[Fix] (DirectVulkan, MG_State): make a shader-written storage buffer readable on Magma

Reading a buffer a compute shader wrote gave zeros: the frontend shadow that MapBuffer
resolves against is only maintained by uploads, and Magma had no path back. Every
KHR-GL40.texture_gather case ends by dispatching a compute shader into an SSBO and
comparing the mapped result, so 66 of 75 failed on it.

Magma needs no readback: EnsureGpuResidentStorage - the same host-visible coherent
adoption the transform feedback capture already uses - makes the shadow BE the memory
the shader writes, so binding a buffer as a shader storage buffer now adopts it. What
coherence does not give is ordering: the writes are visible once they have happened,
and the CPU was reading before the dispatch had retired. The readback op therefore
submits the recorded work and waits.

That exposed a mistake in the frontend flag this rides on: MarkGpuWritten skipped
GPU-resident buffers, reasoning there was no shadow to refresh. True, but the wait is
still needed - "reconcile with the GPU write" is not always "copy it back", and which
of the two it is belongs to the backend. The flag now only says a write is outstanding;
DirectGLES's readback still skips its persistent-mapped buffers when copying.

KHR-GL40.texture_gather on Magma: 66 failures -> 19 (the rest are rectangle textures,
mipmap completeness and tessellation, all still to do). Espryt stays at 75/75.

.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]

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;
 

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;
@@ -300,7 +332,21 @@ 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;
+                GLESProbeFormatInfo outerFallbackInfo;
-                const Bool hasForcedFallback =
+                const Bool outerHasForcedFallback =
-                    BuildFallbackProbeFormatInfo(requestedInternalFormat, forcedOptions, true, fallbackInfo);
+                    BuildFallbackProbeFormatInfo(requestedInternalFormat, forcedOptions, true, outerFallbackInfo);
-                if (!hasForcedFallback) {
+                if (!outerHasForcedFallback) {
-                    BuildFallbackProbeFormatInfo(requestedInternalFormat, driverOptions, false, fallbackInfo);
+                    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;
+                Bool shouldProbeFallbackRenderbuffer = outerHasForcedFallback;
-                if (!hasForcedFallback) {
+                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 &&
+                if (shouldProbeFallbackRenderbuffer && outerFallbackInfo.InternalFormat != GL_UNKNOWN_MGL &&
-                    ProbeRenderbuffer(gl, fallbackInfo.InternalFormat, logicalFormat, false, 1)) {
+                    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
@@ -1025,6 +1131,8 @@ 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 +1142,24 @@ 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
 
@@ -2405,7 +2432,19 @@ 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) {                                                             \
@@ -2679,6 +2718,31 @@ 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) {
@@ -3181,6 +3245,7 @@ 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;
@@ -3232,8 +3297,10 @@ 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;
@@ -3316,6 +3383,21 @@ namespace MobileGL::MG_Backend::DirectGLES {
                     effectiveSpirv = &noperspectiveSpirv;
                 }
 
+                // ES has no rectangle sampler, and SPIRV-Cross refuses the whole module rather
+                // than approximating one. Rewriting the type to 2D is exact for a lookup that
+                // takes integer texel coordinates and needs the coordinate divided by the
+                // texture size for one that does not - see NormalizeRectSamplerCoordinates
+                // below, which the ESSL the transpiler produces goes through. The pass declines
+                // anything neither step can convert.
+                Vector<unsigned int> rectLoweredSpirv;
+                Bool loweredRectImages = false;
+                if (MG_Util::ShaderTranspiler::ShaderCompiler::LowerRectImagesForEssl(*effectiveSpirv,
+                                                                                       rectLoweredSpirv) &&
+                    !rectLoweredSpirv.empty()) {
+                    effectiveSpirv = &rectLoweredSpirv;
+                    loweredRectImages = true;
+                }
+
                 MG_Util::ShaderTranspiler::SpvcSession spvcSession(*effectiveSpirv,
                     MG_Util::ShaderTranspiler::SessionUsageBit::Transpile);
 
@@ -3338,6 +3420,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
                     r.log += spvcSession.GetLastErrorString();
                     r.errc = -5;
                     MGLOG_E("%s", r.log.c_str());
+                    m_backendProgramUsable = false;
                     continue;
                 }
 
@@ -3347,6 +3430,28 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 source = RemoveLayoutBinding(source);
                 source = ProcessOutColorLocations(source);
                 source = ForceFlatIntegerVaryings(source, glShaderType);
+                source = BroadcastLegacyFragColor(std::move(source), glShaderType, m_fragColorBroadcastCount);
+                source = EmulateTextureLodBias(source);
+                if (loweredRectImages) {
+                    // The image type is 2D now, so the transpiled lookups address [0,1]; the
+                    // application wrote them in texels. Only the frontend still knows which
+                    // samplers were declared rectangle.
+                    Vector<String> rectSamplerNames;
+                    const Uint uniformCount = stateProgramObject->GetUniformCount();
+                    for (Uint i = 0; i < uniformCount; ++i) {
+                        switch (stateProgramObject->GetActiveUniformType(i)) {
+                        case GL_SAMPLER_2D_RECT:
+                        case GL_SAMPLER_2D_RECT_SHADOW:
+                        case GL_INT_SAMPLER_2D_RECT:
+                        case GL_UNSIGNED_INT_SAMPLER_2D_RECT:
+                            rectSamplerNames.push_back(stateProgramObject->GetActiveUniformName(i));
+                            break;
+                        default:
+                            break;
+                        }
+                    }
+                    source = NormalizeRectSamplerCoordinates(source, rectSamplerNames);
+                }
                 source = EmulateBaseInstanceInVertexShader(std::move(source), glShaderType);
                 source = PromoteDrawParameterGlobalsToUniforms(std::move(source), glShaderType);
                 source = ForceSupporterOutput(source);
@@ -3377,6 +3482,7 @@ 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;
                 }
 
@@ -3386,12 +3492,33 @@ 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);
@@ -3504,6 +3631,11 @@ 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);
             }
         }
@@ -3512,12 +3644,18 @@ 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);
+            MGLOG_D("Using program %u", programToBind);
-            g_GLESFuncs.glUseProgram(m_backendProgramId);
+            g_GLESFuncs.glUseProgram(programToBind);
-            g_lastUsedBackendProgramId = m_backendProgramId;
+            g_lastUsedBackendProgramId = programToBind;
         }
 
         void BackendProgramObjectImpl::SetBaseInstance(Uint32 baseInstance) const {

MobileGL/MG_Backend/DirectGLES/Managers.h

@@ -270,18 +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
+            // Every desktop-only target is stored on an ES one; see MapToBackendTextureTarget.
-            // either native or emulated (1D -> 2D with height 1, 1D array -> 2D array, see
+            (void)target;
-            // MapToBackendTextureTarget). SPIRV-Cross already emits the matching ESSL samplers and
+            return true;
-            // coordinate padding for 1D/1D-array shaders.
-            return target != TextureTarget::TextureRectangle;
         }
 
-        // ES has no 1D targets: 1D textures are stored as 2D (height 1) and 1D arrays as 2D arrays
+        // ES has none of the desktop-only targets: 1D textures are stored as 2D (height 1), 1D
-        // (height 1, layers in depth). Must match SPIRV-Cross's ES 1D-as-2D shader emulation.
+        // 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::LowerRectImagesForEssl 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;
@@ -297,6 +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;
@@ -438,6 +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
@@ -576,6 +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();
@@ -587,9 +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; }
@@ -616,7 +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;
@@ -629,6 +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 =
+            const Flags<PixelFormatNormalizeOptionBit> forcedOptions = GetApplicablePixelFormatNormalizeOptions(
-                GetApplicablePixelFormatNormalizeOptions(requestedInternalFormat, GetForcedPixelFormatNormalizeOptions());
+                requestedInternalFormat, GetForcedPixelFormatNormalizeOptions() | extraOptions);
             if (forcedOptions) {
                 return forcedOptions;
             }
-            return GetApplicablePixelFormatNormalizeOptions(requestedInternalFormat,
+            return GetApplicablePixelFormatNormalizeOptions(
-                                                            GetDriverPixelFormatNormalizeOptions());
+                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);
@@ -342,6 +436,227 @@ namespace MobileGL::MG_Backend::DirectGLES {
             }
             return result;
         }
+
+        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
+            if (glslCode.find("sampler") == String::npos || glslCode.find("texture") == String::npos) {
+                return glslCode;
+            }
+
+            // 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;
+            }
+
+            // 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;
+                }
+                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;
+        }
+
+        String NormalizeRectSamplerCoordinates(const String& glslCode,
+                                               const Vector<String>& rectSamplerNames) {
+#ifdef TRACY_ENABLE
+            ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
+#endif
+            if (rectSamplerNames.empty() || glslCode.find("texture") == String::npos) {
+                return glslCode;
+            }
+
+            // Lookups whose argument 1 is a plain (non-projective) texel-space coordinate on a
+            // rectangle sampler. texelFetch* is absent on purpose: its coordinates are integer
+            // texels on the 2D target too, so it already lands in the right place.
+            static const char* const kRectCoordinateLookups[] = {
+                "textureGatherOffsets", "textureGatherOffset", "textureGather",
+                "textureOffset",        "texture",
+            };
+
+            String result = glslCode;
+            // Right to left, so the offsets of the not-yet-rewritten calls stay valid.
+            for (SizeT scan = result.size(); scan-- > 0;) {
+                if (result[scan] != 't') continue;
+                if (scan > 0 && IsIdentifierChar(result[scan - 1])) continue;
+
+                SizeT openParen = 0;
+                Bool matched = false;
+                for (const char* name : kRectCoordinateLookups) {
+                    const SizeT nameLength = std::strlen(name);
+                    if (result.compare(scan, nameLength, name) != 0) continue;
+                    const SizeT after = result.find_first_not_of(" \t", scan + nameLength);
+                    if (after == String::npos || result[after] != '(') continue;
+                    openParen = after;
+                    matched = true;
+                    break;
+                }
+                if (!matched) continue;
+
+                const Vector<SizeT> marks = SplitCallArguments(result, openParen);
+                if (marks.size() < 2) continue; // needs a sampler and a coordinate
+
+                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);
+                if (std::find(rectSamplerNames.begin(), rectSamplerNames.end(), samplerName) ==
+                    rectSamplerNames.end()) {
+                    continue;
+                }
+
+                // Wrap argument 1: (coord) / vec2(textureSize(sampler, 0)).
+                const SizeT coordStart = marks[0] + 1;
+                const SizeT coordEnd = marks[1];
+                result.insert(coordEnd, String(") / vec2(textureSize(") + samplerName + ", 0)))");
+                result.insert(coordStart, "((");
+            }
+            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
 
