index);
+        }
+
+        void GLContext::BindTransformFeedbackObject(Uint index) {
+            if (index == m_boundTransformFeedback) return;
+            SaveBoundTransformFeedbackState();
+            m_boundTransformFeedback = index;
+            m_transformFeedbackObjects[index].everBound = true;
+            RestoreBoundTransformFeedbackState();
+        }
+
+        Bool GLContext::IsTransformFeedbackObject(Uint index) const {
+            if (index == 0 || !m_transformFeedbackNames.IsValid(index)) return false;
+            const auto it = m_transformFeedbackObjects.find(index);
+            return it != m_transformFeedbackObjects.end() && it->second.everBound;
+        }
+
+        void GLContext::MarkTransformFeedbackObjectForDeletion(Uint index) {
+            if (index == 0 || !m_transformFeedbackNames.IsValid(index)) return;
+            // Deleting the bound object reverts to the default one (GL 4.6 core 13.2.1);
+            // its state is dropped rather than saved back into the dying object.
+            if (index == m_boundTransformFeedback) {
+                m_boundTransformFeedback = 0;
+                RestoreBoundTransformFeedbackState();
+            }
+            m_transformFeedbackObjects.erase(index);
+            m_transformFeedbackNames.Delete(index);
+        }
+
+        Uint64 GLContext::GetTransformFeedbackRecordedVertices(Uint index) const {
+            const auto it = m_transformFeedbackObjects.find(index);
+            return it == m_transformFeedbackObjects.end() ? 0 : it->second.recordedVertices;
+        }
+
+        Bool GLContext::HasTransformFeedbackCompletedSpan(Uint index) const {
+            const auto it = m_transformFeedbackObjects.find(index);
+            return it != m_transformFeedbackObjects.end() && it->second.hasCompletedSpan;
+        }
+
+        void GLContext::CreateTransformFeedbackObject(Uint index) {
+            // glCreateTransformFeedbacks has no bind step to infer existence from, so the name it
+            // hands out is already the name of an object (GL 4.6 core 13.2.1).
+            m_transformFeedbackObjects[index] = {};
+            m_transformFeedbackObjects[index].everBound = true;
+        }
+
+        Bool GLContext::IsNamedTransformFeedbackActive(Uint index) const {
+            if (index == m_boundTransformFeedback) return m_transformFeedbackActive;
+            const auto it = m_transformFeedbackObjects.find(index);
+            return it != m_transformFeedbackObjects.end() && it->second.active;
+        }
+
+        Bool GLContext::IsNamedTransformFeedbackPaused(Uint index) const {
+            if (index == m_boundTransformFeedback) return m_transformFeedbackPaused;
+            const auto it = m_transformFeedbackObjects.find(index);
+            return it != m_transformFeedbackObjects.end() && it->second.paused;
+        }
+
+        NamedTransformFeedbackBinding GLContext::GetNamedTransformFeedbackBinding(Uint index, Uint bufferIndex) const {
+            NamedTransformFeedbackBinding result;
+            if (bufferIndex >= MAX_TRANSFORM_FEEDBACK_BUFFERS) return result;
+            // The bound object's capture bindings live in the context's own binding points, not in
+            // the saved copy - that one is only written when the object is swapped out.
+            if (index == m_boundTransformFeedback) {
+                const auto& point = m_bufferState.GetBindingPoint(BufferTarget::TransformFeedback, bufferIndex);
+                result.Buffer = point.GetBoundObject();
+                result.Range = point.GetRange();
+                result.HasExplicitRange = point.HasExplicitRange();
+                return result;
+            }
+            const auto it = m_transformFeedbackObjects.find(index);
+            if (it == m_transformFeedbackObjects.end()) return result;
+            const auto& saved = it->second.bindings[bufferIndex];
+            result.Buffer = saved.buffer;
+            result.Range = saved.range;
+            result.HasExplicitRange = saved.hasExplicitRange;
+            return result;
+        }
+
+        void GLContext::SetNamedTransformFeedbackBinding(Uint index, Uint bufferIndex,
+                                                         const SharedPtr<BufferObject>& buffer, Range1D range,
+                                                         Bool hasExplicitRange) {
+            if (bufferIndex >= MAX_TRANSFORM_FEEDBACK_BUFFERS) return;
+            if (index == m_boundTransformFeedback) {
+                auto& point = m_bufferState.GetBindingPoint(BufferTarget::TransformFeedback, bufferIndex);
+                point.Bind(buffer);
+                if (buffer && hasExplicitRange) {
+                    point.SetRange(range, true);
+                } else {
+                    point.ClearRange();
+                }
+                return;
+            }
+            auto& object = m_transformFeedbackObjects[index];
+            object.bindings[bufferIndex] = {buffer, range, hasExplicitRange};
+        }
     } // namespace GLState
 
     // Leak-at-exit storage; see GlobalObjects.cpp.

MobileGL/MG_State/GLState/Core.h

@@ -45,6 +45,14 @@ namespace MobileGL {
             // translates the result into its own API call.
             VertexAttribTypeInfo ClassifyVertexAttribType(GLenum glType);
 
+            // One indexed capture binding of a transform feedback object, as the by-name queries
+            // report it. An empty Buffer means the binding point is unbound.
+            struct NamedTransformFeedbackBinding {
+                SharedPtr<BufferObject> Buffer;
+                Range1D Range{};
+                Bool HasExplicitRange = false;
+            };
+
             class GLContext {
             public:
                 GLContext() = default;
@@ -138,6 +146,8 @@ namespace MobileGL {
                 Float GetLineWidth() const;
                 void SetPointSize(Float size);
                 Float GetPointSize() const;
+                void SetPatchVertices(Uint vertices);
+                Uint GetPatchVertices() const;
                 void SetPolygonOffset(Float factor, Float units);
                 Float GetPolygonOffsetFactor() const;
                 Float GetPolygonOffsetUnits() const;
@@ -211,6 +221,103 @@ namespace MobileGL {
                 void SetScissorBox(IntVec4 box);      // x, y, width, height
                 const IntVec4& GetScissorBox() const; // x, y, width, height
 
+                // Transform feedback. The fields below are the state of the transform
+                // feedback object currently bound to GL_TRANSFORM_FEEDBACK; see the object
+                // block further down for how a bind swaps them.
+                void BeginTransformFeedback(GLenum primitiveMode, const SharedPtr<ProgramObject>& program) {
+                    m_transformFeedbackActive = true;
+                    m_transformFeedbackPaused = false;
+                    m_transformFeedbackPrimitiveMode = primitiveMode;
+                    m_transformFeedbackProgram = program;
+                    m_transformFeedbackGeneration = ++m_transformFeedbackNextGeneration;
+                    m_transformFeedbackCapturedVertices = 0;
+                    m_transformFeedbackInputPrimitives = 0;
+                }
+                void EndTransformFeedback() {
+                    m_transformFeedbackActive = false;
+                    m_transformFeedbackPaused = false;
+                    m_transformFeedbackProgram.reset();
+                    // What glDrawTransformFeedback on this object replays from now on.
+                    auto& object = m_transformFeedbackObjects[m_boundTransformFeedback];
+                    object.recordedVertices = m_transformFeedbackCapturedVertices;
+                    object.hasCompletedSpan = true;
+                }
+                Bool IsTransformFeedbackActive() const { return m_transformFeedbackActive; }
+                Bool IsTransformFeedbackPaused() const { return m_transformFeedbackPaused; }
+                void SetTransformFeedbackPaused(Bool paused) { m_transformFeedbackPaused = paused; }
+                GLenum GetTransformFeedbackPrimitiveMode() const { return m_transformFeedbackPrimitiveMode; }
+                const SharedPtr<ProgramObject>& GetTransformFeedbackProgram() const {
+                    return m_transformFeedbackProgram;
+                }
+                // Bumped on every BeginTransformFeedback; the backend uses it to
+                // distinguish "resume appending" from "fresh capture".
+                Uint64 GetTransformFeedbackGeneration() const { return m_transformFeedbackGeneration; }
+                // CPU-side primitive accounting for the transform feedback queries:
+                // every captured draw adds its primitive count (draws without a
+                // geometry stage write exactly what they generate).
+                void AddTransformFeedbackPrimitives(Uint64 primitives) {
+                    m_transformFeedbackPrimitiveCounter += primitives;
+                }
+                Uint64 GetTransformFeedbackPrimitiveCounter() const { return m_transformFeedbackPrimitiveCounter; }
+                // Primitives a draw assembled while the capture was paused. GL counts those in
+                // PRIMITIVES_GENERATED, but a backend that answers the query with its own
+                // transform feedback counter cannot see them - nothing was being captured.
+                void AddTransformFeedbackPausedPrimitives(Uint64 primitives) {
+                    m_transformFeedbackPausedPrimitiveCounter += primitives;
+                }
+                Uint64 GetTransformFeedbackPausedPrimitiveCounter() const {
+                    return m_transformFeedbackPausedPrimitiveCounter;
+                }
+                // Vertices already captured since BeginTransformFeedback (drives the
+                // buffer-capacity clamp on the primitives-written accounting).
+                void AddTransformFeedbackCapturedVertices(Uint64 vertices) {
+                    m_transformFeedbackCapturedVertices += vertices;
+                }
+                Uint64 GetTransformFeedbackCapturedVertices() const { return m_transformFeedbackCapturedVertices; }
+                // Raw assembled input primitives fed to the capture stage since Begin
+                // (pre-clamp; drives the GS strip capture-order fixup at EndTF).
+                void AddTransformFeedbackInputPrimitives(Uint64 primitives) {
+                    m_transformFeedbackInputPrimitives += primitives;
+                }
+                Uint64 GetTransformFeedbackInputPrimitives() const { return m_transformFeedbackInputPrimitives; }
+
+                // Transform feedback objects (ARB_transform_feedback2 / GL 4.0 core).
+                // The capture state above and the indexed GL_TRANSFORM_FEEDBACK_BUFFER
+                // binding points are object state, but the context keeps exactly one live
+                // copy of both so that every existing reader - the backends' per-draw sync,
+                // the drawing and getter paths - needs no notion of which object owns them.
+                // A bind therefore saves the live copy into the outgoing object and restores
+                // the incoming one's. Object 0 is the default object and always exists.
+                static constexpr Uint MAX_TRANSFORM_FEEDBACK_BUFFERS = 4;
+                void GenTransformFeedbackNames(Uint number, Vector<Uint>& ids);
+                // A name glGenTransformFeedbacks handed out and glDeleteTransformFeedbacks
+                // has not taken back. Name 0 is always valid.
+                Bool ValidateTransformFeedbackName(Uint index) const;
+                // What glIsTransformFeedback reports: a generated name only becomes the name
+                // of an object once it has been bound at least once (GL 4.6 core 13.2.1).
+                Bool IsTransformFeedbackObject(Uint index) const;
+                void BindTransformFeedbackObject(Uint index);
+                void MarkTransformFeedbackObjectForDeletion(Uint index);
+                Uint GetBoundTransformFeedbackName() const { return m_boundTransformFeedback; }
+                // Vertices the object captured in its last completed span; the vertex count
+                // glDrawTransformFeedback replays.
+                Uint64 GetTransformFeedbackRecordedVertices(Uint index) const;
+                // Whether the object has ever completed a capture span. glDrawTransformFeedback
+                // on an object that has not is INVALID_OPERATION, which a zero vertex count
+                // cannot express: an empty completed span is legal and draws nothing.
+                Bool HasTransformFeedbackCompletedSpan(Uint index) const;
+
+                // The by-name (direct state access) view. A named object that happens to be the
+                // bound one is answered from the live copy, since that is where its state actually
+                // is until a bind swaps it out.
+                void CreateTransformFeedbackObject(Uint index);
+                Bool IsNamedTransformFeedbackActive(Uint index) const;
+                Bool IsNamedTransformFeedbackPaused(Uint index) const;
+                NamedTransformFeedbackBinding GetNamedTransformFeedbackBinding(Uint index, Uint bufferIndex) const;
+                void SetNamedTransformFeedbackBinding(Uint index, Uint bufferIndex,
+                                                      const SharedPtr<BufferObject>& buffer, Range1D range,
+                                                      Bool hasExplicitRange);
+
                 // Framebuffer
                 void GenFramebufferNames(Uint number, Vector<Uint>& framebuffers);
                 const SharedPtr<FramebufferObject>& GetFramebufferObject(Uint index);
@@ -243,6 +350,46 @@ namespace MobileGL {
                 BufferState m_bufferState;
                 VertexArrayState m_vertexArrayState;
                 Array<CurrentVertexAttributeValue, VertexArrayObject::MAX_VERTEX_ATTRIBS> m_currentVertexAttributes{};
+                Bool m_transformFeedbackActive = false;
+                Bool m_transformFeedbackPaused = false;
+                GLenum m_transformFeedbackPrimitiveMode = GL_POINTS;
+                SharedPtr<ProgramObject> m_transformFeedbackProgram;
+                Uint64 m_transformFeedbackGeneration = 0;
+                // Source of the per-span ids above; never rolls back with an object switch.
+                Uint64 m_transformFeedbackNextGeneration = 0;
+                // Not object state: the transform feedback queries snapshot it at BeginQuery
+                // and take the delta at EndQuery, which spans whatever objects were used.
+                Uint64 m_transformFeedbackPrimitiveCounter = 0;
+                Uint64 m_transformFeedbackPausedPrimitiveCounter = 0;
+                Uint64 m_transformFeedbackCapturedVertices = 0;
+                Uint64 m_transformFeedbackInputPrimitives = 0;
+
+                // Everything a transform feedback object owns while it is NOT the bound one.
+                struct TransformFeedbackObjectState {
+                    struct SavedBufferBinding {
+                        SharedPtr<BufferObject> buffer;
+                        Range1D range;
+                        Bool hasExplicitRange = false;
+                    };
+                    Array<SavedBufferBinding, MAX_TRANSFORM_FEEDBACK_BUFFERS> bindings;
+                    Bool active = false;
+                    Bool paused = false;
+                    GLenum primitiveMode = GL_POINTS;
+                    SharedPtr<ProgramObject> program;
+                    Uint64 generation = 0;
+                    Uint64 capturedVertices = 0;
+                    Uint64 inputPrimitives = 0;
+                    Uint64 recordedVertices = 0;
+                    Bool hasCompletedSpan = false;
+                    Bool everBound = false;
+                };
+                void SaveBoundTransformFeedbackState();
+                void RestoreBoundTransformFeedbackState();
+                // operator[] materialises an entry with the default state on first touch, so
+                // the default object (name 0) needs no seeding here.
+                UnorderedMap<Uint, TransformFeedbackObjectState> m_transformFeedbackObjects;
+                IndexGenerator<Uint> m_transformFeedbackNames;
+                Uint m_boundTransformFeedback = 0;
                 TextureState m_textureState;
                 ProgramState m_programState;
                 RenderState m_renderState;

MobileGL/MG_State/GLState/ProgramState/ProgramObject.cpp

@@ -170,9 +170,265 @@ namespace MobileGL::MG_State::GLState {
         m_uniformNameMaxLength = 0;
         m_attribInNameMaxLength = 0;
         m_uniformBlockNameMaxLength = 0;
+        m_xfbVaryings.clear();
+        m_xfbStrides.clear();
+        m_xfbBufferMode = GL_INTERLEAVED_ATTRIBS;
+        m_xfbVaryingNameMaxLength = 0;
+        m_xfbNeedsScatteredCapture = false;
+        m_xfbPackedStride = 0;
+        m_gsInputPrimitive = GL_NONE;
         m_linkStatus = false;
     }
 
+    namespace {
+        // GL type enum for a vertex-stage output symbol captured by transform
+        // feedback. Covers the scalar/vector/matrix float+integer types transform
+        // feedback may legally capture in GL 3.3.
+        Bool ResolveXfbSymbolType(const glslang::TType& type, GLenum& outType, GLint& outArraySize,
+                                  Uint32& outBytesPerElement) {
+            outArraySize = type.isArray() ? type.getOuterArraySize() : 1;
+            const Int columns = type.isMatrix() ? type.getMatrixCols() : 1;
+            const Int components = type.isMatrix() ? type.getMatrixRows()
+                                                   : (type.isVector() ? type.getVectorSize() : 1);
+            const glslang::TBasicType basic = type.getBasicType();
+            static constexpr GLenum kFloatTypes[5] = {0, GL_FLOAT, GL_FLOAT_VEC2, GL_FLOAT_VEC3, GL_FLOAT_VEC4};
+            static constexpr GLenum kIntTypes[5] = {0, GL_INT, GL_INT_VEC2, GL_INT_VEC3, GL_INT_VEC4};
+            static constexpr GLenum kUintTypes[5] = {0, GL_UNSIGNED_INT, GL_UNSIGNED_INT_VEC2, GL_UNSIGNED_INT_VEC3,
+                                                     GL_UNSIGNED_INT_VEC4};
+            if (type.isMatrix()) {
+                if (basic != glslang::EbtFloat) return false;
+                static constexpr GLenum kMatTypes[5][5] = {
+                    {}, {},
+                    {0, 0, GL_FLOAT_MAT2, GL_FLOAT_MAT2x3, GL_FLOAT_MAT2x4},
+                    {0, 0, GL_FLOAT_MAT3x2, GL_FLOAT_MAT3, GL_FLOAT_MAT3x4},
+                    {0, 0, GL_FLOAT_MAT4x2, GL_FLOAT_MAT4x3, GL_FLOAT_MAT4},
+                };
+                if (columns < 2 || columns > 4 || components < 2 || components > 4) return false;
+                outType = kMatTypes[columns][components];
+            } else if (components >= 1 && components <= 4) {
+                switch (basic) {
+                case glslang::EbtFloat: outType = kFloatTypes[components]; break;
+                case glslang::EbtInt: outType = kIntTypes[components]; break;
+                case glslang::EbtUint: outType = kUintTypes[components]; break;
+                default: return false;
+                }
+            } else {
+                return false;
+            }
+            outBytesPerElement = static_cast<Uint32>(columns * components) * 4u;
+            return true;
+        }
+    } // namespace
+
+    Bool ProgramObject::ResolveTransformFeedbackVaryings() {
+        m_xfbVaryings.clear();
+        m_xfbStrides.clear();
+        m_xfbBufferMode = m_requestedXfbBufferMode;
+        m_xfbVaryingNameMaxLength = 0;
+        m_xfbNeedsScatteredCapture = false;
+        m_xfbPackedStride = 0;
+        if (m_requestedXfbVaryings.empty()) {
+            return true;
+        }
+
+        // Capture happens at the last vertex-processing stage (geometry, then
+        // tessellation evaluation, then vertex).
+        const glslang::TIntermediate* captureIntermediate = nullptr;
+        for (EShLanguage stage : {EShLangGeometry, EShLangTessEvaluation, EShLangVertex}) {
+            captureIntermediate = m_program->getIntermediate(stage);
+            if (captureIntermediate != nullptr) {
+                break;
+            }
+        }
+        if (captureIntermediate == nullptr) {
+            m_infoLog = "Transform feedback varyings requested but the program has no vertex-processing stage.";
+            return false;
+        }
+        const glslang::TIntermAggregate* linkerObjects = captureIntermediate->findLinkerObjects();
+
+        const Bool interleaved = m_xfbBufferMode == GL_INTERLEAVED_ATTRIBS;
+        Uint32 interleavedOffset = 0;
+        // ARB_transform_feedback3 lets an interleaved capture leave holes (gl_SkipComponents1..4)
+        // and move on to the next buffer (gl_NextBuffer). Both only affect where the following
+        // varyings land, so they are consumed here and never become XfbVaryings of their own -
+        // which also keeps them out of the name list a backend declares on its own driver.
+        Uint32 interleavedBufferIndex = 0;
+        Vector<Uint32> interleavedStrides;
+        for (SizeT i = 0; i < m_requestedXfbVaryings.size(); ++i) {
+            const String& name = m_requestedXfbVaryings[i];
+            if (interleaved && name == "gl_NextBuffer") {
+                interleavedStrides.push_back(interleavedOffset);
+                interleavedOffset = 0;
+                ++interleavedBufferIndex;
+                m_xfbNeedsScatteredCapture = true;
+                continue;
+            }
+            if (interleaved && name.size() == 18 && name.compare(0, 17, "gl_SkipComponents") == 0 &&
+                name[17] >= '1' && name[17] <= '4') {
+                interleavedOffset += static_cast<Uint32>(name[17] - '0') * 4;
+                m_xfbNeedsScatteredCapture = true;
+                continue;
+            }
+            for (SizeT j = 0; j < i; ++j) {
+                if (m_requestedXfbVaryings[j] == name) {
+                    m_infoLog = "Transform feedback varying '" + name + "' is specified more than once.";
+                    return false;
+                }
+            }
+
+            XfbVarying varying;
+            varying.name = name;
+            Uint32 bytesPerElement = 0;
+            Bool resolved = false;
+            if (name == "gl_Position") {
+                varying.type = GL_FLOAT_VEC4;
+                varying.size = 1;
+                bytesPerElement = 16;
+                resolved = true;
+            } else if (name == "gl_PointSize") {
+                varying.type = GL_FLOAT;
+                varying.size = 1;
+                bytesPerElement = 4;
+                resolved = true;
+            } else if (linkerObjects != nullptr) {
+                for (const auto* node : linkerObjects->getSequence()) {
+                    const glslang::TIntermSymbol* symbol = node->getAsSymbolNode();
+                    if (symbol == nullptr || symbol->getType().getQualifier().storage != glslang::EvqVaryingOut) {
+                        continue;
+                    }
+                    if (symbol->getName() != name.c_str()) {
+                        continue;
+                    }
+                    resolved = ResolveXfbSymbolType(symbol->getType(), varying.type, varying.size, bytesPerElement);
+                    break;
+                }
+            }
+            if (!resolved) {
+                m_infoLog = "Transform feedback varying '" + name + "' is not an output of the vertex stage.";
+                return false;
+            }
+
+            varying.byteSize = bytesPerElement * static_cast<Uint32>(varying.size);
+            varying.packedOffsetBytes = m_xfbPackedStride;
+            m_xfbPackedStride += varying.byteSize;
+            if (interleaved) {
+                varying.bufferIndex = interleavedBufferIndex;
+                varying.offsetBytes = interleavedOffset;
+                interleavedOffset += varying.byteSize;
+            } else {
+                varying.bufferIndex = static_cast<Uint32>(m_xfbVaryings.size());
+                varying.offsetBytes = 0;
+            }
+            m_xfbVaryingNameMaxLength =
+                std::max(m_xfbVaryingNameMaxLength, static_cast<Int>(name.size()) + 1);
+            m_xfbVaryings.push_back(Move(varying));
+        }
+
+        constexpr Uint32 kMaxSeparateAttribs = 4;
+        constexpr Uint32 kMaxSeparateComponents = 4;
+        constexpr Uint32 kMaxInterleavedComponents = 64;
+        constexpr Uint32 kMaxTransformFeedbackBuffers = 4;
+        if (interleaved) {
+            interleavedStrides.push_back(interleavedOffset);
+            if (interleavedStrides.size() > kMaxTransformFeedbackBuffers) {
+                m_infoLog = "Transform feedback capture uses more buffers than "
+                            "GL_MAX_TRANSFORM_FEEDBACK_BUFFERS.";
+                return false;
+            }
+            for (const Uint32 stride : interleavedStrides) {
+                if (stride > kMaxInterleavedComponents * 4) {
+                    m_infoLog = "Transform feedback interleaved capture exceeds "
+                                "GL_MAX_TRANSFORM_FEEDBACK_INTERLEAVED_COMPONENTS.";
+                    return false;
+                }
+            }
+            m_xfbStrides = Move(interleavedStrides);
+        } else {
+            if (m_xfbVaryings.size() > kMaxSeparateAttribs) {
+                m_infoLog = "Transform feedback separate capture exceeds "
+                            "GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_ATTRIBS.";
+                return false;
+            }
+            m_xfbStrides.resize(m_xfbVaryings.size());
+            for (SizeT i = 0; i < m_xfbVaryings.size(); ++i) {
+                if (m_xfbVaryings[i].byteSize > kMaxSeparateComponents * 4) {
+                    m_infoLog = "Transform feedback varying '" + m_xfbVaryings[i].name +
+                                "' exceeds GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_COMPONENTS.";
+                    return false;
+                }
+                m_xfbStrides[i] = m_xfbVaryings[i].byteSize;
+            }
+        }
+
+        ResolveGsTriangleStripCapture(captureIntermediate);
+        return true;
+    }
+
+    namespace {
+        // Extracts a geometry shader's per-invocation EmitVertex/EndPrimitive sequence
+        // when it is statically knowable (no emit inside selection/loop/switch). Vulkan
+        // transform feedback captures triangle strips in plain (i, i+1, i+2) order while
+        // GL decomposes odd strip triangles as (i+1, i, i+2) (GL 4.6 table 10.1); with
+        // the static strip lengths the capture buffer can be reordered after EndTF.
+        class GsEmitSequenceTraverser final : public glslang::TIntermTraverser {
+        public:
+            bool visitAggregate(glslang::TVisit, glslang::TIntermAggregate* node) override {
+                if (node->getOp() == glslang::EOpEmitVertex) {
+                    ++emitCount;
+                    hasEmit = true;
+                } else if (node->getOp() == glslang::EOpEndPrimitive) {
+                    FlushStrip();
+                }
+                return true;
+            }
+            bool visitSelection(glslang::TVisit, glslang::TIntermSelection*) override {
+                inControlFlow = true;
+                return true;
+            }
+            bool visitLoop(glslang::TVisit, glslang::TIntermLoop*) override {
+                inControlFlow = true;
+                return true;
+            }
+            bool visitSwitch(glslang::TVisit, glslang::TIntermSwitch*) override {
+                inControlFlow = true;
+                return true;
+            }
+            void FlushStrip() {
+                if (emitCount >= 3) {
+                    stripTriangles.push_back(static_cast<Uint32>(emitCount - 2));
+                }
+                emitCount = 0;
+            }
+
+            Vector<Uint32> stripTriangles;
+            Uint32 emitCount = 0;
+            Bool hasEmit = false;
+            Bool inControlFlow = false;
+        };
+    } // namespace
+
+    void ProgramObject::ResolveGsTriangleStripCapture(const glslang::TIntermediate* captureIntermediate) {
+        m_gsStripTriangles.clear();
+        m_gsStripCaptureFixup = false;
+        if (captureIntermediate == nullptr || m_program == nullptr) {
+            return;
+        }
+        if (m_program->getIntermediate(EShLangGeometry) != captureIntermediate) {
+            return;
+        }
+        if (captureIntermediate->getOutputPrimitive() != glslang::ElgTriangleStrip) {
+            return;
+        }
+        GsEmitSequenceTraverser traverser;
+        const_cast<glslang::TIntermediate*>(captureIntermediate)->getTreeRoot()->traverse(&traverser);
+        traverser.FlushStrip(); // the invocation end acts as an implicit EndPrimitive
+        if (!traverser.hasEmit || traverser.inControlFlow || traverser.stripTriangles.empty()) {
+            return;
+        }
+        m_gsStripTriangles = Move(traverser.stripTriangles);
+        m_gsStripCaptureFixup = true;
+    }
+
     bool ProgramObject::ShaderIsAttached(const SharedPtr<ShaderObject>& shader) {
         MGLOG_D("ProgramObject %u: ShaderIsAttached check for shader %p", m_externalIndex, shader.get());
         auto it = std::find_if(m_shaders.begin(), m_shaders.end(),
@@ -321,12 +577,32 @@ namespace MobileGL::MG_State::GLState {
             return;
         }
 