@@ -104,7 +109,35 @@ 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);
+        // 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);
+        // GL_TEXTURE_RECTANGLE is emulated on an ES 2D texture and LowerRectImagesForEssl
+        // rewrites the image type to match, but a rectangle lookup addresses texels
+        // directly while a 2D one addresses [0,1] - so every lookup that takes normalized
+        // coordinates has to divide by the texture's size. `rectSamplerNames` is the set of
+        // samplers the program declared as rectangle; texelFetch is left alone (its
+        // coordinates are unnormalized on both targets) and so is anything projective,
+        // which LowerRectImagesForEssl still declines outright.
+        String NormalizeRectSamplerCoordinates(const String& glslCode,
+                                               const Vector<String>& rectSamplerNames);
     } // 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>
 
@@ -525,10 +526,11 @@ 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};
         if (shaderSubgroupSupported && !MG_Config::Features.DisableSubgroup) {
             extensions.push_back(E_GL_KHR_shader_subgroup);
         }
@@ -634,6 +636,15 @@ 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;
@@ -796,6 +807,23 @@ 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

@@ -1250,10 +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;
@@ -1266,6 +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;
@@ -1334,6 +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;
@@ -1483,8 +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
@@ -1493,6 +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
 
@@ -1574,6 +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
@@ -1606,6 +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

@@ -123,6 +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) {
@@ -202,17 +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.commandBufferCount = commandBufferCount;
-        packet.submitInfo.pCommandBuffers = shouldSubmitCommandBuffer ? &packet.commandBuffer : nullptr;
+        packet.submitInfo.pCommandBuffers = commandBufferCount > 0 ? packet.commandBuffers : nullptr;
         packet.submitInfo.signalSemaphoreCount = 1;
         packet.submitInfo.pSignalSemaphores = &packet.signalSemaphore;
         return packet;
@@ -276,12 +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;
@@ -289,10 +340,20 @@ 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;
         }
+        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});

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};
         };
 
@@ -52,10 +54,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         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),
             // appended in submit order; freed once their submission is known
@@ -77,6 +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,
@@ -91,7 +109,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // can restart while the submitted buffer is still executing. Retired
         // buffers are freed after the slot's fence is next waited, or as soon
         // as their submission is observed complete.
-        VkResult RetireCurrentCommandBuffer();
+        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

MobileGL/MG_Backend/DirectVulkan/Renderer/ProgramFactory.cpp

@@ -12,10 +12,7 @@
 #include "MG_Util/ShaderTranspiler/ShaderCompiler.h"
 #include "MG_Util/ShaderTranspiler/SpvcSession.h"
 #include "MG_Util/ShaderTranspiler/Types.h"
-#include <cmath>
-#include <cstdio>
 #include <cstring>
-#include <unordered_set>
 #include <spirv-tools/libspirv.h>
 #include <spirv-tools/optimizer.hpp>
 #include <source/opt/build_module.h>
@@ -926,6 +923,163 @@ 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
@@ -1102,374 +1256,6 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return spvtools::Optimizer::PassToken(MakeUnique<ForceExplicitLod0SamplePass>());
         }
 
-        // TEMP-PERFDIAG: measure what fragment-stage fp32 costs on this GPU. Desktop GLSL carries
-        // no precision qualifiers, so everything reaches the driver as full fp32 while Adreno runs
-        // fp16 at twice the rate. Decorating every float-typed result in a fragment entry point
-        // with RelaxedPrecision is the blunt "all mediump" upper bound - it changes results, so it
-        // is a probe, not a shipping transform. Toggled by /sdcard/MG/exp_relaxed_precision.
-        class RelaxedPrecisionProbePass final : public spvtools::opt::Pass {
-        public:
-            const char* name() const override { return "relaxed-precision-probe"; }
-
-            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;
-
-                // Every 32-bit-float scalar/vector/matrix type in the module. Anything wider (f64)
-                // or narrower is left alone: RelaxedPrecision only has meaning for 32-bit floats.
-                std::unordered_set<Uint32> relaxableTypes;
-                for (auto& type : get_module()->types_values()) {
-                    const Uint32 typeId = type.result_id();
-                    if (typeId == 0) continue;
-                    switch (type.opcode()) {
-                    case spv::Op::OpTypeFloat:
-                        if (type.GetSingleWordInOperand(0) == 32) relaxableTypes.insert(typeId);
-                        break;
-                    case spv::Op::OpTypeVector:
-                    case spv::Op::OpTypeMatrix:
-                        if (relaxableTypes.count(type.GetSingleWordInOperand(0)) != 0) {
-                            relaxableTypes.insert(typeId);
-                        }
-                        break;
-                    default:
-                        break;
-                    }
-                }
-                if (relaxableTypes.empty()) return Status::SuccessWithoutChange;
-
-                Vector<Uint32> targets;
-                for (auto& function : *get_module()) {
-                    for (auto& block : function) {
-                        for (auto& inst : block) {
-                            const Uint32 resultId = inst.result_id();
-                            if (resultId == 0) continue;
-                            if (relaxableTypes.count(inst.type_id()) == 0) continue;
-                            targets.push_back(resultId);
-                        }
-                    }
-                }
-                if (targets.empty()) return Status::SuccessWithoutChange;
-
-                for (const Uint32 id : targets) {
-                    context()->get_decoration_mgr()->AddDecoration(
-                        id, static_cast<Uint32>(spv::Decoration::RelaxedPrecision));
-                }
-                context()->InvalidateAnalysesExceptFor(spvtools::opt::IRContext::kAnalysisNone);
-                return Status::SuccessWithChange;
-            }
-        };
-
-        // Relax fragment-stage arithmetic that provably came out of a texture read. Desktop GLSL
-        // has no precision qualifiers, so every fragment value reaches the driver as fp32 while
-        // Adreno runs fp16 at twice the rate - and a texel is at most 8 bits per channel, which
-        // fp16's 11-bit mantissa carries exactly. Seeding at image reads and propagating only
-        // through operations whose every input is already relaxed keeps everything the shader
-        // computes from other sources (screen coordinates, depth, wide-range uniforms) at full
-        // precision, which is where fp16 would actually go wrong: fp16 cannot even represent a
-        // 3044-pixel gl_FragCoord.x exactly.
-        class RelaxTextureDerivedPrecisionPass final : public spvtools::opt::Pass {
-        public:
-            const char* name() const override { return "relax-texture-derived-precision"; }
-
-            Status Process() override {
-                if (!IsFragmentEntryPoint()) return Status::SuccessWithoutChange;
-                // A shader that drives depth or coverage itself is out of scope: those values must
-                // stay exact, and proving which computations feed them is not worth it here.
-                if (WritesDepthOrSampleMask()) return Status::SuccessWithoutChange;
-
-                CollectRelaxableFloatTypes();
-                if (m_relaxableTypes.empty()) return Status::SuccessWithoutChange;
-
-                // Whitelisting from texture reads captures nothing in practice: MC's fragment
-                // shaders multiply every texel by an interpolated colour and a UBO value, so one
-                // un-relaxed operand vetoes the whole expression (measured: no fps change).
-                // Taint the few genuinely precision-critical sources instead and relax the rest.
-                std::unordered_set<Uint32> tainted;
-                CollectPrecisionCriticalSeeds(tainted);
-                Bool grew = true;
-                while (grew) {
-                    grew = false;
-                    for (auto& function : *get_module()) {
-                        for (auto& block : function) {
-                            for (auto& inst : block) {
-                                const Uint32 resultId = inst.result_id();
-                                if (resultId == 0 || tainted.count(resultId) != 0) continue;
-                                if (!AnyOperandTainted(inst, tainted)) continue;
-                                tainted.insert(resultId);
-                                grew = true;
-                            }
-                        }
-                    }
-                }
-
-                std::unordered_set<Uint32> relaxed;
-                for (auto& function : *get_module()) {
-                    for (auto& block : function) {
-                        for (auto& inst : block) {
-                            const Uint32 resultId = inst.result_id();
-                            if (resultId == 0 || tainted.count(resultId) != 0) continue;
-                            if (m_relaxableTypes.count(inst.type_id()) == 0) continue;
-                            relaxed.insert(resultId);
-                        }
-                    }
-                }
-                if (relaxed.empty()) return Status::SuccessWithoutChange;
-
-                for (const Uint32 id : relaxed) {
-                    context()->get_decoration_mgr()->AddDecoration(
-                        id, static_cast<Uint32>(spv::Decoration::RelaxedPrecision));
-                }
-                context()->InvalidateAnalysesExceptFor(spvtools::opt::IRContext::kAnalysisNone);
-                return Status::SuccessWithChange;
-            }
-
-        private:
-            std::unordered_set<Uint32> m_relaxableTypes;
-
-            Bool IsFragmentEntryPoint() const {
-                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) {
-                        return true;
-                    }
-                }
-                return false;
-            }
-
-            Bool WritesDepthOrSampleMask() const {
-                for (auto& annotation : get_module()->annotations()) {
-                    if (annotation.opcode() != spv::Op::OpDecorate) continue;
-                    if (static_cast<spv::Decoration>(annotation.GetSingleWordInOperand(1)) !=
-                        spv::Decoration::BuiltIn) {
-                        continue;
-                    }
-                    const auto builtIn = static_cast<spv::BuiltIn>(annotation.GetSingleWordInOperand(2));
-                    if (builtIn == spv::BuiltIn::FragDepth || builtIn == spv::BuiltIn::SampleMask) {
-                        return true;
-                    }
-                }
-                return false;
-            }
-
-            void CollectRelaxableFloatTypes() {
-                m_relaxableTypes.clear();
-                for (auto& type : get_module()->types_values()) {
-                    const Uint32 typeId = type.result_id();
-                    if (typeId == 0) continue;
-                    switch (type.opcode()) {
-                    case spv::Op::OpTypeFloat:
-                        if (type.GetSingleWordInOperand(0) == 32) m_relaxableTypes.insert(typeId);
-                        break;
-                    case spv::Op::OpTypeVector:
-                        if (m_relaxableTypes.count(type.GetSingleWordInOperand(0)) != 0) {
-                            m_relaxableTypes.insert(typeId);
-                        }
-                        break;
-                    default:
-                        break;
-                    }
-                }
-            }
-
-            void CollectImageReadSeeds(std::unordered_set<Uint32>& relaxed) const {
-                for (auto& function : *get_module()) {
-                    for (auto& block : function) {
-                        for (auto& inst : block) {
-                            const Uint32 resultId = inst.result_id();
-                            if (resultId == 0 || m_relaxableTypes.count(inst.type_id()) == 0) continue;
-                            // Interpolated user varyings seed too, or propagation dies at the
-                            // first `texel * vertexColour`: the load of an Input can never be
-                            // relaxed by the rule below (its operand is a pointer), so a single
-                            // varying vetoes every downstream operation. This is what ESSL's
-                            // mediump varyings already mean. Built-ins are excluded - gl_FragCoord
-                            // carries pixel coordinates that fp16 cannot represent exactly.
-                            if (inst.opcode() == spv::Op::OpLoad && IsNonBuiltInFragmentInput(inst)) {
-                                relaxed.insert(resultId);
-                                continue;
-                            }
-                            switch (inst.opcode()) {
-                            case spv::Op::OpImageSampleImplicitLod:
-                            case spv::Op::OpImageSampleExplicitLod:
-                            case spv::Op::OpImageSampleProjImplicitLod:
-                            case spv::Op::OpImageSampleProjExplicitLod:
-                            case spv::Op::OpImageSampleDrefImplicitLod:
-                            case spv::Op::OpImageSampleDrefExplicitLod:
-                            case spv::Op::OpImageFetch:
-                            case spv::Op::OpImageRead:
-                            case spv::Op::OpImageGather:
-                                relaxed.insert(resultId);
-                                break;
-                            default:
-                                break;
-                            }
-                        }
-                    }
-                }
-            }
-
-            // OpLoad straight out of a fragment Input variable that carries no BuiltIn decoration.
-            // Only a direct load counts: a load through an access chain could be indexing a
-            // structure whose other members are not interpolated colour data.
-            Bool IsNonBuiltInFragmentInput(const spvtools::opt::Instruction& load) const {
-                const Uint32 pointerId = load.GetSingleWordInOperand(0);
-                const auto* pointer = context()->get_def_use_mgr()->GetDef(pointerId);
-                if (pointer == nullptr || pointer->opcode() != spv::Op::OpVariable) return false;
-                if (static_cast<spv::StorageClass>(pointer->GetSingleWordInOperand(0)) !=
-                    spv::StorageClass::Input) {
-                    return false;
-                }
-                Bool isBuiltIn = false;
-                context()->get_decoration_mgr()->ForEachDecoration(
-                    pointerId, static_cast<Uint32>(spv::Decoration::BuiltIn),
-                    [&isBuiltIn](const spvtools::opt::Instruction&) { isBuiltIn = true; });
-                return !isBuiltIn;
-            }
-
-            // A float constant small enough that fp16 represents it without surprise. Colour math
-            // constants (0, 1, 0.5, 255, gamma exponents) all live here; anything larger is
-            // treated as unknown so it stops propagation.
-            Bool IsBoundedFloatConstant(Uint32 id) const {
-                const auto* constant = context()->get_constant_mgr()->FindDeclaredConstant(id);
-                if (constant == nullptr) return false;
-                if (const auto* scalar = constant->AsFloatConstant()) {
-                    const float value = scalar->GetFloat();
-                    return std::isfinite(value) && std::fabs(value) <= 1024.0f;
-                }
-                if (const auto* composite = constant->AsVectorConstant()) {
-                    for (const auto* component : composite->GetComponents()) {
-                        const auto* scalar = component->AsFloatConstant();
-                        if (scalar == nullptr) return false;
-                        const float value = scalar->GetFloat();
-                        if (!std::isfinite(value) || std::fabs(value) > 1024.0f) return false;
-                    }
-                    return true;
-                }
-                return false;
-            }
-
-            // Precision-critical sources: a built-in fragment input. gl_FragCoord is the one that
-            // matters - fp16 cannot represent a 3044-pixel x coordinate exactly, and anything
-            // derived from it (screen-space effects, manual depth reconstruction) would visibly
-            // quantise. Everything else a fragment shader reads is colour-range data.
-            void CollectPrecisionCriticalSeeds(std::unordered_set<Uint32>& tainted) const {
-                for (auto& function : *get_module()) {
-                    for (auto& block : function) {
-                        for (auto& inst : block) {
-                            if (inst.opcode() != spv::Op::OpLoad || inst.result_id() == 0) continue;
-                            if (IsBuiltInInputLoad(inst)) tainted.insert(inst.result_id());
-                        }
-                    }
-                }
-            }
-
-            Bool IsBuiltInInputLoad(const spvtools::opt::Instruction& load) const {
-                const Uint32 pointerId = load.GetSingleWordInOperand(0);
-                const auto* pointer = context()->get_def_use_mgr()->GetDef(pointerId);
-                if (pointer == nullptr || pointer->opcode() != spv::Op::OpVariable) return false;
-                if (static_cast<spv::StorageClass>(pointer->GetSingleWordInOperand(0)) !=
-                    spv::StorageClass::Input) {
-                    return false;
-                }
-                Bool isBuiltIn = false;
-                context()->get_decoration_mgr()->ForEachDecoration(
-                    pointerId, static_cast<Uint32>(spv::Decoration::BuiltIn),
-                    [&isBuiltIn](const spvtools::opt::Instruction&) { isBuiltIn = true; });
-                return isBuiltIn;
-            }
-
-            Bool AnyOperandTainted(const spvtools::opt::Instruction& inst,
-                                   const std::unordered_set<Uint32>& tainted) const {
-                const Uint32 operandCount = inst.NumInOperands();
-                for (Uint32 i = 0; i < operandCount; ++i) {
-                    const auto& operand = inst.GetInOperand(i);
-                    if (!spvIsIdType(operand.type)) continue;
-                    if (IsNonNumericOperand(inst, i)) continue;
-                    if (tainted.count(operand.words[0]) != 0) return true;
-                }
-                return false;
-            }
-
-            Bool AllValueOperandsRelaxed(const spvtools::opt::Instruction& inst,
-                                         const std::unordered_set<Uint32>& relaxed) const {
-                switch (inst.opcode()) {
-                // Pointer-typed plumbing: relaxing the loaded value would say nothing about the
-                // memory it came from, and the pointer operand can never be in the set.
-                case spv::Op::OpLoad:
-                case spv::Op::OpStore:
-                case spv::Op::OpAccessChain:
-                case spv::Op::OpInBoundsAccessChain:
-                case spv::Op::OpFunctionCall:
-                    return false;
-                default:
-                    break;
-                }
-
-                Bool sawValueOperand = false;
-                Bool allRelaxed = true;
-                const Uint32 operandCount = inst.NumInOperands();
-                for (Uint32 i = 0; i < operandCount; ++i) {
-                    const auto& operand = inst.GetInOperand(i);
-                    if (!spvIsIdType(operand.type)) continue; // literals: selectors, swizzle indices
-                    const Uint32 id = operand.words[0];
-                    // OpPhi's block labels, OpSelect's condition and OpExtInst's instruction-set id
-                    // are ids that carry no numeric precision; skip them rather than let them veto.
-                    if (IsNonNumericOperand(inst, i)) continue;
-                    sawValueOperand = true;
-                    if (relaxed.count(id) != 0) continue;
-                    if (IsBoundedFloatConstant(id)) continue;
-                    allRelaxed = false;
-                    break;
-                }
-                return sawValueOperand && allRelaxed;
-            }
-
-            static Bool IsNonNumericOperand(const spvtools::opt::Instruction& inst, Uint32 index) {
-                switch (inst.opcode()) {
-                case spv::Op::OpPhi:
-                    return (index % 2) == 1; // parent block labels
-                case spv::Op::OpSelect:
-                    return index == 0; // condition
-                case spv::Op::OpExtInst:
-                    return index == 0; // extended instruction set
-                default:
-                    return false;
-                }
-            }
-        };
-
-        // TEMP-PERFDIAG: A/B switch between the scoped transform and the all-float upper bound.
-        Bool PerfDiagRelaxAllPrecision() {
-            static const Bool enabled = [] {
-                std::FILE* probe = std::fopen("/sdcard/MG/exp_relaxed_precision_all", "rb");
-                if (probe == nullptr) return false;
-                std::fclose(probe);
-                MGLOG_I("[PERFDIAG] fragment RelaxedPrecision: ALL floats (upper-bound probe)");
-                return true;
-            }();
-            return enabled;
-        }
-
-        // TEMP-PERFDIAG: lets a run turn the transform off entirely for an A/B baseline.
-        Bool PerfDiagRelaxedPrecisionEnabled() {
-            static const Bool disabled = [] {
-                std::FILE* probe = std::fopen("/sdcard/MG/exp_no_relaxed_precision", "rb");
-                if (probe == nullptr) return false;
-                std::fclose(probe);
-                MGLOG_I("[PERFDIAG] fragment RelaxedPrecision DISABLED");
-                return true;
-            }();
-            return !disabled;
-        }
-
         Bool TransformSpirvForExplicitLod0Sampling(const Vector<Uint>& input, Vector<Uint>& output) {
             if (input.empty()) {
                 output.clear();
@@ -1499,32 +1285,36 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return spvtools::Optimizer::PassToken(MakeUnique<GlToVulkanPositionFixPass>(transformFlags));
         }
 