+        // GL_GEOMETRY_INPUT_TYPE. A draw's primitive type has to be compatible with it
+        // (GL 4.6 core 11.3.1), so it is resolved for every link, not only a capturing one.
+        m_gsInputPrimitive = GL_NONE;
+        if (const glslang::TIntermediate* gs = m_program->getIntermediate(EShLangGeometry)) {
+            switch (gs->getInputPrimitive()) {
+            case glslang::ElgPoints: m_gsInputPrimitive = GL_POINTS; break;
+            case glslang::ElgLines: m_gsInputPrimitive = GL_LINES; break;
+            case glslang::ElgLinesAdjacency: m_gsInputPrimitive = GL_LINES_ADJACENCY; break;
+            case glslang::ElgTriangles: m_gsInputPrimitive = GL_TRIANGLES; break;
+            case glslang::ElgTrianglesAdjacency: m_gsInputPrimitive = GL_TRIANGLES_ADJACENCY; break;
+            default: break;
+            }
+        }
+
         MGLOG_D("ProgramObject %u: Starting reflection", m_externalIndex);
         DoReflection();
         MGLOG_D("ProgramObject %u: Reflection done (linkStatus=%d)", m_externalIndex, (int)m_linkStatus);
         if (!ValidateFragmentOutputLocations()) {
             return;
         }
+        if (!ResolveTransformFeedbackVaryings()) {
+            m_linkStatus = false;
+            MGLOG_E("ProgramObject %u: transform feedback varying resolution failed: %s", m_externalIndex,
+                    m_infoLog.c_str());
+            return;
+        }
 
         MGLOG_D("ProgramObject %u: Starting binary generation", m_externalIndex);
         GenerateBinary();

MobileGL/MG_State/GLState/ProgramState/ProgramObject.h

@@ -334,16 +334,32 @@ namespace MobileGL::MG_State::GLState {
         // draw. The memo is keyed by (backendStateVersion, flags); ResetLinkArtifacts and
         // the binding setters below invalidate it by bumping m_backendStateVersion.
         Bool GetBackendHashMemo(Uint flags, Uint64& outHash) const {
-            if (m_backendHashMemoVersion != m_backendStateVersion || m_backendHashMemoFlags != flags) {
-                return false;
+            if (m_backendHashMemoVersion != m_backendStateVersion) return false;
+            for (const auto& slot : m_backendHashMemoSlots) {
+                if (slot.valid && slot.flags == flags) {
+                    outHash = slot.hash;
+                    return true;
+                }
             }
-            outHash = m_backendHashMemo;
-            return true;
+            return false;
         }
         void SetBackendHashMemo(Uint flags, Uint64 hash) const {
-            m_backendHashMemo = hash;
-            m_backendHashMemoVersion = m_backendStateVersion;
-            m_backendHashMemoFlags = flags;
+            if (m_backendHashMemoVersion != m_backendStateVersion) {
+                for (auto& slot : m_backendHashMemoSlots) slot.valid = false;
+                m_backendHashMemoVersion = m_backendStateVersion;
+                m_backendHashMemoNextSlot = 0;
+            }
+            for (auto& slot : m_backendHashMemoSlots) {
+                if (slot.valid && slot.flags == flags) {
+                    slot.hash = hash;
+                    return;
+                }
+            }
+            auto& slot = m_backendHashMemoSlots[m_backendHashMemoNextSlot];
+            slot.flags = flags;
+            slot.hash = hash;
+            slot.valid = true;
+            m_backendHashMemoNextSlot = (m_backendHashMemoNextSlot + 1) % kBackendHashMemoSlotCount;
         }
 
         void SetUniformSamplerOrImageUnitIndex(Uint location, Int unit) {
@@ -361,6 +377,17 @@ namespace MobileGL::MG_State::GLState {
 
         Bool GetDeleteStatus() const { return m_deleteStatus; }
         Bool GetLinkStatus() const { return m_linkStatus; }
+        // GL_PROGRAM_BINARY_RETRIEVABLE_HINT. MobileGL exposes no program binary format
+        // (GL_NUM_PROGRAM_BINARY_FORMATS is 0), so the hint is pure state - which is all
+        // ARB_get_program_binary requires of it.
+        Bool GetBinaryRetrievableHint() const { return m_binaryRetrievableHint; }
+        void SetBinaryRetrievableHint(Bool hint) { m_binaryRetrievableHint = hint; }
+        // glProgramBinary always fails here (there is no format it could accept) and the
+        // spec then requires the program's LINK_STATUS to read FALSE.
+        void MarkLinkFailedByProgramBinary() {
+            ResetLinkArtifacts();
+            m_infoLog = "No program binary format is supported.";
+        }
         Bool GetValidateStatus() const { return m_validateStatus; }
         Int GetActiveAtomicCounterCount() const { return m_program->getNumAtomicCounters(); }
         Int GetActiveAttributesCount() const { return m_program->getNumPipeInputs(); }
@@ -449,6 +476,55 @@ namespace MobileGL::MG_State::GLState {
             return it == m_shaders.end() ? -1 : (Int)std::distance(m_shaders.begin(), it);
         }
 
+        // Transform feedback (GL 3.0 core: glTransformFeedbackVaryings applies on
+        // the NEXT link; the linked snapshot below is what draws and queries see).
+        struct XfbVarying {
+            String name;
+            GLenum type = GL_FLOAT;
+            GLint size = 1;           // array element count
+            Uint32 bufferIndex = 0;   // capture buffer slot
+            Uint32 offsetBytes = 0;   // offset within the capture buffer
+            Uint32 byteSize = 0;      // bytes captured per vertex for this varying
+            // Offset within the gap-free record a backend that cannot express the GL
+            // layout captures into; see NeedsScatteredTransformFeedbackCapture.
+            Uint32 packedOffsetBytes = 0;
+        };
+        void SetTransformFeedbackVaryings(Vector<String>&& names, GLenum bufferMode) {
+            m_requestedXfbVaryings = Move(names);
+            m_requestedXfbBufferMode = bufferMode;
+        }
+        GLenum GetTransformFeedbackBufferMode() const { return m_xfbBufferMode; }
+        SizeT GetTransformFeedbackVaryingCount() const { return m_xfbVaryings.size(); }
+        const XfbVarying* GetTransformFeedbackVarying(SizeT index) const {
+            return index < m_xfbVaryings.size() ? &m_xfbVaryings[index] : nullptr;
+        }
+        const Vector<XfbVarying>& GetTransformFeedbackVaryings() const { return m_xfbVaryings; }
+        // Stride of one captured vertex in the given capture buffer slot.
+        Uint32 GetTransformFeedbackStride(Uint32 bufferIndex) const {
+            return bufferIndex < m_xfbStrides.size() ? m_xfbStrides[bufferIndex] : 0;
+        }
+        SizeT GetTransformFeedbackBufferCount() const { return m_xfbStrides.size(); }
+        Int GetTransformFeedbackVaryingMaxLength() const { return m_xfbVaryingNameMaxLength; }
+        // True when the capture layout uses gl_SkipComponents / gl_NextBuffer
+        // (ARB_transform_feedback3), which no ES driver can express: it can only pack every
+        // captured varying into one record with no gaps. A backend that captures through
+        // such a driver has to capture into scratch storage and scatter the records into the
+        // application's buffers itself, using packedOffsetBytes as the source offset and
+        // (bufferIndex, offsetBytes, stride) as the destination.
+        Bool NeedsScatteredTransformFeedbackCapture() const { return m_xfbNeedsScatteredCapture; }
+        // Bytes one gap-free captured record occupies.
+        Uint32 GetTransformFeedbackPackedStride() const { return m_xfbPackedStride; }
+        // True when the capture stage is a triangle-strip geometry shader with a
+        // statically-known emit sequence: the Vulkan capture order then needs the GL
+        // odd-triangle vertex swap after EndTransformFeedback.
+        Bool HasGsTriangleStripCaptureFixup() const { return m_gsStripCaptureFixup; }
+        // Triangles per strip, in emission order, for ONE geometry invocation.
+        const Vector<Uint32>& GetGsStripTriangles() const { return m_gsStripTriangles; }
+        // GL_GEOMETRY_INPUT_TYPE of the linked geometry stage (GL_POINTS, GL_LINES,
+        // GL_LINES_ADJACENCY, GL_TRIANGLES or GL_TRIANGLES_ADJACENCY), or GL_NONE when the
+        // program has no geometry stage. Draws must present a compatible primitive type.
+        GLenum GetGeometryInputType() const { return m_gsInputPrimitive; }
+
         Uint GetExternalIndex() const { return m_externalIndex; }
         // Globally-unique, never-reused id for this program object's lifetime. Unlike the GL
         // name (external index), which is freed to a LIFO list and immediately handed back by
@@ -459,6 +535,11 @@ namespace MobileGL::MG_State::GLState {
     private:
         void ResetLinkArtifacts();
         void DoReflection();
+        // Resolves the requested transform feedback varyings against the linked
+        // vertex stage; fails the link (GL semantics) on unknown or duplicate
+        // names or exceeded capture limits.
+        Bool ResolveTransformFeedbackVaryings();
+        void ResolveGsTriangleStripCapture(const glslang::TIntermediate* captureIntermediate);
         void GenerateBinary();
         void WaitUntilGenerationCompleted() const;
         void AddDefaultFragmentShaderIfMissing();
@@ -523,15 +604,38 @@ namespace MobileGL::MG_State::GLState {
         String m_infoLog;
         Bool m_deleteStatus = false;
         Bool m_linkStatus = false;
+        Bool m_binaryRetrievableHint = false;
         Bool m_validateStatus = true;
         Uint32 m_backendStateVersion = 0;
 
         // Backend-owned content-hash memo (see GetBackendHashMemo): valid only while
-        // m_backendStateVersion and the compile flags match the recorded values.
-        mutable Uint64 m_backendHashMemo = 0;
+        // m_backendStateVersion matches. Several slots, not one: a backend may resolve the same
+        // program under more than one compile-flag set within a frame (surface rotation, and the
+        // explicit-LOD sampling variant), and a single slot would then miss on every lookup and
+        // re-hash the program's whole SPIR-V once per draw.
+        static constexpr SizeT kBackendHashMemoSlotCount = 4;
+        struct BackendHashMemoSlot {
+            Uint64 hash = 0;
+            Uint flags = 0;
+            Bool valid = false;
+        };
+        mutable Array<BackendHashMemoSlot, kBackendHashMemoSlotCount> m_backendHashMemoSlots{};
+        mutable SizeT m_backendHashMemoNextSlot = 0;
         mutable Uint32 m_backendHashMemoVersion = ~0u;
-        mutable Uint m_backendHashMemoFlags = 0;
         Uint32 m_uboContentVersion = 0;
         Uint32 m_linkVersion = 0;
+
+        // Transform feedback: request (applies at next link) and linked snapshot.
+        Vector<String> m_requestedXfbVaryings;
+        GLenum m_requestedXfbBufferMode = GL_INTERLEAVED_ATTRIBS;
+        Vector<XfbVarying> m_xfbVaryings;
+        Vector<Uint32> m_xfbStrides;
+        Vector<Uint32> m_gsStripTriangles;
+        Bool m_gsStripCaptureFixup = false;
+        GLenum m_gsInputPrimitive = GL_NONE;
+        GLenum m_xfbBufferMode = GL_INTERLEAVED_ATTRIBS;
+        Int m_xfbVaryingNameMaxLength = 0;
+        Bool m_xfbNeedsScatteredCapture = false;
+        Uint32 m_xfbPackedStride = 0;
     };
 } // namespace MobileGL::MG_State::GLState

MobileGL/MG_State/GLState/ProgramState/ProgramState.cpp

@@ -29,17 +29,25 @@ namespace MobileGL::MG_State::GLState {
         if (!CheckIndexAvail(program, m_programObjects)) return; // FIXME: add error reporting here
         auto& programObject = m_programObjects[program];
         if (programObject != nullptr) {
-            // Snapshot the attachments: deleting the program is a detach point for shaders
-            // that were flagged with glDeleteShader while still attached.
-            const Vector<SharedPtr<ShaderObject>> attachedShaders = programObject->GetAttachedShaders();
             programObject->MarkAsDeleted();
-            programObject.reset();
-            m_programIndexGenerator.Delete(program);
-            for (const auto& shader : attachedShaders) {
-                const Uint shaderName = shader->GetExternalIndex();
-                if (CheckIndexAvail(shaderName, m_shaderObjects) && m_shaderObjects[shaderName] == shader) {
-                    ReleaseShaderNameIfOrphaned(shaderName);
-                }
+            // A program in use is only FLAGGED: its name (and every program query) stays
+            // valid until it stops being current, at which point UseProgram finishes the job.
+            if (programObject == m_currentProgram) return;
+            DestroyProgramSlot(program);
+        }
+    }
+
+    void ProgramState::DestroyProgramSlot(const Uint program) {
+        auto& programObject = m_programObjects[program];
+        // Snapshot the attachments: deleting the program is a detach point for shaders
+        // that were flagged with glDeleteShader while still attached.
+        const Vector<SharedPtr<ShaderObject>> attachedShaders = programObject->GetAttachedShaders();
+        programObject.reset();
+        m_programIndexGenerator.Delete(program);
+        for (const auto& shader : attachedShaders) {
+            const Uint shaderName = shader->GetExternalIndex();
+            if (CheckIndexAvail(shaderName, m_shaderObjects) && m_shaderObjects[shaderName] == shader) {
+                ReleaseShaderNameIfOrphaned(shaderName);
             }
         }
     }
@@ -49,10 +57,22 @@ namespace MobileGL::MG_State::GLState {
     }
 
     void ProgramState::UseProgram(Uint program) {
+        const SharedPtr<ProgramObject> previous = m_currentProgram;
+
         if (program == 0) m_currentProgram.reset();
 
-        if (!CheckIndexAvail(program, m_programObjects)) return;
-        m_currentProgram = m_programObjects[program];
+        if (CheckIndexAvail(program, m_programObjects)) {
+            m_currentProgram = m_programObjects[program];
+        }
+
+        // A deletion flagged while the program was current takes effect the moment it
+        // stops being current.
+        if (previous != nullptr && previous != m_currentProgram && previous->GetDeleteStatus()) {
+            const Uint previousName = previous->GetExternalIndex();
+            if (CheckIndexAvail(previousName, m_programObjects) && m_programObjects[previousName] == previous) {
+                DestroyProgramSlot(previousName);
+            }
+        }
     }
 
     Uint ProgramState::CreateShader(ShaderStage stage) {

MobileGL/MG_State/GLState/ProgramState/ProgramState.h

@@ -35,6 +35,9 @@ namespace MobileGL::MG_State::GLState {
 
     private:
         Bool ShaderHasGLVisibleAttachment(const SharedPtr<ShaderObject>& shaderObject) const;
+        // Frees the name slot and releases orphaned attached shaders; the immediate half
+        // of glDeleteProgram (deferred while the program is current).
+        void DestroyProgramSlot(Uint program);
 
         template <typename T>
         static Bool CheckIndexAvail(const SizeT idx, const Vector<T>& vec) {

MobileGL/MG_State/GLState/ProgramState/ShaderObject.cpp

@@ -169,6 +169,15 @@ namespace MobileGL::MG_State::GLState {
             }
         }
 
+        const std::optional<String> reservedError =
+            MG_Util::ShaderTranspiler::FindReservedIdentifierViolation(compileSource);
+        if (reservedError) {
+            m_compileStatus = false;
+            m_shader.reset();
+            m_infoLog = *reservedError;
+            return;
+        }
+
         // Compile for OpenGL here, so that we can do validation and link
         // like a real OpenGL driver at linking stage
         // Will compile for other backends later.

MobileGL/MG_State/GLState/RenderState/RenderState.cpp

@@ -154,6 +154,17 @@ namespace MobileGL {
                 return m_parameters.PointSize;
             }
 
+            void RenderState::SetPatchVertices(Uint vertices) {
+                if (m_parameters.PatchVertices == vertices) return;
+
+                m_parameters.PatchVertices = vertices;
+                ++m_version;
+            }
+
+            Uint RenderState::GetPatchVertices() const {
+                return m_parameters.PatchVertices;
+            }
+
             void RenderState::SetPolygonOffset(Float factor, Float units) {
                 if (m_parameters.PolygonOffsetFactor == factor && m_parameters.PolygonOffsetUnits == units) return;
 

MobileGL/MG_State/GLState/RenderState/RenderState.h

@@ -224,6 +224,8 @@ namespace MobileGL {
         IntVec4 Viewport = IntVec4(0, 0, 0, 0); // x, y, width, height
         Float LineWidth = 1.0f;
         Float PointSize = 1.0f;
+        // GL_PATCH_VERTICES: how many vertices one tessellation patch consumes.
+        Uint PatchVertices = 3;
         Float PolygonOffsetFactor = 0.0f;
         Float PolygonOffsetUnits = 0.0f;
 
@@ -319,6 +321,8 @@ namespace MobileGL {
                 Float GetLineWidth() const;
                 void SetPointSize(Float size);
                 Float GetPointSize() const;
+                void SetPatchVertices(Uint vertices);
+                Uint GetPatchVertices() const;
                 void SetPolygonOffset(Float factor, Float units);
                 Float GetPolygonOffsetFactor() const;
                 Float GetPolygonOffsetUnits() const;

MobileGL/MG_State/GLState/TextureState/TextureEnum.h

@@ -162,6 +162,7 @@ namespace MobileGL {
         DepthComponent32F,
         Depth24Stencil8,
         Depth32FStencil8,
+        StencilIndex8,
 
         DepthComponent,
         DepthStencil,

MobileGL/MG_State/GLState/TextureState/TextureObject.cpp

@@ -24,6 +24,19 @@ namespace MobileGL {
             TextureObjectBase::TextureObjectBase(TextureTarget target, Uint externalIndex)
                 : m_externalIndex(externalIndex), m_lifetimeId(AllocateLifetimeId()), m_target(target) {
                 m_sampler = MakeShared<SamplerObject>(0);
+                if (target == TextureTarget::TextureRectangle) {
+                    // A rectangle texture has no mip chain, so its initial sampler state is not
+                    // the shared one: TEXTURE_MIN_FILTER is LINEAR and TEXTURE_WRAP_S/T are
+                    // CLAMP_TO_EDGE (GL 4.6 core table 23.15). Leaving the 2D default of
+                    // NEAREST_MIPMAP_LINEAR in place makes the texture mipmap-incomplete from
+                    // birth, and every lookup that the application never re-filtered reads
+                    // (0, 0, 0, 1) instead of its contents.
+                    m_sampler->SetMinFilter(SamplerFilterMode::Linear);
+                    m_sampler->SetMipmapMode(SamplerMipmapMode::None);
+                    m_sampler->SetWrapS(SamplerWrapMode::ClampToEdge);
+                    m_sampler->SetWrapT(SamplerWrapMode::ClampToEdge);
+                    m_sampler->SetWrapR(SamplerWrapMode::ClampToEdge);
+                }
             }
 
             TextureInternalFormat TextureObjectBase::GetFormat() const {
@@ -335,6 +348,64 @@ namespace MobileGL {
 
             // TODO: add other texture types as needed
 
+        Bool SamplesAsIncompleteTexture(const ITextureObject* texture, const SamplerObject* effectiveSampler) {
+            const Bool mipmapped =
+                effectiveSampler != nullptr && effectiveSampler->GetMipmapMode() != SamplerMipmapMode::None;
+            return !IsMipmapCompleteForFilter(texture, mipmapped);
+        }
+
+        Bool IsMipmapCompleteForFilter(const ITextureObject* texture, Bool mipmapped) {
+            if (texture == nullptr) return true;
+            if (!texture->IsComplete()) return false;
+            if (!mipmapped) return true;
+
+            const auto* mipmapTexture = AsMipmapTexture(texture);
+            if (mipmapTexture == nullptr) return true; // no mip chain to be incomplete about
+
+            const UintVec2& levelRange = texture->GetLevelRange();
+            const Uint baseLevel = levelRange.x();
+            const Uint storedLevels = mipmapTexture->GetMipmapLevelCount();
+            if (baseLevel >= storedLevels) return false;
+
+            // An array texture's layer count is not a dimension of the image: it stays put all
+            // the way down the chain (GL 4.6 core 8.14.3). GetMipmapTexelSize reports it in the
+            // slot after the image's own dimensions.
+            const TextureTarget target = texture->GetTarget();
+            Int shrinkingComponents = 3;
+            if (target == TextureTarget::Texture1DArray) {
+                shrinkingComponents = 1;
+            } else if (target == TextureTarget::Texture2DArray || target == TextureTarget::TextureCubeMapArray) {
+                shrinkingComponents = 2;
+            }
+
+            for (const auto uploadTarget : texture->GetUploadTargets()) {
+                const IntVec3 baseSize = mipmapTexture->GetMipmapTexelSize(uploadTarget, baseLevel);
+                Int largest = 0;
+                for (Int component = 0; component < shrinkingComponents; ++component) {
+                    largest = std::max(largest, baseSize[component]);
+                }
+                if (largest <= 0) return false;
+
+                // p = log2 of the largest base dimension: the last level the chain needs
+                // before every dimension has reached 1. TEXTURE_MAX_LEVEL can cut it short.
+                Uint p = 0;
+                for (Int extent = largest; extent > 1; extent >>= 1) ++p;
+                const Uint lastLevel = std::min(baseLevel + p, levelRange.y());
+
+                for (Uint level = baseLevel; level <= lastLevel; ++level) {
+                    if (level >= storedLevels) return false;
+                    const IntVec3 actual = mipmapTexture->GetMipmapTexelSize(uploadTarget, level);
+                    for (Int component = 0; component < 3; ++component) {
+                        const Int expected = component < shrinkingComponents
+                            ? std::max(1, baseSize[component] >> (level - baseLevel))
+                            : baseSize[component];
+                        if (actual[component] != expected) return false;
+                    }
+                }
+            }
+            return true;
+        }
+
         } // namespace GLState
     } // namespace MG_State
 } // namespace MobileGL

MobileGL/MG_State/GLState/TextureState/TextureObject.h

@@ -15,7 +15,11 @@
 #include <MG_Util/Math/VectorTypes.h>
 
 namespace MobileGL::MG_State::GLState {
-    class ITextureObject {
+    // Texture objects are always SharedPtr-owned (TextureState creates every instance via
+    // MakeShared, including the per-target default objects). enable_shared_from_this lets
+    // backends that only receive a reference (e.g. syncing a name-deleted texture kept
+    // alive by an FBO attachment) still register a weak liveness reference for GC.
+    class ITextureObject : public std::enable_shared_from_this<ITextureObject> {
     public:
         using TargetEnum = TextureTarget;
         virtual ~ITextureObject() = default;
@@ -152,6 +156,20 @@ namespace MobileGL::MG_State::GLState {
                    ? static_cast<TextureObjectMipmap*>(texture)
                    : nullptr;
     }
+    // Whether the texture satisfies the mipmap-completeness rules a minification filter
+    // that samples the mip chain imposes (GL 4.6 core 8.17): every level from the base to
+    // the effective max must exist at exactly half the previous one's size. `mipmapped` is
+    // the effective sampler's answer to "does this filter read more than the base level" -
+    // when it is false only base-level completeness matters, which the ordinary
+    // IsComplete() already covers. Sampling an incomplete texture returns (0, 0, 0, 1).
+    Bool IsMipmapCompleteForFilter(const ITextureObject* texture, Bool mipmapped);
+
+    // The rule above asked as the backends need it: does a lookup on this texture read
+    // (0, 0, 0, 1) instead of its contents? `effectiveSampler` is the sampler object bound
+    // to the unit when there is one, otherwise the texture's own. A backend answers yes by
+    // routing the texture to whatever it already uses for "nothing is bound there".
+    Bool SamplesAsIncompleteTexture(const ITextureObject* texture, const SamplerObject* effectiveSampler);
+
     inline const TextureObjectMipmap* AsMipmapTexture(const ITextureObject* texture) {
         return (texture && texture->GetStorageType() == TextureStorageType::Mipmap)
                    ? static_cast<const TextureObjectMipmap*>(texture)

MobileGL/MG_State/GLState/TextureState/TextureObjectBuffer.h

@@ -21,10 +21,31 @@ namespace MobileGL {
                 BindingSlot<BufferObject>& GetBufferBindingSlot(
                     TextureUploadTarget target = TextureUploadTarget::TextureBuffer);
 
+                // The window of the attached buffer the texture addresses. glTexBuffer attaches
+                // the whole buffer, which is expressed here as an offset of 0 and a size of
+                // kWholeBuffer so a later respecify of the buffer keeps being followed - a stored
+                // size would freeze the texture at the size the buffer happened to have.
+                static constexpr SizeT kWholeBuffer = ~static_cast<SizeT>(0);
+                void SetBufferRange(SizeT offset, SizeT size) {
+                    m_bufferRangeOffset = offset;
+                    m_bufferRangeSize = size;
+                }
+                SizeT GetBufferRangeOffset() const { return m_bufferRangeOffset; }
+                // Resolved against the buffer's current size, so kWholeBuffer tracks it.
+                SizeT GetBufferRangeSizeInBytes() const {
+                    const auto& buffer = m_bufferBindingSlot.GetBoundObject();
+                    const SizeT bufferSize = buffer != nullptr ? buffer->GetSize() : 0;
+                    if (m_bufferRangeSize == kWholeBuffer) return bufferSize;
+                    const SizeT available = bufferSize > m_bufferRangeOffset ? bufferSize - m_bufferRangeOffset : 0;
+                    return std::min(m_bufferRangeSize, available);
+                }
+
             protected:
                 Uint GetIndexOfTextureUploadTarget(TextureUploadTarget target) const override;
 
                 BindingSlot<BufferObject> m_bufferBindingSlot = BindingSlot<BufferObject>(BufferTarget::Texture);
+                SizeT m_bufferRangeOffset = 0;
+                SizeT m_bufferRangeSize = kWholeBuffer;
                 const Vector<TextureUploadTarget> m_uploadTargets{TextureUploadTarget::TextureBuffer};
             };
         } // namespace GLState

MobileGL/MG_State/GLState/VertexArrayState/VertexArrayObject.cpp

@@ -77,6 +77,26 @@ namespace MobileGL::MG_State::GLState {
         BumpAttributeFormatVersion(index);
     }
 
+    void VertexArrayObject::MirrorPointerIntoBinding(Uint index, const SharedPtr<BufferObject>& buffer, SizeT offset,
+                                                     int effectiveStride) {
+        if (index >= MAX_VERTEX_ATTRIBS || index >= MAX_VERTEX_ATTRIB_BINDINGS) return;
+
+        // glVertexAttribPointer is defined in terms of the binding model (GL 4.6 core 10.3.2): it
+        // also sets binding point `index` to the buffer, the pointer as the offset, and the
+        // *effective* stride, and points the attribute at that binding point with relative offset 0.
+        // The flat attribute view keeps the raw stride, because VERTEX_ATTRIB_ARRAY_STRIDE reports
+        // that argument verbatim, so the binding point is recorded alongside the resolved attribute
+        // rather than being resolved into it.
+        m_attributeBindingIndex[index] = index;
+        m_attributeRelativeOffset[index] = 0;
+
+        auto& binding = m_bindingPoints[index];
+        binding.Buffer = buffer;
+        binding.Offset = offset;
+        binding.Stride = effectiveStride;
+        binding.Divisor = m_attributes[index].Divisor;
+    }
+
     void VertexArrayObject::BindAttributeBuffer(Uint index, const SharedPtr<BufferObject>& buffer) {
         if (index >= MAX_VERTEX_ATTRIBS) return;
 