-        // TEMP-PERFDIAG
+        Bool TransformSpirvForXfbCapture(const Vector<Uint>& input, Vector<Uint>& output,
-        Bool TransformSpirvForRelaxedPrecisionProbe(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: relaxed-precision probe: %s", message != nullptr ? message : "");
+                MGLOG_E("Vulkan: xfb capture pass: %s", message != nullptr ? message : "");
             });
-            // SSA promotion first: glslang emits function-local variables with stores and loads,
+            optimizer.RegisterPass(spvtools::Optimizer::PassToken(
-            // and a load can never be relaxed (its operand is a pointer), so without this the
+                MakeUnique<XfbCaptureDecoratePass>(Move(varyings), Move(strides))));
-            // propagation below dies at the first temporary.
+
-            optimizer.RegisterPass(spvtools::CreateLocalMultiStoreElimPass());
-            if (PerfDiagRelaxAllPrecision()) {
-                optimizer.RegisterPass(spvtools::Optimizer::PassToken(MakeUnique<RelaxedPrecisionProbePass>()));
-            } else {
-                optimizer.RegisterPass(
-                    spvtools::Optimizer::PassToken(MakeUnique<RelaxTextureDerivedPrecisionPass>()));
-            }
             const Bool success = optimizer.Run(input.data(), input.size(), &output, options);
             if (!success) {
-                MGLOG_E("Vulkan: relaxed-precision probe failed; keeping the original module");
+                MGLOG_E("Vulkan: xfb capture decoration pass failed; keeping the original module");
                 output = input;
             }
             return success;
@@ -2575,7 +2365,17 @@ 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;
             }
@@ -2588,15 +2388,6 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 }
             }
 
-            if ((flags & ProgramFactory::CompileOptionBit::RelaxedFragmentPrecision) &&
-                PerfDiagRelaxedPrecisionEnabled() && shaders[i] &&
-                shaders[i]->GetShaderStage() == ShaderStage::Fragment) {
-                Vector<Uint> relaxedSpirv;
-                if (TransformSpirvForRelaxedPrecisionProbe(moduleSpirvs[i], relaxedSpirv)) {
-                    moduleSpirvs[i] = Move(relaxedSpirv);
-                }
-            }
-
             // 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

MobileGL/MG_Backend/DirectVulkan/Renderer/ProgramFactory.h

@@ -47,11 +47,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // 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,
-            // Fragment arithmetic may run at relaxed (fp16) precision. Only requested for draws
+            // Decorates the last vertex-processing stage's captured varyings with
-            // where every sampled texture and every colour attachment is an 8-bit-or-less
+            // XfbBuffer/XfbStride/Offset (VK_EXT_transform_feedback). Set only for draws
-            // normalized format, so nothing the shader reads or writes carries more precision
+            // recorded while GL transform feedback is active, so plain draws keep the
-            // than fp16 already represents exactly.
+            // undecorated variant.
-            RelaxedFragmentPrecision = 1 << 6,
+            XfbCapture = 1 << 6,
         };
         using CompileOptionFlags = Flags<CompileOptionBit>;
         using HashType = Uint64;

MobileGL/MG_Backend/DirectVulkan/Renderer/SwapchainObject.cpp

@@ -262,6 +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);
@@ -433,9 +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

@@ -52,6 +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);
@@ -77,5 +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

@@ -16,9 +16,9 @@
 #include "MG_Util/Converters/GLToMG/TextureEnumConverter.h"
 #include "MG_Util/Converters/MGToStr/FramebufferEnumConverter.h"
 #include "MG_Util/Converters/MGToVk/TextureEnumConverter.h"
-#include <vulkan/utility/vk_format_utils.h>
 #include "MG_Util/Metrics/TextureMetrics.h"
 #include <Config.h>
+#include <algorithm>
 #include <cstdio>
 #include <cstdlib>
 #include <cstring>
@@ -205,6 +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) {
@@ -447,64 +448,6 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         return outImageInfo.sampler != VK_NULL_HANDLE;
     }
 