@@ -111,6 +131,11 @@ namespace MobileGL::MG_State::GLState {
 
     void VertexArrayObject::SetAttributeDivisor(Uint index, Uint divisor) {
         if (index >= MAX_VERTEX_ATTRIBS) return;
+        // glVertexAttribDivisor is VertexBindingDivisor on the attribute's own binding point
+        // (GL 4.6 core 10.3.2), so the binding-point view has to follow the resolved attribute.
+        if (index < MAX_VERTEX_ATTRIB_BINDINGS && m_attributeBindingIndex[index] == index) {
+            m_bindingPoints[index].Divisor = divisor;
+        }
         if (m_attributes[index].Divisor == divisor) return;
         m_attributes[index].Divisor = divisor;
         BumpAttributeFormatVersion(index);
@@ -187,18 +212,18 @@ namespace MobileGL::MG_State::GLState {
     }
 
     void VertexArrayObject::SetAttributeFormatSeparate(Uint attribIndex, int size, DataType type, Bool normalized,
-                                                       Bool isInteger, Uint relativeOffset) {
+                                                       Bool isInteger, Uint relativeOffset, Bool isBgra) {
         if (attribIndex >= MAX_VERTEX_ATTRIBS) return;
         if (size < 1 || size > 4) return;
 
         auto& attr = m_attributes[attribIndex];
         if (attr.Size != size || attr.Type != type || attr.Normalized != normalized || attr.IsInteger != isInteger ||
-            attr.IsBgra || m_attributeRelativeOffset[attribIndex] != relativeOffset) {
+            attr.IsBgra != isBgra || m_attributeRelativeOffset[attribIndex] != relativeOffset) {
             attr.Size = size;
             attr.Type = type;
             attr.Normalized = normalized;
             attr.IsInteger = isInteger;
-            attr.IsBgra = false; // the binding-format path (glVertexAttribFormat) does not carry BGRA
+            attr.IsBgra = isBgra;
             m_attributeRelativeOffset[attribIndex] = relativeOffset;
             BumpAttributeFormatVersion(attribIndex);
         }

MobileGL/MG_State/GLState/VertexArrayState/VertexArrayObject.h

@@ -64,6 +64,12 @@ namespace MobileGL {
 
                 void BindAttributeBuffer(Uint index, const SharedPtr<BufferObject>& buffer);
 
+                // Record what the pointer-style API implies for the binding-point view: attribute
+                // `index` bound to binding point `index` with relative offset 0, and that binding
+                // point carrying the buffer, the pointer offset and the effective stride.
+                void MirrorPointerIntoBinding(Uint index, const SharedPtr<BufferObject>& buffer, SizeT offset,
+                                              int effectiveStride);
+
                 BindingSlot<BufferObject>& GetIndexBufferBindingSlot();
                 const BindingSlot<BufferObject>& GetIndexBufferBindingSlot() const;
 
@@ -82,7 +88,22 @@ namespace MobileGL {
                 void SetBindingDivisor(Uint bindingIndex, Uint divisor);
                 void SetAttributeBinding(Uint attribIndex, Uint bindingIndex);
                 void SetAttributeFormatSeparate(Uint attribIndex, int size, DataType type, Bool normalized,
-                                                Bool isInteger, Uint relativeOffset);
+                                                Bool isInteger, Uint relativeOffset, Bool isBgra = false);
+
+                // The binding-point view the attributes were resolved from. Kept queryable
+                // because glGetVertexArrayIndexed[64]iv reports it verbatim, and the resolved
+                // flat attribute cannot always be inverted back into it.
+                Uint GetAttributeRelativeOffset(Uint attribIndex) const {
+                    return attribIndex < m_attributeRelativeOffset.size() ? m_attributeRelativeOffset[attribIndex] : 0;
+                }
+                Uint GetAttributeBindingIndex(Uint attribIndex) const {
+                    return attribIndex < m_attributeBindingIndex.size() ? m_attributeBindingIndex[attribIndex]
+                                                                       : attribIndex;
+                }
+                const VertexBufferBindingPoint& GetBindingPoint(Uint bindingIndex) const {
+                    static const VertexBufferBindingPoint kEmpty{};
+                    return bindingIndex < m_bindingPoints.size() ? m_bindingPoints[bindingIndex] : kEmpty;
+                }
 
                 const VertexAttributeVersion& GetAttributeVersion(Uint index) const;
                 const Array<VertexAttributeVersion, MAX_VERTEX_ATTRIBS>& GetAllAttributeVersions() const;
@@ -104,6 +125,23 @@ namespace MobileGL {
                     m_backendHashMemoVersion = m_configVersion;
                 }
 
+                // Backend-owned resolved-state memo: an opaque pointer into the
+                // backend's vertex-input-state cache plus the cache's eviction
+                // epoch, valid while the config version matches. Lets the
+                // per-draw path skip the content hash AND the cache lookup; the
+                // epoch guards against the cache evicting the pointee.
+                Bool GetBackendStateMemo(const void*& outState, Uint64& outEpoch) const {
+                    if (m_backendStateMemoVersion != m_configVersion) return false;
+                    outState = m_backendStateMemo;
+                    outEpoch = m_backendStateMemoEpoch;
+                    return true;
+                }
+                void SetBackendStateMemo(const void* state, Uint64 epoch) const {
+                    m_backendStateMemo = state;
+                    m_backendStateMemoEpoch = epoch;
+                    m_backendStateMemoVersion = m_configVersion;
+                }
+
             private:
                 void BumpAttributeFormatVersion(Uint index);
                 void BumpAttributeBufferVersion(Uint index);
@@ -137,6 +175,9 @@ namespace MobileGL {
                 Uint32 m_configVersion = 0;
                 mutable Uint64 m_backendHashMemo = 0;
                 mutable Uint32 m_backendHashMemoVersion = ~0u;
+                mutable const void* m_backendStateMemo = nullptr;
+                mutable Uint64 m_backendStateMemoEpoch = 0;
+                mutable Uint32 m_backendStateMemoVersion = ~0u;
             };
         } // namespace GLState
     } // namespace MG_State

MobileGL/MG_Test/BackendLoader/BackendLoaderTest.cpp

@@ -34,6 +34,11 @@ namespace {
         GLint maxFragmentImageUniforms = 4;
         GLint maxComputeImageUniforms = 5;
         bool maxGeometryImageUniformsQueried = false;
+        GLfloat minFragmentInterpolationOffset = -0.75f;
+        GLfloat maxFragmentInterpolationOffset = 0.625f;
+        GLint fragmentInterpolationOffsetBits = 6;
+        bool fragmentInterpolationLimitsQueried = false;
+        bool fragmentInterpolationQueryRaisesError = false;
         // Emulates ANGLE-on-Vulkan: the draw reads the indirect command's
         // baseInstance word and exposes it through gl_InstanceID.
         bool drawLeaksBaseInstanceWord = false;
@@ -108,6 +113,14 @@ namespace {
             case GL_MAX_COMPUTE_IMAGE_UNIFORMS:
                 *data = g_fake.maxComputeImageUniforms;
                 break;
+            case GL_FRAGMENT_INTERPOLATION_OFFSET_BITS:
+                g_fake.fragmentInterpolationLimitsQueried = true;
+                if (g_fake.fragmentInterpolationQueryRaisesError) {
+                    g_fake.pendingError = GL_INVALID_ENUM;
+                } else {
+                    *data = g_fake.fragmentInterpolationOffsetBits;
+                }
+                break;
             // FillInGLESCapabilities reads the context version before running the
             // baseInstance probe, which requires ES >= 3.1.
             case GL_MAJOR_VERSION:
@@ -155,6 +168,22 @@ namespace {
                 g_fake.maxTextureMaxAnisotropyQueried = true;
                 data[0] = g_fake.maxTextureMaxAnisotropy;
                 break;
+            case GL_MIN_FRAGMENT_INTERPOLATION_OFFSET:
+                g_fake.fragmentInterpolationLimitsQueried = true;
+                if (g_fake.fragmentInterpolationQueryRaisesError) {
+                    g_fake.pendingError = GL_INVALID_ENUM;
+                } else {
+                    data[0] = g_fake.minFragmentInterpolationOffset;
+                }
+                break;
+            case GL_MAX_FRAGMENT_INTERPOLATION_OFFSET:
+                g_fake.fragmentInterpolationLimitsQueried = true;
+                if (g_fake.fragmentInterpolationQueryRaisesError) {
+                    g_fake.pendingError = GL_INVALID_ENUM;
+                } else {
+                    data[0] = g_fake.maxFragmentInterpolationOffset;
+                }
+                break;
             // Two-component range queries.
             case GL_ALIASED_LINE_WIDTH_RANGE:
             case GL_SMOOTH_LINE_WIDTH_RANGE:
@@ -462,6 +491,52 @@ TEST(ImageUniformCapabilities, QueriesRealPerStageLimitsAndConservativelyGatesGe
     EXPECT_TRUE(g_fake.maxGeometryImageUniformsQueried);
 }
 
+TEST(FragmentInterpolationCapabilities, QueriesOnlyWhenSupportedAndPreservesDriverLimits) {
+    const auto funcs = MakeFakeGLESFunctions();
+
+    ResetFakeDriver();
+    g_fake.maxVertexSsboBlocks = 0;
+    MobileGL::MG_External::GLESCapabilities unsupportedCaps;
+    ASSERT_TRUE(MobileGL::MG_Util::BackendLoader::FillInGLESCapabilities(unsupportedCaps, funcs));
+    EXPECT_FALSE(unsupportedCaps.SupportsShaderMultisampleInterpolation);
+    EXPECT_FALSE(g_fake.fragmentInterpolationLimitsQueried);
+    EXPECT_FLOAT_EQ(unsupportedCaps.MinFragmentInterpolationOffset, -0.5f);
+    EXPECT_FLOAT_EQ(unsupportedCaps.MaxFragmentInterpolationOffset, 0.4375f);
+    EXPECT_EQ(unsupportedCaps.FragmentInterpolationOffsetBits, 4);
+
+    ResetFakeDriver();
+    g_fake.maxVertexSsboBlocks = 0;
+    g_fake.extensions.emplace_back("GL_OES_shader_multisample_interpolation");
+    // A stale error from an earlier capability probe must not make the optional
+    // interpolation query look like it failed.
+    g_fake.pendingError = GL_INVALID_OPERATION;
+    MobileGL::MG_External::GLESCapabilities supportedCaps;
+    ASSERT_TRUE(MobileGL::MG_Util::BackendLoader::FillInGLESCapabilities(supportedCaps, funcs));
+    EXPECT_TRUE(supportedCaps.SupportsShaderMultisampleInterpolation);
+    EXPECT_TRUE(g_fake.fragmentInterpolationLimitsQueried);
+    EXPECT_FLOAT_EQ(supportedCaps.MinFragmentInterpolationOffset, g_fake.minFragmentInterpolationOffset);
+    EXPECT_FLOAT_EQ(supportedCaps.MaxFragmentInterpolationOffset, g_fake.maxFragmentInterpolationOffset);
+    EXPECT_EQ(supportedCaps.FragmentInterpolationOffsetBits, g_fake.fragmentInterpolationOffsetBits);
+    EXPECT_EQ(funcs.glGetError(), GL_NO_ERROR);
+}
+
+TEST(FragmentInterpolationCapabilities, QueryErrorIsDrainedAndFallsBackToCoreMinimums) {
+    ResetFakeDriver();
+    g_fake.maxVertexSsboBlocks = 0;
+    g_fake.extensions.emplace_back("GL_OES_shader_multisample_interpolation");
+    g_fake.fragmentInterpolationQueryRaisesError = true;
+    const auto funcs = MakeFakeGLESFunctions();
+
+    MobileGL::MG_External::GLESCapabilities caps;
+    ASSERT_TRUE(MobileGL::MG_Util::BackendLoader::FillInGLESCapabilities(caps, funcs));
+
+    EXPECT_TRUE(g_fake.fragmentInterpolationLimitsQueried);
+    EXPECT_FLOAT_EQ(caps.MinFragmentInterpolationOffset, -0.5f);
+    EXPECT_FLOAT_EQ(caps.MaxFragmentInterpolationOffset, 0.4375f);
+    EXPECT_EQ(caps.FragmentInterpolationOffsetBits, 4);
+    EXPECT_EQ(funcs.glGetError(), GL_NO_ERROR);
+}
+
 // The extension string is what apps gate on (LWJGL builds GLCapabilities from it), so advertising
 // it on a driver that cannot filter anisotropically would leave them silently on trilinear.
 TEST(TextureAnisotropyCapabilities, ExtensionIsAdvertisedOnlyWhenTheHostDriverSupportsIt) {

MobileGL/MG_Test/Program/ProgramUtilTest.cpp

@@ -223,7 +223,7 @@ void main() {
 
     PreprocessShaderSource(ShaderStage::Vertex, source);
 
-    EXPECT_EQ(source.find("#version 330 core\n"), 0);
+    EXPECT_EQ(source.find("#version 330 core "), 0);
     EXPECT_NE(source.find("in vec3 position;"), String::npos);
     EXPECT_NE(source.find("out vec2 uv;"), String::npos);
     EXPECT_EQ(source.find("attribute"), String::npos);
@@ -323,7 +323,7 @@ void main() {
 
     PreprocessShaderSource(ShaderStage::Fragment, source);
 
-    EXPECT_EQ(source.find("#version 330 core\n"), 0);
+    EXPECT_EQ(source.find("#version 330 core "), 0);
     EXPECT_NE(source.find("out vec4 mg_FragColor;\n"), String::npos);
     EXPECT_NE(source.find("in vec2 uv;"), String::npos);
     EXPECT_NE(source.find("texture(texture0, uv)"), String::npos);
@@ -379,7 +379,7 @@ void main() {
 
     PreprocessShaderSource(ShaderStage::Fragment, source);
 
-    EXPECT_EQ(source.find("#version 330 core\n"), 0);
+    EXPECT_EQ(source.find("#version 330 core "), 0);
     EXPECT_NE(source.find("vec4 sample = texture(DiffuseSampler"), String::npos);
     EXPECT_NE(source.find("totalAlpha = totalAlpha + sample.a;"), String::npos);
     EXPECT_NE(source.find("float totalSamples = 0.0;"), String::npos);
@@ -405,7 +405,7 @@ void main() {
 )";
     PreprocessShaderSource(ShaderStage::Vertex, vertexSource);
 