-    namespace {
-        // fp16 carries an 11-bit mantissa, so an 8-bit normalized channel round-trips exactly.
-        // Anything wider - 16-bit normalized, half float, full float, and every packed HDR
-        // encoding - holds precision or range that relaxing the arithmetic would throw away.
-        Bool IsLowPrecisionNormalizedFormat(VkFormat format) {
-            if (format == VK_FORMAT_UNDEFINED) return false;
-            if (!vkuFormatIsUNORM(format) && !vkuFormatIsSNORM(format) && !vkuFormatIsSRGB(format)) {
-                return false;
-            }
-            const struct VKU_FORMAT_INFO info = vkuGetFormatInfo(format);
-            for (Uint32 i = 0; i < info.component_count; ++i) {
-                if (info.components[i].size > 8) return false;
-            }
-            return info.component_count > 0;
-        }
-    } // namespace
-
-    Bool UniformManager::DrawTargetIsLowPrecision(const MG_State::GLState::FramebufferObject* drawFramebuffer) {
-        // Default framebuffer: the swapchain is an 8-bit normalized surface.
-        if (drawFramebuffer == nullptr) return true;
-
-        Bool sawColour = false;
-        for (Int i = static_cast<Int>(FramebufferAttachmentType::Color0);
-             i < static_cast<Int>(FramebufferAttachmentType::FramebufferAttachmentTypeCount);
-             ++i) {
-            const auto& attachment =
-                drawFramebuffer->GetAttachment(static_cast<FramebufferAttachmentType>(i));
-            VkFormat format = VK_FORMAT_UNDEFINED;
-            if (const auto& texture = attachment.GetTexture()) {
-                format = MG_Util::ConvertTextureInternalFormatToVkEnum(texture->GetFormat());
-            } else if (const auto& renderbuffer = attachment.GetRenderbuffer()) {
-                format = MG_Util::ConvertTextureInternalFormatToVkEnum(
-                    renderbuffer->GetInternalFormat());
-            } else {
-                continue;
-            }
-            if (!IsLowPrecisionNormalizedFormat(format)) return false;
-            sawColour = true;
-        }
-        return sawColour;
-    }
-
-    Bool UniformManager::ProgramSamplesOnlyLowPrecisionTextures(
-        const MG_State::GLState::ProgramObject& program, const ProgramFactory::VkProgramObject& programObj) {
-        for (Uint32 binding = 0; binding < programObj.bindingKinds.size(); ++binding) {
-            if (programObj.bindingKinds[binding] != ProgramFactory::DescriptorBindingKind::CombinedImageSampler) {
-                continue;
-            }
-            const auto* texture = ResolveSamplerTextureRaw(program, programObj, binding);
-            // An unresolvable binding is unknown territory, not licence to relax.
-            if (texture == nullptr) return false;
-            const VkFormat format =
-                MG_Util::ConvertTextureInternalFormatToVkEnum(texture->GetFormat());
-            if (!IsLowPrecisionNormalizedFormat(format)) return false;
-        }
-        return true;
-    }
-
     Bool UniformManager::ProgramSamplesOnlySingleLevelTextures(
         const MG_State::GLState::ProgramObject& program, const ProgramFactory::VkProgramObject& programObj) {
         Bool sawSampler = false;
@@ -707,6 +650,14 @@ 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'",
@@ -1248,16 +1199,47 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                         bufferInfo.range = ubo.range;
                         dynOffset = static_cast<Uint32>(ubo.dynamicOffset);
                     } else {
-                        BufferSlice slice{};
+                        // Global-UBO slice reuse (see GlobalUboSliceMemo): unchanged
-                        if (!m_bufferManager->UploadTransient(BufferKind::Uniform, frameIndex, ubo.payload,
+                        // uniform bytes re-use the slice already uploaded this frame.
-                                                              ubo.payloadSize, m_minDynamicOffsetAlignment, slice)) {
+                        const Bool isGlobalUbo =
-                            MOBILEGL_ASSERT(false, "UniformDescriptorBinder::BindProgramUniformBuffers failed: UBO upload failed on binding %u element %u",
+                            programObj.globalUboBinding == static_cast<Int>(binding) && element == 0;
-                                    binding, element);
+                        const Uint64 uboFrameSerial = m_bufferManager->GetFrameSerial();
-                            return false;
+                        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,
@@ -1396,8 +1378,34 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             m_hasLastDescriptor = cacheable;
         }
 
-        vkCmdBindDescriptorSets(commandBuffer, bindPoint, programObj.pipelineLayout, 0, 1,
+        // Skip the driver call when this exact binding is already live on the
-                                &descriptorSet, static_cast<Uint32>(dynamicOffsets.size()), dynamicOffsets.data());
+        // 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 +39,9 @@ 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
@@ -76,15 +79,6 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // 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.
-        // True when every texture this program samples is an 8-bit-or-less normalized format, so
-        // relaxing the fragment stage to fp16 cannot lose a bit the texel ever carried. Says
-        // nothing about the render target - the caller must check that too.
-        static Bool ProgramSamplesOnlyLowPrecisionTextures(const MG_State::GLState::ProgramObject& program,
-                                                          const ProgramFactory::VkProgramObject& programObj);
-        // True when every colour attachment the draw writes is an 8-bit-or-less normalized
-        // format (nullptr = default framebuffer, which is). Blending happens at attachment
-        // precision, so a wider target must keep the fragment stage at full precision.
-        static Bool DrawTargetIsLowPrecision(const MG_State::GLState::FramebufferObject* drawFramebuffer);
         static Bool ProgramSamplesOnlySingleLevelTextures(const MG_State::GLState::ProgramObject& program,
                                                           const ProgramFactory::VkProgramObject& programObj);
 
@@ -194,6 +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

MobileGL/MG_Backend/DirectVulkan/Renderer/VertexInputStateFactory.cpp

@@ -58,15 +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()) {
-            it->second.lastUsedFrameBoundary = m_frameBoundaryCounter;
+            it->second->lastUsedFrameBoundary = m_frameBoundaryCounter;
-            return it->second;
+            return *it->second;
         }
 
         VertexInputStateBuilder builder;
@@ -172,11 +184,37 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         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.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)));
+        }
+        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);
@@ -205,8 +243,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
 
         for (auto it = m_cache.begin(); it != m_cache.end();) {
-            if (m_frameBoundaryCounter - it->second.lastUsedFrameBoundary > kRetireAgeBoundaries) {
+            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;
             }

MobileGL/MG_Backend/DirectVulkan/Renderer/VertexInputStateFactory.h

@@ -27,9 +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).
-            Uint64 lastUsedFrameBoundary = 0;
+            // 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;
@@ -41,6 +50,9 @@ 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;
             VkPipelineVertexInputStateCreateInfo state{
                 VK_STRUCTURE_TYPE_PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO
             };
@@ -80,9 +92,19 @@ 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
 
@@ -465,7 +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;

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).

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) {
+        // GL promises "at least the requested samples", so a non-power-of-two
-        case 1:
+        // 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;
+        if (requestedSamples > 64) {
-            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:
             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;
@@ -231,10 +225,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         if (resource.image == VK_NULL_HANDLE && resource.view == VK_NULL_HANDLE) {
             return;
         }
-        m_deferredRenderbufferReleases.push_back({resource.image, resource.allocation, resource.view, m_frameCounter});
+        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) {
@@ -249,6 +245,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             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);
             }
@@ -304,7 +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),
@@ -314,6 +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 ||
@@ -351,6 +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(
@@ -385,6 +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;
@@ -481,12 +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();
@@ -560,9 +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);
@@ -617,14 +721,49 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             combineFramebufferAttachmentObjHash(drawbuf);
         }
 
-        combineFramebufferAttachmentObjHash(FramebufferAttachmentType::Depth);
+        // The depth-less default-FBO flavor omits the depth/stencil attachment
-        combineFramebufferAttachmentObjHash(FramebufferAttachmentType::Stencil);
+        // 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) {
@@ -674,6 +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()) {
@@ -682,7 +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()) {
@@ -699,10 +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;
@@ -789,8 +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
@@ -825,7 +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);
 
@@ -894,6 +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,
@@ -907,12 +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,
                         });
@@ -976,6 +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);
@@ -994,6 +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;
@@ -1077,6 +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,
                 });
@@ -1122,6 +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;
@@ -1330,6 +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) {
@@ -1382,11 +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;
@@ -188,8 +193,22 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         HashType ComputeHash(
             const MG_State::GLState::FramebufferObject& fbo,
             Uint32 swapchainImageIndex,
-            Bool includePendingClear = true);
+            Bool includePendingClear = true,
-        RenderPassEntry& GetOrCreateRenderPass(const MG_State::GLState::FramebufferObject& fbo, Uint32 swapchainImageIndex);
+            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,
@@ -219,6 +238,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // 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 /
@@ -231,6 +257,10 @@ 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 {
@@ -244,6 +274,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             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;
@@ -277,12 +310,16 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             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;

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) {
+        // GL promises "at least the requested samples", so a non-power-of-two
-        case 1:
+        // 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;
+        if (requestedSamples > 64) {
-            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:
             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) {
@@ -607,7 +597,11 @@ 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();
@@ -629,6 +623,7 @@ 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
@@ -659,6 +654,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
         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) {
@@ -714,45 +712,63 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             }
         }
 
-        auto aliveIt = m_aliveObjects.find(identity);
+        // Cross-draw memo probe (see SyncedTextureMemoEntry): skips both map
-        if (aliveIt != m_aliveObjects.end() && aliveIt->second.expired()) {
+        // lookups and the (re)registration path for repeat-bound textures.
-            EraseTrackedTexture(aliveIt->first);
+        TextureResource* resourcePtr = nullptr;
-            aliveIt = m_aliveObjects.end();
+        for (Uint32 i = 0; i < kSyncedTextureMemoSize; ++i) {
-        }
+            const SyncedTextureMemoEntry& memo = m_syncedTextureMemo[i];
-
+            if (memo.texture == &texture && memo.lifetimeId == identity.lifetimeId &&
-        // Only (re)register and prune when this (texture, lifetime) pair is new: stale
+                memo.eraseEpoch == m_resourceEraseEpoch) {
-        // aliases can only come into existence through an address reuse, which by
+                resourcePtr = memo.resource;
-        // construction introduces a new identity. Doing this unconditionally made every
+                break;
-        // 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 (resourcePtr == nullptr) {
-        if (it == m_textureResources.end()) {
+            auto aliveIt = m_aliveObjects.find(identity);
-            TextureResource initial{};
+            if (aliveIt != m_aliveObjects.end() && aliveIt->second.expired()) {
-            auto [insertIt, _] = m_textureResources.emplace(identity, Move(initial));
+                EraseTrackedTexture(aliveIt->first);
-            it = insertIt;
+                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;
         }
@@ -766,11 +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) {
@@ -814,7 +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);
         }
 
@@ -823,6 +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()) {
@@ -833,7 +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",
@@ -1049,6 +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));
@@ -1076,6 +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;
@@ -1160,6 +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;
     }
 
@@ -1189,6 +1225,8 @@ 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;
     }
 
@@ -1208,6 +1246,20 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                !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 TextureIdentity identity = MakeTextureIdentity(&texture);
         const auto it = m_textureResources.find(identity);
@@ -1406,11 +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;
@@ -1420,8 +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,
@@ -1473,6 +1546,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             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 |
@@ -1485,6 +1566,44 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             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()) &&
                                 resource.extent.height == static_cast<Uint32>(texelSize.y()) &&
@@ -1614,8 +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),
+        // Soft failure like the unsupported-sample-count path above: a driver can pass the
-                  "vmaCreateImage(texture)");
+        // 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;
@@ -1683,6 +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();
@@ -1884,17 +2038,119 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return true;
         }
 
-        // Combined depth-stencil images need per-aspect de-interleaved copies (VkBufferImageCopy
+        // Combined depth-stencil images need per-aspect copies (VkBufferImageCopy aspectMask
-        // aspectMask must have exactly one bit set). Until that is implemented, skip the upload
+        // must have exactly one bit set), so de-interleave the shadow's GL wire format into
-        // instead of recording an invalid command buffer that kills the process.
+        // 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)) {
+        const Bool isCombinedDepthStencil =
-            MGLOG_E("UploadDirtyMipLevels: skipping unimplemented depth-stencil data upload for textureId=%d",
+            (uploadAspectMask & VK_IMAGE_ASPECT_DEPTH_BIT) && (uploadAspectMask & VK_IMAGE_ASPECT_STENCIL_BIT);
-                    mipmapTexture.GetExternalIndex());
+        if (isCombinedDepthStencil) {
-            for (const auto& item : uploadItems) {
+            const Bool srcIsD24S8 = outResource.format == VK_FORMAT_D24_UNORM_S8_UINT;
-                mipmapTexture.MarkStorageDirty(item.target, item.level, false);
+            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;
@@ -1947,7 +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;
@@ -1955,10 +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);
         }
@@ -1989,11 +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)");
+        // Do NOT wait the fence here: this submit sits behind the previous
-        vkDestroyFence(m_device, uploadFence, nullptr);
+        // frame's rendering on the queue, so a synchronous wait stalls the CPU
-        vkFreeCommandBuffers(m_device, m_commandPool, 1, &commandBuffer);
+        // until the GPU drains - a per-frame vkQueueWaitIdle for any workload
-
+        // with animated textures. Ordering against the current frame's draws is
-        vmaDestroyBuffer(m_allocator, stagingBuffer, stagingAllocation);
+        // 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;
@@ -64,12 +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;
             }
         };
 
@@ -80,6 +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;
             }
         };
@@ -172,6 +181,13 @@ public:
         // 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;
@@ -207,6 +223,7 @@ public:
             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);
         }
@@ -307,6 +324,21 @@ 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
@@ -318,6 +350,10 @@ public:
     // 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
@@ -364,6 +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);
@@ -394,6 +433,11 @@ 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);
@@ -423,6 +467,23 @@ 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;
@@ -430,7 +491,21 @@ private:
     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 {
@@ -151,6 +155,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         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);
 
@@ -188,6 +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,
@@ -273,6 +304,15 @@ 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
@@ -391,6 +431,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         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
@@ -443,6 +492,66 @@ 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;
+        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);
+        // 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;
@@ -455,14 +564,30 @@ 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;
+        // Small N-way pipeline-resolution memo (round-robin replacement). A
-        GLenum m_lastPipelineMode = 0;
+        // single-entry memo thrashed on draw sequences that alternate a few
-        Uint64 m_lastPipelineProgramHash = 0;
+        // pipelines (GUI text/quad program ping-pong), paying the full
-        Uint64 m_lastPipelineVertexInputHash = 0;
+        // payload-hash lookup per draw; eight entries cover such working sets
-        Uint64 m_lastPipelineRenderPassHash = 0;
+        // while keeping the hit path a trivial linear scan.
-        Uint m_lastPipelineRenderStateVersion = 0;
+        struct PipelineMemoEntry {
-        ProgramFactory::CompileOptionFlags m_lastPipelineTransformFlags = {};
+            GLenum mode = 0;
-        VkPipeline m_lastPipelineResult = VK_NULL_HANDLE;
+            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;
@@ -491,9 +616,61 @@ 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;
@@ -618,6 +795,14 @@ 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

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/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 +689,321 @@ 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 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 +11,21 @@
 
 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 DeleteTransformFeedbacks(GLsizei n, const GLuint* ids);
+    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_HEAD(void, BeginTransformFeedback, GLenum primitiveMode) DECLARE_GL_FUNCTION_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, 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_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_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, 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_HEAD(void, BindTransformFeedback, GLenum target, GLuint id) DECLARE_GL_FUNCTION_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_HEAD(void, DeleteTransformFeedbacks, GLsizei n, const GLuint* ids) DECLARE_GL_FUNCTION_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)
+DECLARE_GL_FUNCTION_HEAD(void, GenTransformFeedbacks, GLsizei n, GLuint* ids) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GenTransformFeedbacks, n, ids)
-// Transform feedback objects are not implemented, so no name is ever a live object. The shared
+DECLARE_GL_FUNCTION_HEAD(GLboolean, IsTransformFeedback, GLuint id) DECLARE_GL_FUNCTION_END(GLboolean, IsTransformFeedback, id)
-// stub returns (type)1, telling a probing caller that every id it invents already exists; GL_FALSE
+DECLARE_GL_FUNCTION_HEAD(void, PauseTransformFeedback) DECLARE_GL_FUNCTION_END_NO_RETURN(void, PauseTransformFeedback)
-// is both truthful and what the spec requires for a name that was never generated.
+DECLARE_GL_FUNCTION_HEAD(void, ResumeTransformFeedback) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ResumeTransformFeedback)
-MOBILEGL_GL_API GLboolean glIsTransformFeedback(GLuint id) {
+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)
-    MGLOG_W("Stub function: %s(...)", __FUNCTION__);
+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)
-    return GL_FALSE;
+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, 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_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)
@@ -430,7 +424,7 @@ DECLARE_GL_FUNCTION_STUB_HEAD(void, GetnUniformfv, GLuint program, GLint locatio
 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)
@@ -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_HEAD(void, Uniform1d, GLint location, GLdouble x) DECLARE_GL_FUNCTION_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_HEAD(void, Uniform2d, GLint location, GLdouble x, GLdouble y) DECLARE_GL_FUNCTION_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_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_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_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_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_HEAD(void, Uniform1dv, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_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_HEAD(void, Uniform2dv, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_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_HEAD(void, Uniform3dv, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_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_HEAD(void, Uniform4dv, GLint location, GLsizei count, const GLdouble* value) DECLARE_GL_FUNCTION_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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, 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_HEAD(void, DrawTransformFeedback, GLenum mode, GLuint id) DECLARE_GL_FUNCTION_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_HEAD(void, DrawTransformFeedbackStream, GLenum mode, GLuint id, GLuint stream) DECLARE_GL_FUNCTION_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_HEAD(void, BeginQueryIndexed, GLenum target, GLuint index, GLuint id) DECLARE_GL_FUNCTION_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_HEAD(void, EndQueryIndexed, GLenum target, GLuint index) DECLARE_GL_FUNCTION_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_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_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_HEAD(void, ProgramUniform1d, GLuint program, GLint location, GLdouble v0) DECLARE_GL_FUNCTION_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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_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, 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_HEAD(void, DrawTransformFeedbackInstanced, GLenum mode, GLuint id, GLsizei instancecount) DECLARE_GL_FUNCTION_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, 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)