-    EXPECT_EQ(vertexSource.find("#version 330 core\n"), 0);
+    EXPECT_EQ(vertexSource.find("#version 330 core "), 0);
     EXPECT_NE(vertexSource.find("in vec3 sample;"), String::npos);
 
     ShaderAttrib vertexAttrib{.shaderType = GL_VERTEX_SHADER, .sourceStr = vertexSource};
@@ -425,7 +425,7 @@ void main() {
 )";
     PreprocessShaderSource(ShaderStage::Fragment, fragmentSource);
 
-    EXPECT_EQ(fragmentSource.find("#version 330 core\n"), 0);
+    EXPECT_EQ(fragmentSource.find("#version 330 core "), 0);
     EXPECT_NE(fragmentSource.find("uniform sampler2D sample;"), String::npos);
     EXPECT_NE(fragmentSource.find("texture(sample, texCoord)"), String::npos);
 
@@ -483,7 +483,9 @@ void main() {
 )";
         PreprocessShaderSource(ShaderStage::Fragment, source);
 
-        EXPECT_EQ(source.find("#version 460 core\n"), 0);
+        // An explicitly declared modern core version keeps its number (see "keep declared modern
+        // GLSL versions strict"); only the BOM goes.
+        EXPECT_EQ(source.find(String(inputVersion) + "\n"), 0);
         EXPECT_EQ(source.find("\xef\xbb\xbf"), String::npos);
 
         ShaderAttrib attrib{.shaderType = GL_FRAGMENT_SHADER, .sourceStr = source};
@@ -568,10 +570,12 @@ void main() {
 )";
     PreprocessShaderSource(ShaderStage::Fragment, source);
 
-    const SizeT versionPos = source.find("#version 330 core\n");
+    const SizeT versionPos = source.find("#version 330 core ");
     const SizeT outputPos = source.find("out vec4 mg_FragColor;\n");
     EXPECT_NE(versionPos, String::npos);
-    EXPECT_EQ(outputPos, versionPos + std::strlen("#version 330 core\n"));
+    // The normalized directive carries a marker recording that this 330 came from a legacy
+    // declaration, so measure the line rather than assuming its length.
+    EXPECT_EQ(outputPos, source.find('\n', versionPos) + 1);
     EXPECT_NE(source.find("// #version 460 core"), String::npos);
     // This #line sits ahead of the version directive, where GLSL would never have honoured it, so
     // it is still dropped. Directives that follow the version line are kept - see
@@ -684,7 +688,7 @@ void main() {
     }
 }
 
-TEST_F(ProgramUtilTest, PreprocessModernSampleQualifierStaysAtVersion460) {
+TEST_F(ProgramUtilTest, PreprocessModernSampleQualifierStaysAtItsDeclaredVersion) {
     using namespace MG_Util::ShaderTranspiler;
 
     String source = R"(#version 400 core
@@ -697,7 +701,7 @@ void main() {
 )";
     PreprocessShaderSource(ShaderStage::Fragment, source);
 
-    EXPECT_EQ(source.find("#version 460 core\n"), 0);
+    EXPECT_EQ(source.find("#version 400 core\n"), 0);
     EXPECT_NE(source.find("sample in vec4 interpolatedColor;"), String::npos);
 
     ShaderAttrib attrib{.shaderType = GL_FRAGMENT_SHADER, .sourceStr = source};
@@ -746,7 +750,7 @@ void main() {
 
     PreprocessShaderSource(ShaderStage::Fragment, source);
 
-    EXPECT_EQ(source.find("#version 330 core\n"), 0);
+    EXPECT_EQ(source.find("#version 330 core "), 0);
     EXPECT_NE(source.find("layout(location = 0) out vec4 mg_FragData[8];\n"), String::npos);
     EXPECT_NE(source.find("mg_FragData[0] = vec4(1.0);"), String::npos);
     EXPECT_NE(source.find("mg_FragData[1].a = 0.5;"), String::npos);
@@ -902,16 +906,17 @@ TEST_F(ProgramUtilTest, CompileSimpleVertexShader) {
 }
 
 // Legacy desktop sources are normalized to "#version 330 core", which is stricter than the 460 they
-// used to be forced to. A shader declaring 330 while using 420-era syntax without the matching
-// #extension line is accepted by real drivers, so CompileShader retries it at 460 instead of failing.
-TEST_F(ProgramUtilTest, CompileShaderRetriesAt460WhenLegacyVersionRejects420Syntax) {
+// used to be forced to. A legacy shader using 420-era syntax without the matching #extension line
+// is accepted by real drivers, so CompileShader retries the normalized source at 460 rather than
+// failing. Only MobileGL's own normalization is rescued this way - an application-declared
+// "#version 330" keeps strict 3.30 semantics, which is what the CTS negative-compile cases need.
+TEST_F(ProgramUtilTest, CompileShaderRetriesAt460WhenNormalizedLegacyVersionRejects420Syntax) {
     using namespace MG_Util::ShaderTranspiler;
-    String source = R"(#version 330
+    String source = R"(#version 130
 layout(binding = 0) uniform sampler2D InSampler;
-in vec2 texCoord;
-out vec4 fragColor;
+varying vec2 texCoord;
 void main() {
-    fragColor = texture(InSampler, texCoord);
+    gl_FragColor = texture2D(InSampler, texCoord);
 })";
     PreprocessShaderSource(ShaderStage::Fragment, source);
     // The normal path still emits 330 - the retry must not become the default.
@@ -952,10 +957,21 @@ void main() {
 TEST_F(ProgramUtilTest, RetargetLegacyVersionDirectiveOnlyTouchesNormalizedDesktopCore) {
     using namespace MG_Util::ShaderTranspiler;
 
-    String normalized = "#version 330 core\nvoid main() {}\n";
+    // Only MobileGL's own normalization is retargetable, and it is recognised by the marker the
+    // preprocessor leaves on the directive line - so normalize a legacy source rather than
+    // hand-writing the directive the marker belongs to.
+    String normalized = "#version 130\nvoid main() {}\n";
+    PreprocessShaderSource(ShaderStage::Vertex, normalized);
+    ASSERT_EQ(normalized.find("#version 330 core "), 0u);
     EXPECT_TRUE(RetargetLegacyVersionDirectiveTo460(normalized));
     EXPECT_EQ(normalized.find("#version 460 core"), 0u);
 
+    // An application that declared 330 itself keeps strict 3.30 semantics: raising it would
+    // re-legalize the CTS negative-compile cases.
+    String declared330 = "#version 330 core\nvoid main() {}\n";
+    EXPECT_FALSE(RetargetLegacyVersionDirectiveTo460(declared330));
+    EXPECT_EQ(declared330.find("#version 330 core"), 0u);
+
     // Already modern: nothing to retarget.
     String modern = "#version 460 core\nvoid main() {}\n";
     EXPECT_FALSE(RetargetLegacyVersionDirectiveTo460(modern));

MobileGL/MG_Test/SanityTest.cpp

@@ -33,6 +33,7 @@
 #include <MG_Util/ShaderTranspiler/ShaderCompiler.h>
 #include <MG_Util/ShaderTranspiler/ShaderSourceProcessor.h>
 #include <MG_Util/Debug/Log.h>
+#include <FastSTL/UnorderedMap.h>
 
 namespace {
     class DynamicParameterBackend final : public MobileGL::MG_Backend::BackendObject {
@@ -196,7 +197,9 @@ TEST(DirectGLESSanity, AdvertisesDepthTextureForGlmarkShadowScenes) {
     EXPECT_NE(std::find(extensions.begin(), extensions.end(), MobileGL::E_GL_ARB_depth_texture), extensions.end());
 }
 
-TEST(DirectGLESSanity, AdvertisesVoxyRequiredRenderingExtensionsWithoutRaisingGLVersion) {
+// The advertised target went back to 3.3 (see "restore target GL version to 3.3"); Voxy only ever
+// needed the extensions, which stay advertised, so assert what the backend really reports.
+TEST(DirectGLESSanity, AdvertisesVoxyRequiredRenderingExtensions) {
     MobileGL::MG_Backend::DirectGLES::BackendObject_DirectGLES backend;
     const auto& rendererInfo = backend.GetRendererInfo().RendererGLInfo;
     const auto& extensions = rendererInfo.Extensions;
@@ -396,7 +399,8 @@ TEST(DirectVulkanSanity, RenderPassExtentUsesSwapchainSizeOnlyForDefaultFramebuf
               MobileGL::IntVec2(512, 512));
 }
 
-TEST(DirectVulkanSanity, AdvertisesVoxyRequiredRenderingExtensionsWithoutRaisingGLVersion) {
+// See the DirectGLES twin above: the target version is 3.3 again, the extensions are what matter.
+TEST(DirectVulkanSanity, AdvertisesVoxyRequiredRenderingExtensions) {
     MobileGL::MG_Backend::DirectVulkan::BackendObject_DirectVulkan backend;
     const auto& rendererInfo = backend.GetRendererInfo().RendererGLInfo;
     const auto& extensions = rendererInfo.Extensions;
@@ -515,6 +519,50 @@ TEST(DirectGLESSanity, PreservesHostPerStageImageUniformLimits) {
     EXPECT_EQ(params.MaxComputeImageUniforms, 5);
 }
 
+TEST(FragmentInterpolationCapabilities, PlumbsGLESAndBothVulkanPropertyPaths) {
+    using namespace MobileGL;
+
+    MG_External::GLESCapabilities glesCaps;
+    glesCaps.MinFragmentInterpolationOffset = -0.75f;
+    glesCaps.MaxFragmentInterpolationOffset = 0.625f;
+    glesCaps.FragmentInterpolationOffsetBits = 6;
+    MG_Backend::DirectGLES::BackendObject_DirectGLES glesBackend;
+    glesBackend.ApplyGLESCapabilitiesForTesting(glesCaps);
+    EXPECT_FLOAT_EQ(glesBackend.GetDynamicParameters().MinFragmentInterpolationOffset, -0.75f);
+    EXPECT_FLOAT_EQ(glesBackend.GetDynamicParameters().MaxFragmentInterpolationOffset, 0.625f);
+    EXPECT_EQ(glesBackend.GetDynamicParameters().FragmentInterpolationOffsetBits, 6);
+
+    VkPhysicalDeviceProperties properties{};
+    // A common Vulkan limit pair: max is one representable 4-bit step below 0.5.
+    properties.limits.minInterpolationOffset = -0.5f;
+    properties.limits.maxInterpolationOffset = 0.4375f;
+    properties.limits.subPixelInterpolationOffsetBits = 4;
+    MG_External::VulkanCapabilities vkCaps;
+    MG_Util::BackendLoader::FillInVulkanCapabilities(vkCaps, properties);
+    EXPECT_FLOAT_EQ(vkCaps.MinFragmentInterpolationOffset, -0.5f);
+    EXPECT_FLOAT_EQ(vkCaps.MaxFragmentInterpolationOffset, 0.4375f);
+    EXPECT_EQ(vkCaps.FragmentInterpolationOffsetBits, 4);
+
+    vkCaps.MinFragmentInterpolationOffset = -0.875f;
+    vkCaps.MaxFragmentInterpolationOffset = 0.75f;
+    vkCaps.FragmentInterpolationOffsetBits = 7;
+    MG_Backend::DirectVulkan::BackendObject_DirectVulkan vkBackend;
+    vkBackend.ApplyVulkanCapabilitiesForTesting(vkCaps);
+    EXPECT_FLOAT_EQ(vkBackend.GetDynamicParameters().MinFragmentInterpolationOffset, -0.875f);
+    EXPECT_FLOAT_EQ(vkBackend.GetDynamicParameters().MaxFragmentInterpolationOffset, 0.75f);
+    EXPECT_EQ(vkBackend.GetDynamicParameters().FragmentInterpolationOffsetBits, 7);
+
+    // Invalid/zero host data cannot under-advertise the OpenGL 4 minimums.
+    MG_External::VulkanCapabilities invalidCaps;
+    invalidCaps.MinFragmentInterpolationOffset = 0.0f;
+    invalidCaps.MaxFragmentInterpolationOffset = 0.0f;
+    invalidCaps.FragmentInterpolationOffsetBits = 0;
+    vkBackend.ApplyVulkanCapabilitiesForTesting(invalidCaps);
+    EXPECT_LE(vkBackend.GetDynamicParameters().MinFragmentInterpolationOffset, -0.5f);
+    EXPECT_FLOAT_EQ(vkBackend.GetDynamicParameters().MaxFragmentInterpolationOffset, 0.4375f);
+    EXPECT_EQ(vkBackend.GetDynamicParameters().FragmentInterpolationOffsetBits, 4);
+}
+
 TEST(DirectVulkanSanity, AdvertisesSubgroupOnlyWhenVulkanReportsUsableSupport) {
     using namespace MobileGL;
 
@@ -637,6 +685,48 @@ TEST(GetterSanity, ClampsMaxVertexAttribsToCurrentValueStorageCapacity) {
     MG_State::pGLContext.reset();
 }
 