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;
+                    const Bool singleTarget = capabilityTargetIndex < MG_Backend::kFormatCapabilityTargetCount;
-                         ++targetIndex) {
+                    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>(
@@ -1422,8 +1658,8 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (HasNonRenderableColorAttachment(*framebufferObject)) {
             return GL_FRAMEBUFFER_UNSUPPORTED;
         }
-        if (IsActiveBackendDirectVulkan() &&
+        if (ActiveBackendRejectsDistinctDepthStencil() &&
-            IsUnsupportedFramebufferForDirectVulkan(*framebufferObject)) {
+            HasUnsupportedDistinctDepthStencilAttachments(*framebufferObject)) {
             return GL_FRAMEBUFFER_UNSUPPORTED;
         }
         return GL_FRAMEBUFFER_COMPLETE;
@@ -1450,8 +1686,8 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (HasNonRenderableColorAttachment(*framebufferObject)) {
             return GL_FRAMEBUFFER_UNSUPPORTED;
         }
-        if (IsActiveBackendDirectVulkan() &&
+        if (ActiveBackendRejectsDistinctDepthStencil() &&
-            IsUnsupportedFramebufferForDirectVulkan(*framebufferObject)) {
+            HasUnsupportedDistinctDepthStencilAttachments(*framebufferObject)) {
             return GL_FRAMEBUFFER_UNSUPPORTED;
         }
         return GL_FRAMEBUFFER_COMPLETE;
@@ -1468,20 +1704,40 @@ 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 +1783,10 @@ 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>(

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;
+                if (attachment.IsRenderbuffer() && attachment.GetRenderbuffer()) {
-                maxSamples = std::max(maxSamples, static_cast<GLint>(attachment.GetRenderbuffer()->GetSamples()));
+                    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;
@@ -1004,6 +1030,11 @@ 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;
@@ -1886,6 +1917,21 @@ 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 +1947,25 @@ 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 +2022,15 @@ 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;
@@ -317,11 +327,16 @@ 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 +620,18 @@ 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 +650,14 @@ 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 +832,8 @@ 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 +843,46 @@ 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 +896,13 @@ namespace MobileGL::MG_Impl::GLImpl {
     }
 
     GLboolean IsProgram_State(GLuint program) {
-        /* FIXME: Handle situations that:
+        // Deletion-flagged names stay valid while the object is still GL-visible (program in
-         * A program object marked for deletion with glDeleteProgram but still in use as part of current
+        // use, shader attached), so name validity is exactly the Is* answer.
-         * rendering state is still considered a program object and glIsProgram will return GL_TRUE.
-         */
         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 +912,18 @@ 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 +967,18 @@ 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 +1101,38 @@ 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 +1941,312 @@ 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);
     }
@@ -2217,4 +2642,150 @@ 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>
@@ -27,6 +28,8 @@ namespace MobileGL::MG_Impl::GLImpl {
             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 +44,11 @@ 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 +134,11 @@ 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.
@@ -180,7 +193,24 @@ 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) {
@@ -204,10 +234,15 @@ namespace MobileGL::MG_Impl::GLImpl {
     }
 
     void BeginQuery(GLenum target, GLuint id) {
-        if (target != GL_TIME_ELAPSED) {
+        const Bool isTransformFeedbackQuery =
-            // Only GL_TIME_ELAPSED timer queries are implemented (occlusion and
+            target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN || target == GL_PRIMITIVES_GENERATED;
-            // primitive queries remain stubs); GL_TIMESTAMP is not a valid
+        const Bool isOcclusionQuery =
-            // BeginQuery target either.
+            (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 +256,13 @@ 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 +278,72 @@ 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;
+        if (isTransformFeedbackQuery) {
-        queryObject->backendHandle =
+            // Prefer real GPU transform-feedback queries (exact with geometry shaders);
-            (!TimerQueryDisabled() && beginTimeElapsedQuery) ? beginTimeElapsedQuery() : nullptr;
+            // the CPU accounting delta stays as the fallback when the backend lacks them.
-        g_activeTimeElapsedQueryId = id;
+            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) {
+        GLuint& activeQueryId = isTransformFeedbackQuery
-            RecordQueryError(ErrorCode::InvalidOperation, __FUNCTION__, "No query is active on GL_TIME_ELAPSED.");
+            ? (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 +389,25 @@ 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
+            switch (target) {
-            // never are, and other targets remain unimplemented.
+            case GL_TIME_ELAPSED:
-            *params = target == GL_TIME_ELAPSED ? static_cast<GLint>(g_activeTimeElapsedQueryId) : 0;
+                *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 +415,13 @@ 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 +467,42 @@ 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

@@ -16,6 +16,9 @@ namespace MobileGL::MG_Impl::GLImpl {
     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/Texture/GL_Texture.cpp

@@ -2820,7 +2820,22 @@ 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);

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/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 +753,95 @@ 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;
+        }
     } // namespace GLState
 
     // Leak-at-exit storage; see GlobalObjects.cpp.

MobileGL/MG_State/GLState/Core.h

@@ -138,6 +138,8 @@ 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 +213,92 @@ 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;
+
                 // Framebuffer
                 void GenFramebufferNames(Uint number, Vector<Uint>& framebuffers);
                 const SharedPtr<FramebufferObject>& GetFramebufferObject(Uint index);
@@ -243,6 +331,46 @@ 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

@@ -377,6 +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(); }
@@ -465,6 +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
@@ -475,6 +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();
@@ -539,6 +604,7 @@ 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;
 
@@ -558,5 +624,18 @@ namespace MobileGL::MG_State::GLState {
         mutable Uint32 m_backendHashMemoVersion = ~0u;
         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();
+            // A program in use is only FLAGGED: its name (and every program query) stays
-            m_programIndexGenerator.Delete(program);
+            // valid until it stops being current, at which point UseProgram finishes the job.
-            for (const auto& shader : attachedShaders) {
+            if (programObject == m_currentProgram) return;
-                const Uint shaderName = shader->GetExternalIndex();
+            DestroyProgramSlot(program);
-                if (CheckIndexAvail(shaderName, m_shaderObjects) && m_shaderObjects[shaderName] == shader) {
+        }
-                    ReleaseShaderNameIfOrphaned(shaderName);
+    }
-                }
+
+    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;
+        if (CheckIndexAvail(program, m_programObjects)) {
-        m_currentProgram = m_programObjects[program];
+            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 +348,58 @@ namespace MobileGL {
 
             // TODO: add other texture types as needed
 
+        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

@@ -156,6 +156,14 @@ 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);
+
     inline const TextureObjectMipmap* AsMipmapTexture(const ITextureObject* texture) {
         return (texture && texture->GetStorageType() == TextureStorageType::Mipmap)
                    ? static_cast<const TextureObjectMipmap*>(texture)

MobileGL/MG_State/GLState/VertexArrayState/VertexArrayObject.h

@@ -104,6 +104,23 @@ 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 +154,9 @@ 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
+// used to be forced to. A legacy shader using 420-era syntax without the matching #extension line
-// #extension line is accepted by real drivers, so CompileShader retries it at 460 instead of failing.
+// is accepted by real drivers, so CompileShader retries the normalized source at 460 rather than
-TEST_F(ProgramUtilTest, CompileShaderRetriesAt460WhenLegacyVersionRejects420Syntax) {
+// 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;
+varying vec2 texCoord;
-out vec4 fragColor;
 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

@@ -197,7 +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;
@@ -397,7 +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;
@@ -516,6 +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;
 
@@ -638,6 +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;
 

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
+    // Every desktop-only target is stored on an ES one (MapToBackendTextureTarget): 1D and
-    // SPIRV-Cross's ES 1D-as-2D shader emission; only rectangle textures stay unsupported.
+    // 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 LowerRectImagesForEssl 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;
@@ -1007,6 +1073,19 @@ 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;
@@ -1120,6 +1134,9 @@ 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) {
@@ -171,6 +191,7 @@ 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);
@@ -261,6 +282,7 @@ 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

@@ -68,6 +68,9 @@ 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:

MobileGL/MG_Util/Converters/GLToMG/TextureEnumConverter.cpp

@@ -251,6 +251,8 @@ namespace MobileGL {
                 return TextureInternalFormat::Depth24Stencil8;
             case GL_DEPTH32F_STENCIL8:
                 return TextureInternalFormat::Depth32FStencil8;
+            case GL_STENCIL_INDEX8:
+                return TextureInternalFormat::StencilIndex8;
             case GL_DEPTH_COMPONENT:
                 return TextureInternalFormat::DepthComponent;
             case GL_DEPTH_STENCIL:

MobileGL/MG_Util/Converters/MGToGL/BufferEnumConverter.cpp

@@ -73,7 +73,7 @@ namespace MobileGL {
             }
         }
 
-        GLbitfield ConvertBufferMappingAccessToGLEnum(BufferMappingAccessBit access) {
+        GLbitfield ConvertBufferMappingAccessToGLEnum(Flags<BufferMappingAccessBit> access) {
             GLbitfield result = 0;
             if (access & BufferMappingAccessBit::Read) result |= GL_MAP_READ_BIT;
             if (access & BufferMappingAccessBit::Write) result |= GL_MAP_WRITE_BIT;

MobileGL/MG_Util/Converters/MGToGL/BufferEnumConverter.h

@@ -14,6 +14,6 @@ namespace MobileGL {
     namespace MG_Util {
         GLenum ConvertBufferTargetToGLEnum(BufferTarget bufferTarget);
         GLenum ConvertBufferUsageToGLEnum(BufferUsage usage);
-        GLbitfield ConvertBufferMappingAccessToGLEnum(BufferMappingAccessBit access);
+        GLbitfield ConvertBufferMappingAccessToGLEnum(Flags<BufferMappingAccessBit> access);
     } // namespace MG_Util
 } // namespace MobileGL