+TEST(GetterSanity, ReportsFragmentInterpolationLimitsForFloatAndIntegerQueries) {
+    using namespace MobileGL;
+
+    auto previousContext = Move(MG_State::pGLContext);
+    auto previousBackend = Move(MG_Backend::pActiveBackendObject);
+    MG_State::pGLContext = MakeUnique<MG_State::GLState::GLContext>();
+
+    MG_Backend::DynamicBackendParameters params;
+    params.MinFragmentInterpolationOffset = -0.75f;
+    params.MaxFragmentInterpolationOffset = 0.4375f;
+    params.FragmentInterpolationOffsetBits = 6;
+    MG_Backend::pActiveBackendObject = MakeUnique<DynamicParameterBackend>(params);
+
+    GLfloat floatValue = 0.0f;
+    MG_Impl::GLImpl::GetFloatv(GL_MIN_FRAGMENT_INTERPOLATION_OFFSET, &floatValue);
+    EXPECT_FLOAT_EQ(floatValue, -0.75f);
+    MG_Impl::GLImpl::GetFloatv(GL_MAX_FRAGMENT_INTERPOLATION_OFFSET, &floatValue);
+    EXPECT_FLOAT_EQ(floatValue, 0.4375f);
+    MG_Impl::GLImpl::GetFloatv(GL_FRAGMENT_INTERPOLATION_OFFSET_BITS, &floatValue);
+    EXPECT_FLOAT_EQ(floatValue, 6.0f);
+
+    GLint intValue = 0;
+    MG_Impl::GLImpl::GetIntegerv(GL_MIN_FRAGMENT_INTERPOLATION_OFFSET, &intValue);
+    EXPECT_EQ(intValue, -1);
+    MG_Impl::GLImpl::GetIntegerv(GL_MAX_FRAGMENT_INTERPOLATION_OFFSET, &intValue);
+    EXPECT_EQ(intValue, 0);
+    MG_Impl::GLImpl::GetIntegerv(GL_FRAGMENT_INTERPOLATION_OFFSET_BITS, &intValue);
+    EXPECT_EQ(intValue, 6);
+
+    GLboolean boolValue = GL_FALSE;
+    MG_Impl::GLImpl::GetBooleanv(GL_MIN_FRAGMENT_INTERPOLATION_OFFSET, &boolValue);
+    EXPECT_EQ(boolValue, GL_TRUE);
+    MG_Impl::GLImpl::GetBooleanv(GL_MAX_FRAGMENT_INTERPOLATION_OFFSET, &boolValue);
+    EXPECT_EQ(boolValue, GL_TRUE);
+    MG_Impl::GLImpl::GetBooleanv(GL_FRAGMENT_INTERPOLATION_OFFSET_BITS, &boolValue);
+    EXPECT_EQ(boolValue, GL_TRUE);
+    EXPECT_EQ(MG_Impl::GLImpl::GetError(), GL_NO_ERROR);
+
+    MG_Backend::pActiveBackendObject = Move(previousBackend);
+    MG_State::pGLContext = Move(previousContext);
+}
+
 TEST(GetterSanity, PerStageImageUniformQueriesMatchShaderCompilerLimits) {
     using namespace MobileGL;
 
@@ -1736,3 +1826,61 @@ TEST(DirectGLESStateGuards, DefaultFramebufferBindGoesThroughShadow) {
     FramebufferImpl::BindFramebufferId(GL_DRAW_FRAMEBUFFER, 7); // must reach the driver again
     EXPECT_EQ(mocks.log.Count("BindFramebuffer:"), 3u);
 }
+
+// FastSTL::unordered_map::erase(iterator) regression coverage. The open-addressing
+// iterator constructor snaps forward from a tombstoned slot to the successor, so
+// erase must NOT advance the rebuilt iterator again: the old double-advance skipped
+// one live element per erase, and erasing the element in the highest occupied
+// bucket pushed the returned index past bucket_count where it never compared equal
+// to end() again - erase-while-iterating sweeps (pipeline/program cache eviction)
+// then ran off the bucket array and fed garbage handles to vkDestroyPipeline
+// (device crash on first mass eviction during world load).
+TEST(FastSTLSanity, EraseWhileIteratingVisitsEveryElementExactlyOnce) {
+    FastSTL::unordered_map<MobileGL::Uint64, MobileGL::Uint64> map;
+    constexpr MobileGL::Uint64 kCount = 1000;
+    for (MobileGL::Uint64 key = 0; key < kCount; ++key) {
+        map.emplace(key * 0x9e3779b97f4a7c15ull, key);
+    }
+    ASSERT_EQ(map.size(), kCount);
+
+    MobileGL::SizeT visited = 0;
+    for (auto it = map.begin(); it != map.end();) {
+        it = map.erase(it);
+        ++visited;
+        ASSERT_LE(visited, kCount); // old code: runaway past end / skipped entries
+    }
+    EXPECT_EQ(visited, kCount);
+    EXPECT_EQ(map.size(), 0u);
+}
+
+TEST(FastSTLSanity, EraseReturnsTheSuccessorElement) {
+    FastSTL::unordered_map<MobileGL::Uint32, MobileGL::Uint32> map;
+    for (MobileGL::Uint32 key = 1; key <= 64; ++key) {
+        map.emplace(key, key);
+    }
+
+    // Erasing every other visited element must still visit all 64 exactly once:
+    // the iterator returned by erase names the very next element, not one past it.
+    MobileGL::SizeT visited = 0;
+    MobileGL::SizeT erased = 0;
+    for (auto it = map.begin(); it != map.end();) {
+        ++visited;
+        if ((visited & 1) != 0) {
+            it = map.erase(it);
+            ++erased;
+        } else {
+            ++it;
+        }
+        ASSERT_LE(visited, 64u);
+    }
+    EXPECT_EQ(visited, 64u);
+    EXPECT_EQ(map.size(), 64u - erased);
+}
+
+TEST(FastSTLSanity, ErasingTheOnlyElementReturnsEnd) {
+    FastSTL::unordered_map<MobileGL::Uint32, MobileGL::Uint32> map;
+    map.emplace(42u, 1u);
+    auto next = map.erase(map.begin());
+    EXPECT_EQ(next, map.end());
+    EXPECT_TRUE(map.empty());
+}

MobileGL/MG_Test/Texture/TextureTest.cpp

@@ -1867,11 +1867,13 @@ TEST_F(TextureTest, DirectGLESTreats2DArrayAsSupportedTextureTarget) {
     EXPECT_TRUE(IsSupportedTextureTarget(TextureTarget::Texture2DArray));
     EXPECT_TRUE(IsSupportedTextureTarget(TextureTarget::Texture3D));
     EXPECT_TRUE(IsSupportedTextureTarget(TextureTarget::Texture2D));
-    // 1D and 1D-array are emulated as 2D / 2D-array (MapToBackendTextureTarget), matching
-    // SPIRV-Cross's ES 1D-as-2D shader emission; only rectangle textures stay unsupported.
+    // Every desktop-only target is stored on an ES one (MapToBackendTextureTarget): 1D and
+    // 1D-array as 2D / 2D-array, matching SPIRV-Cross's ES 1D-as-2D shader emission, and
+    // rectangle as a plain 2D - it is single-level and already clamps, so only the
+    // non-normalized coordinates differ and LowerRectImages handles those.
     EXPECT_TRUE(IsSupportedTextureTarget(TextureTarget::Texture1D));
     EXPECT_TRUE(IsSupportedTextureTarget(TextureTarget::Texture1DArray));
-    EXPECT_FALSE(IsSupportedTextureTarget(TextureTarget::TextureRectangle));
+    EXPECT_TRUE(IsSupportedTextureTarget(TextureTarget::TextureRectangle));
 }
 
 // 2D-array textures keep their layer count constant across mip levels (GL 3.3 §3.9);

MobileGL/MG_Util/BackendLoaders/OpenGL/Loader.cpp

@@ -9,6 +9,7 @@
 #include "Loader.h"
 #include "MG_Util/Types.h"
 #include <Config.h>
+#include <cmath>
 #if defined(_WIN32)
 #ifndef WIN32_LEAN_AND_MEAN
 #define WIN32_LEAN_AND_MEAN 1
@@ -823,6 +824,12 @@ namespace MobileGL::MG_Util::BackendLoader {
                 if (std::strcmp(extension, "GL_EXT_texture_norm16") == 0) {
                     caps.SupportsNorm16Texture = true;
                 }
+                if (std::strcmp(extension, "GL_EXT_render_snorm") == 0) {
+                    caps.SupportsRenderSnorm = true;
+                }
+                if (std::strcmp(extension, "GL_EXT_sRGB_write_control") == 0) {
+                    caps.SupportsSrgbWriteControl = true;
+                }
                 if (std::strcmp(extension, "GL_EXT_texture_filter_anisotropic") == 0) {
                     caps.SupportsTextureFilterAnisotropy = true;
                 }
@@ -838,8 +845,14 @@ namespace MobileGL::MG_Util::BackendLoader {
                 if (std::strcmp(extension, "GL_NV_shader_noperspective_interpolation") == 0) {
                     caps.SupportsNoperspectiveInterpolation = true;
                 }
+                if (std::strcmp(extension, "GL_OES_shader_multisample_interpolation") == 0) {
+                    caps.SupportsShaderMultisampleInterpolation = true;
+                }
             }
         }
+        caps.SupportsShaderMultisampleInterpolation =
+            caps.SupportsShaderMultisampleInterpolation || caps.GLESVersion.Major > 3 ||
+            (caps.GLESVersion.Major == 3 && caps.GLESVersion.Minor >= 2);
 
         // Detect optional raster/color-mask entry points by whether they loaded. glColorMaski is GLES
         // 3.2 core (no extension string), so pointer presence is the reliable signal for all of these.
@@ -900,6 +913,15 @@ namespace MobileGL::MG_Util::BackendLoader {
         GLint maxColorAttachments = 8;
         GLint maxClipDistances = 8;
         GLint maxViewports = 16;
+        GLfloat minFragmentInterpolationOffset = -0.5f;
+        GLfloat maxFragmentInterpolationOffset = 0.4375f;
+        GLint fragmentInterpolationOffsetBits = 4;
+        // Core minimums of both APIs (GL 4.6 table 23.53, ES 3.1 table 20.40); the probe
+        // below only ever widens them.
+        GLint minProgramTextureGatherOffset = -8;
+        GLint maxProgramTextureGatherOffset = 7;
+        GLint maxPatchVertices = 32;
+        GLint maxTessGenLevel = 64;
         glesFuncs.glGetFloatv(GL_ALIASED_LINE_WIDTH_RANGE, aliasedLineWidthRange);
         glesFuncs.glGetFloatv(GL_SMOOTH_LINE_WIDTH_RANGE, smoothLineWidthRange);
         glesFuncs.glGetFloatv(GL_SMOOTH_LINE_WIDTH_GRANULARITY, &smoothLineWidthGranularity);
@@ -919,6 +941,23 @@ namespace MobileGL::MG_Util::BackendLoader {
         glesFuncs.glGetIntegerv(GL_MAX_INTEGER_SAMPLES, &maxIntegerSamples);
         glesFuncs.glGetIntegerv(GL_MAX_SAMPLES, &maxSamples);
         glesFuncs.glGetIntegerv(GL_MAX_SAMPLE_MASK_WORDS, &maxSampleMaskWords);
+        // MobileGL's sample-mask state only stores a single 32-bit word (see
+        // RenderState::SampleMaskValue) and SampleMaski_State() rejects any
+        // maskNumber other than 0. Advertising the real driver's value here (e.g.
+        // NVIDIA's GLES driver reports 2) makes glSampleMaski(1, ...) - which
+        // dEQP's per-case gluStateReset always issues up to GL_MAX_SAMPLE_MASK_WORDS -
+        // raise GL_INVALID_VALUE and aborts the whole glcts process after every
+        // single test case. 1 is a spec-legal value (the minimum required), so cap
+        // to what is actually implemented instead of forwarding the raw driver limit.
+        maxSampleMaskWords = std::min(maxSampleMaskWords, 1);
+        glesFuncs.glGetIntegerv(GL_MAX_PATCH_VERTICES, &maxPatchVertices);
+        glesFuncs.glGetIntegerv(GL_MAX_TESS_GEN_LEVEL, &maxTessGenLevel);
+        glesFuncs.glGetIntegerv(GL_MIN_PROGRAM_TEXTURE_GATHER_OFFSET, &minProgramTextureGatherOffset);
+        glesFuncs.glGetIntegerv(GL_MAX_PROGRAM_TEXTURE_GATHER_OFFSET, &maxProgramTextureGatherOffset);
+        // A driver that leaves the probe untouched (pre-ES 3.1, or an ignored enum) must not
+        // drag the advertised range below what GL 4.0 requires of us.
+        minProgramTextureGatherOffset = std::min(minProgramTextureGatherOffset, -8);
+        maxProgramTextureGatherOffset = std::max(maxProgramTextureGatherOffset, 7);
         glesFuncs.glGetIntegerv(GL_MAX_TEXTURE_IMAGE_UNITS, &maxTextureImageUnits);
         glesFuncs.glGetIntegerv(GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS, &maxVertexTextureImageUnits);
         glesFuncs.glGetIntegerv(GL_MAX_COMPUTE_TEXTURE_IMAGE_UNITS, &maxComputeTextureImageUnits);
@@ -950,6 +989,29 @@ namespace MobileGL::MG_Util::BackendLoader {
         glesFuncs.glGetIntegerv(GL_MAX_VIEWPORTS, &maxViewports);
         glesFuncs.glGetIntegerv(GL_MAX_VIEWPORT_DIMS, maxViewportDims);
         glesFuncs.glGetIntegerv(GL_VIEWPORT_SUBPIXEL_BITS, &viewportSubpixelBits);
+        if (caps.SupportsShaderMultisampleInterpolation && glesFuncs.glGetFloatv) {
+            const auto drainErrors = [&glesFuncs]() {
+                Bool hadError = false;
+                if (glesFuncs.glGetError) {
+                    while (glesFuncs.glGetError() != GL_NO_ERROR) hadError = true;
+                }
+                return hadError;
+            };
+
+            // Isolate these optional queries from errors raised by preceding capability
+            // probes, then consume any query error so initialization never leaks it into
+            // the application's first glGetError call.
+            drainErrors();
+            glesFuncs.glGetFloatv(GL_MIN_FRAGMENT_INTERPOLATION_OFFSET, &minFragmentInterpolationOffset);
+            glesFuncs.glGetFloatv(GL_MAX_FRAGMENT_INTERPOLATION_OFFSET, &maxFragmentInterpolationOffset);
+            glesFuncs.glGetIntegerv(GL_FRAGMENT_INTERPOLATION_OFFSET_BITS, &fragmentInterpolationOffsetBits);
+            if (drainErrors()) {
+                MGLOG_W("Fragment interpolation limit query failed; using OpenGL minimums");
+                minFragmentInterpolationOffset = -0.5f;
+                maxFragmentInterpolationOffset = 0.4375f;
+                fragmentInterpolationOffsetBits = 4;
+            }
+        }
         // Only legal to query once the extension has been seen in the loop above, hence not batched
         // with the unconditional probes: on a driver without it this raises GL_INVALID_ENUM.
         if (caps.SupportsTextureFilterAnisotropy) {
@@ -979,6 +1041,10 @@ namespace MobileGL::MG_Util::BackendLoader {
         caps.MaxIntegerSamples = maxIntegerSamples;
         caps.MaxSamples = maxSamples;
         caps.MaxSampleMaskWords = maxSampleMaskWords;
+        caps.MaxPatchVertices = maxPatchVertices;
+        caps.MaxTessGenLevel = maxTessGenLevel;
+        caps.MinProgramTextureGatherOffset = minProgramTextureGatherOffset;
+        caps.MaxProgramTextureGatherOffset = maxProgramTextureGatherOffset;
         caps.MaxTextureImageUnits = maxTextureImageUnits;
         caps.MaxVertexTextureImageUnits = maxVertexTextureImageUnits;
         caps.MaxComputeTextureImageUnits = maxComputeTextureImageUnits;
@@ -990,6 +1056,18 @@ namespace MobileGL::MG_Util::BackendLoader {
         caps.MaxComputeWorkGroupInvocations = maxComputeWorkGroupInvocations;
         caps.MaxShaderStorageBufferBindings = maxShaderStorageBufferBindings;
         caps.MaxTextureBufferSize = maxTextureBufferSize;
+        // Through glesFuncs, like every other capability query here: a bare glGetIntegerv resolves
+        // to MobileGL's own exported entry point, which answers this pname from the very
+        // capability table being filled in - so the driver's real alignment never arrived and the
+        // backend reported an unconstrained offset it cannot honour.
+        GLint textureBufferOffsetAlignment = 1;
+        glesFuncs.glGetIntegerv(GL_TEXTURE_BUFFER_OFFSET_ALIGNMENT, &textureBufferOffsetAlignment);
+        // Core in ES 3.2 and in EXT_texture_buffer; an older context rejects the pname and leaves
+        // the default in place.
+        if (glesFuncs.glGetError) {
+            while (glesFuncs.glGetError() != GL_NO_ERROR) {}
+        }
+        caps.TextureBufferOffsetAlignment = std::max(1, textureBufferOffsetAlignment);
         caps.MaxUniformBufferBindings = maxUniformBufferBindings;
         caps.MaxUniformBlockSize = maxUniformBlockSize;
         caps.MaxImageUnits = maxImageUnits;
@@ -1007,6 +1085,19 @@ namespace MobileGL::MG_Util::BackendLoader {
         caps.ViewportBoundsRangeMin = viewportBoundsRange[0];
         caps.ViewportBoundsRangeMax = viewportBoundsRange[1];
         caps.ViewportSubpixelBits = viewportSubpixelBits;
+        caps.MinFragmentInterpolationOffset =
+            std::isfinite(minFragmentInterpolationOffset) && minFragmentInterpolationOffset <= -0.5f
+                ? minFragmentInterpolationOffset
+                : -0.5f;
+        caps.MaxFragmentInterpolationOffset = 0.4375f;
+        caps.FragmentInterpolationOffsetBits = 4;
+        if (fragmentInterpolationOffsetBits >= 4 && std::isfinite(maxFragmentInterpolationOffset)) {
+            const Float requiredMaxOffset = 0.5f - std::ldexp(1.0f, -fragmentInterpolationOffsetBits);
+            if (maxFragmentInterpolationOffset >= requiredMaxOffset) {
+                caps.MaxFragmentInterpolationOffset = maxFragmentInterpolationOffset;
+                caps.FragmentInterpolationOffsetBits = fragmentInterpolationOffsetBits;
+            }
+        }
         MGLOG_I("    GL_ALIASED_LINE_WIDTH_RANGE: [%.3f, %.3f]", caps.AliasedLineWidthRangeMin,
                 caps.AliasedLineWidthRangeMax);
         MGLOG_I("    GL_SMOOTH_LINE_WIDTH_RANGE: [%.3f, %.3f]", caps.SmoothLineWidthRangeMin,