MobileGL/MG_Util/Converters/MGToGL/TextureEnumConverter.cpp

@@ -233,6 +233,8 @@ namespace MobileGL {
                 return GL_DEPTH24_STENCIL8;
             case TextureInternalFormat::Depth32FStencil8:
                 return GL_DEPTH32F_STENCIL8;
+            case TextureInternalFormat::StencilIndex8:
+                return GL_STENCIL_INDEX8;
             case TextureInternalFormat::DepthComponent32:
                 return GL_DEPTH_COMPONENT32;
             case TextureInternalFormat::DepthStencil:

MobileGL/MG_Util/Converters/MGToStr/TextureEnumConverter.cpp

@@ -234,6 +234,8 @@ namespace MobileGL {
                 return "Depth24Stencil8";
             case TextureInternalFormat::Depth32FStencil8:
                 return "Depth32FStencil8";
+            case TextureInternalFormat::StencilIndex8:
+                return "StencilIndex8";
             case TextureInternalFormat::Red:
                 return "Red";
             case TextureInternalFormat::RG:

MobileGL/MG_Util/Converters/MGToVk/RenderStateEnumConverter.cpp

@@ -25,6 +25,19 @@ namespace MobileGL {
             case GL_TRIANGLE_FAN:
                 return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_FAN;
             case GL_LINE_LOOP:
+                // DrawArrays/DrawElements rewrite line loops into closed indexed
+                // strips; entry points without that rewrite (instanced/indirect)
+                // degrade to an open strip, which only misses the closing segment.
+                MGLOG_W("GL_LINE_LOOP without index rewrite; drawing as LINE_STRIP");
+                return VK_PRIMITIVE_TOPOLOGY_LINE_STRIP;
+            case GL_LINES_ADJACENCY:
+                return VK_PRIMITIVE_TOPOLOGY_LINE_LIST_WITH_ADJACENCY;
+            case GL_LINE_STRIP_ADJACENCY:
+                return VK_PRIMITIVE_TOPOLOGY_LINE_STRIP_WITH_ADJACENCY;
+            case GL_TRIANGLES_ADJACENCY:
+                return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST_WITH_ADJACENCY;
+            case GL_TRIANGLE_STRIP_ADJACENCY:
+                return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP_WITH_ADJACENCY;
             default:
                 MGLOG_W("Unrecognized primitive topology");
                 return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST;

MobileGL/MG_Util/Converters/MGToVk/TextureEnumConverter.cpp

@@ -226,6 +226,8 @@ namespace MobileGL {
                 return VK_FORMAT_D24_UNORM_S8_UINT;
             case TextureInternalFormat::Depth32FStencil8:
                 return VK_FORMAT_D32_SFLOAT_S8_UINT;
+            case TextureInternalFormat::StencilIndex8:
+                return VK_FORMAT_S8_UINT;
             case TextureInternalFormat::DepthComponent32:
                 return VK_FORMAT_D32_SFLOAT;
             case TextureInternalFormat::DepthStencil:

MobileGL/MG_Util/Metrics/TextureMetrics.cpp

@@ -18,6 +18,7 @@ namespace MobileGL {
             case TextureInternalFormat::Red: // UNorm8 shadow layout
             case TextureInternalFormat::R8Snorm:
             case TextureInternalFormat::R8I:
+            case TextureInternalFormat::StencilIndex8:
             case TextureInternalFormat::R8UI:
                 return 1;
 
@@ -43,8 +44,11 @@ namespace MobileGL {
             case TextureInternalFormat::SRGB8:
             case TextureInternalFormat::RGB8I:
             case TextureInternalFormat::RGB8UI:
-            case TextureInternalFormat::DepthComponent24:
                 return 3;
+            // Canonical depth shadow is a full 32-bit unorm word (see PixelStoreProcessor),
+            // converted at upload to the image's own 24/32-bit layout.
+            case TextureInternalFormat::DepthComponent24:
+                return 4;
 
             case TextureInternalFormat::RGBA2:
             case TextureInternalFormat::RGBA4:
@@ -91,6 +95,9 @@ namespace MobileGL {
             case TextureInternalFormat::RG32F:
             case TextureInternalFormat::RG32I:
             case TextureInternalFormat::RG32UI:
+            // Shadow bytes hold the GL_FLOAT_32_UNSIGNED_INT_24_8_REV wire format
+            // (float depth + a word whose low 8 bits are stencil), 8 bytes/texel.
+            case TextureInternalFormat::Depth32FStencil8:
                 return 8;
 
             case TextureInternalFormat::RGB32F:
@@ -101,7 +108,6 @@ namespace MobileGL {
             case TextureInternalFormat::RGBA32F:
             case TextureInternalFormat::RGBA32I:
             case TextureInternalFormat::RGBA32UI:
-            case TextureInternalFormat::Depth32FStencil8:
                 return 16;
 
             case TextureInternalFormat::R11FG11FB10F:
@@ -253,8 +259,10 @@ namespace MobileGL {
             case TexturePixelDataType::UnsignedInt101111Rev:
             case TexturePixelDataType::UnsignedInt5999Rev:
             case TexturePixelDataType::UnsignedInt248:
-            case TexturePixelDataType::Float32UnsignedInt248Rev:
                 return 4;
+            case TexturePixelDataType::Float32UnsignedInt248Rev:
+                // A 32-bit float depth word followed by a 32-bit word holding stencil.
+                return 8;
             default:
                 return 0;
             }
@@ -501,9 +509,16 @@ namespace MobileGL {
                 s.Depth = 32;
                 s.Stencil = 8;
                 break;
+            case TextureInternalFormat::StencilIndex8:
+                s.Stencil = 8;
+                break;
+            case TextureInternalFormat::Unknown:
+                // Queried for attachments that have no storage yet (e.g. framebuffer
+                // parameter queries on the initial state); every size stays 0.
+                break;
             default:
-                MOBILEGL_ASSERT(false, "Unimplemented internal format in GetComponentSizesForInternalFormat: %d",
+                MGLOG_W("Unimplemented internal format in GetComponentSizesForInternalFormat: %d",
-                                static_cast<Int>(internal));
+                        static_cast<Int>(internal));
                 break;
             }
 

MobileGL/MG_Util/ShaderTranspiler/ShaderCompiler.cpp

@@ -220,11 +220,14 @@ namespace MobileGL {
                 auto result = ParseShaderSource(lang, shaderType, source, attrib.flags);
                 if (result) return result;
 
-                // Legacy desktop sources are normalized to "#version 330 core", which parses under
+                // Legacy desktop sources are normalized to "#version 330 core" (with a marker on the
-                // stricter rules than the 460 they used to be forced to: a shader declaring 330 while
+                // directive), which parses under stricter rules than the 460 they used to be forced
-                // using e.g. layout(binding=...) without the matching #extension line compiles on real
+                // to: a shader declaring 110-150 while using e.g. layout(binding=...) without the
-                // drivers but is rejected here. Retry once at 460 before reporting failure; a genuinely
+                // matching #extension line compiles on real drivers but is rejected here. Retry once
-                // broken shader fails both attempts and keeps its original diagnostics.
+                // at 460 before reporting failure; a genuinely broken shader fails both attempts and
+                // keeps its original diagnostics. Application-declared 330+ sources carry no marker
+                // and keep their declared version's strict rules (the GL CTS negative-compile cases
+                // depend on that).
                 String retrySource = source;
                 if (!MG_Util::ShaderTranspiler::RetargetLegacyVersionDirectiveTo460(retrySource)) {
                     return result;
@@ -360,6 +363,92 @@ namespace MobileGL {
                 return optimizer.Run(inputBinary.data(), inputBinary.size(), &outputBinary, options);
             }
 
+            bool ShaderCompiler::LowerRectImagesForEssl(const Vector<Uint32>& inputBinary,
+                                                        Vector<uint32_t>& outputBinary) {
+                constexpr SizeT kSpirvHeaderWordCount = 5;
+                // OpTypeImage: [0] opcode/wordcount, [1] result id, [2] sampled type, [3] Dim, ...
+                constexpr SizeT kTypeImageDimWordIndex = 3;
+                constexpr SizeT kTypeImageMinWordCount = 9;
+                outputBinary.clear();
+                if (inputBinary.size() < kSpirvHeaderWordCount || inputBinary[0] != spv::MagicNumber) {
+                    return false;
+                }
+
+                Vector<SizeT> rectDimWordOffsets;
+                Vector<SizeT> rectCapabilityWordOffsets;
+                Bool hasNormalizedCoordinateLookup = false;
+                for (SizeT offset = kSpirvHeaderWordCount; offset < inputBinary.size();) {
+                    const Uint32 instructionWord = inputBinary[offset];
+                    const SizeT wordCount = instructionWord >> 16u;
+                    const auto opcode = static_cast<spv::Op>(instructionWord & 0xffffu);