MobileGL/MG_Util/BackendLoaders/OpenGL/Loader.h

@@ -1031,6 +1031,13 @@ namespace MobileGL {
             String GLESShadingLanguageVersionString;
             Bool SupportsPersistentMapping = false;
             Bool SupportsNorm16Texture = false;
+            // GL_EXT_render_snorm is present, so the signed-normalized formats are colour-renderable
+            // (and usable as multisample texture storage) rather than texture-only.
+            Bool SupportsRenderSnorm = false;
+            // GL_EXT_sRGB_write_control is present, so GL_FRAMEBUFFER_SRGB can be turned off.
+            // GLES has no such switch in core: writes into an sRGB attachment are ALWAYS encoded,
+            // while desktop GL leaves GL_FRAMEBUFFER_SRGB disabled by default and writes raw.
+            Bool SupportsSrgbWriteControl = false;
             // GL_EXT_texture_filter_anisotropic is present, so sampler/texture
             // anisotropy may be forwarded without raising GL_INVALID_ENUM in GLES.
             Bool SupportsTextureFilterAnisotropy = false;
@@ -1055,6 +1062,9 @@ namespace MobileGL {
             // SPIRV-Cross's `#extension ... : require` would fail to compile and MobileGL falls back
             // to stripping the NoPerspective decoration (smooth interpolation) via StripNoPerspectivePass.
             Bool SupportsNoperspectiveInterpolation = false;
+            // GLES 3.2 core or GL_OES_shader_multisample_interpolation exposes
+            // interpolateAtOffset and the three fragment-offset limit queries.
+            Bool SupportsShaderMultisampleInterpolation = false;
             // GL_RENDERER contains "ANGLE".
             Bool IsAngleRenderer = false;
             // GL_RENDERER contains both "ANGLE" and "llvmpipe".
@@ -1092,6 +1102,10 @@ namespace MobileGL {
             Int MaxIntegerSamples = 1;
             Int MaxSamples = 1;
             Int MaxSampleMaskWords = 1;
+            Int MaxPatchVertices = 32;
+            Int MaxTessGenLevel = 64;
+            Int MinProgramTextureGatherOffset = -8;
+            Int MaxProgramTextureGatherOffset = 7;
             Int MaxTextureImageUnits = 32;
             Int MaxVertexTextureImageUnits = 32;
             Int MaxComputeTextureImageUnits = 32;
@@ -1103,6 +1117,8 @@ namespace MobileGL {
             Int MaxComputeWorkGroupInvocations = 128;
             Int MaxShaderStorageBufferBindings = 8;
             Int MaxTextureBufferSize = 65536;
+            // GL_TEXTURE_BUFFER_OFFSET_ALIGNMENT; 1 means the offset is unconstrained.
+            Int TextureBufferOffsetAlignment = 1;
             Int MaxUniformBufferBindings = 24;
             Int MaxUniformBlockSize = 16384;
             Int MaxImageUnits = 8;
@@ -1120,6 +1136,9 @@ namespace MobileGL {
             Float ViewportBoundsRangeMin = 0.0f;
             Float ViewportBoundsRangeMax = 0.0f;
             Int ViewportSubpixelBits = 0;
+            Float MinFragmentInterpolationOffset = -0.5f;
+            Float MaxFragmentInterpolationOffset = 0.4375f;
+            Int FragmentInterpolationOffsetBits = 4;
         };
     } // namespace MG_External
 

MobileGL/MG_Util/BackendLoaders/Vulkan/Loader.cpp

@@ -9,6 +9,7 @@
 #include "Loader.h"
 
 #include <Config.h>
+#include <cmath>
 
 namespace MobileGL::MG_Util::BackendLoader {
     namespace {
@@ -40,6 +41,25 @@ namespace MobileGL::MG_Util::BackendLoader {
             return MaxSampleCountFromFlags(commonFlags);
         }
 
+        void FillFragmentInterpolationLimits(MG_External::VulkanCapabilities& caps,
+                                             const VkPhysicalDeviceLimits& limits) {
+            caps.MinFragmentInterpolationOffset =
+                std::isfinite(limits.minInterpolationOffset) && limits.minInterpolationOffset <= -0.5f
+                    ? limits.minInterpolationOffset
+                    : -0.5f;
+
+            caps.MaxFragmentInterpolationOffset = 0.4375f;
+            caps.FragmentInterpolationOffsetBits = 4;
+            const Int bits = static_cast<Int>(limits.subPixelInterpolationOffsetBits);
+            if (bits >= 4 && std::isfinite(limits.maxInterpolationOffset)) {
+                const Float requiredMaxOffset = 0.5f - std::ldexp(1.0f, -bits);
+                if (limits.maxInterpolationOffset >= requiredMaxOffset) {
+                    caps.MaxFragmentInterpolationOffset = limits.maxInterpolationOffset;
+                    caps.FragmentInterpolationOffsetBits = bits;
+                }
+            }
+        }
+
         VulkanDynamicFunctions LoadVulkanDynamicFunctions(VkInstance instance) {
             VulkanDynamicFunctions loaded{};
             if (instance == VK_NULL_HANDLE) {
@@ -157,6 +177,8 @@ namespace MobileGL::MG_Util::BackendLoader {
         caps.MaxComputeWorkGroupInvocations = static_cast<Int>(p.limits.maxComputeWorkGroupInvocations);
         caps.MaxShaderStorageBufferBindings = static_cast<Int>(p.limits.maxDescriptorSetStorageBuffers);
         caps.MaxTextureBufferSize = static_cast<Int>(p.limits.maxTexelBufferElements);
+        caps.TextureBufferOffsetAlignment =
+            static_cast<Int>(std::max<VkDeviceSize>(1, p.limits.minTexelBufferOffsetAlignment));
         caps.MaxUniformBufferBindings = static_cast<Int>(p.limits.maxDescriptorSetUniformBuffers);
         caps.MaxUniformBlockSize = static_cast<Int>(p.limits.maxUniformBufferRange);
         caps.MaxImageUnits = static_cast<Int>(p.limits.maxPerStageDescriptorStorageImages);
@@ -171,6 +193,7 @@ namespace MobileGL::MG_Util::BackendLoader {
         caps.ViewportBoundsRangeMin = p.limits.viewportBoundsRange[0];
         caps.ViewportBoundsRangeMax = p.limits.viewportBoundsRange[1];
         caps.ViewportSubpixelBits = static_cast<Int>(p.limits.viewportSubPixelBits);
+        FillFragmentInterpolationLimits(caps, p.limits);
 
         VkPhysicalDeviceFeatures supportedFeatures{};
         vkGetPhysicalDeviceFeatures(physicalDevice, &supportedFeatures);
@@ -247,6 +270,8 @@ namespace MobileGL::MG_Util::BackendLoader {
         caps.MaxComputeWorkGroupInvocations = static_cast<Int>(properties.limits.maxComputeWorkGroupInvocations);
         caps.MaxShaderStorageBufferBindings = static_cast<Int>(properties.limits.maxDescriptorSetStorageBuffers);
         caps.MaxTextureBufferSize = static_cast<Int>(properties.limits.maxTexelBufferElements);
+        caps.TextureBufferOffsetAlignment =
+            static_cast<Int>(std::max<VkDeviceSize>(1, properties.limits.minTexelBufferOffsetAlignment));
         caps.MaxUniformBufferBindings = static_cast<Int>(properties.limits.maxDescriptorSetUniformBuffers);
         caps.MaxUniformBlockSize = static_cast<Int>(properties.limits.maxUniformBufferRange);
         caps.MaxImageUnits = static_cast<Int>(properties.limits.maxPerStageDescriptorStorageImages);
@@ -261,6 +286,7 @@ namespace MobileGL::MG_Util::BackendLoader {
         caps.ViewportBoundsRangeMin = properties.limits.viewportBoundsRange[0];
         caps.ViewportBoundsRangeMax = properties.limits.viewportBoundsRange[1];
         caps.ViewportSubpixelBits = static_cast<Int>(properties.limits.viewportSubPixelBits);
+        FillFragmentInterpolationLimits(caps, properties.limits);
         caps.SupportsWideLines = false;
         // This helper only receives properties, not VkPhysicalDeviceFeatures. Leave optional
         // stage writes disabled rather than inferring them from descriptor limits alone.

MobileGL/MG_Util/BackendLoaders/Vulkan/Loader.h

@@ -54,6 +54,8 @@ namespace MobileGL {
             Int MaxComputeWorkGroupInvocations = 128;
             Int MaxShaderStorageBufferBindings = 8;
             Int MaxTextureBufferSize = 65536;
+            // GL_TEXTURE_BUFFER_OFFSET_ALIGNMENT; 1 means the offset is unconstrained.
+            Int TextureBufferOffsetAlignment = 1;
             Int MaxUniformBufferBindings = 24;
             Int MaxUniformBlockSize = 16384;
             Int MaxImageUnits = 8;
@@ -68,6 +70,9 @@ namespace MobileGL {
             Float ViewportBoundsRangeMin = 0.0f;
             Float ViewportBoundsRangeMax = 0.0f;
             Int ViewportSubpixelBits = 0;
+            Float MinFragmentInterpolationOffset = -0.5f;
+            Float MaxFragmentInterpolationOffset = 0.4375f;
+            Int FragmentInterpolationOffsetBits = 4;
             Bool SupportsWideLines = false;
             // Storage-image descriptors are limited per stage by
             // maxPerStageDescriptorStorageImages, but writes/atomics outside compute additionally

MobileGL/MG_Util/Classifiers/TextureEnumClassifier.cpp

@@ -28,6 +28,7 @@ namespace MobileGL {
 
         bool IsStencilFormatInternalFormat(TextureInternalFormat internalformat) {
             switch (internalformat) {
+            case TextureInternalFormat::StencilIndex8:
             case TextureInternalFormat::Depth24Stencil8:
             case TextureInternalFormat::Depth32FStencil8:
             case TextureInternalFormat::DepthStencil: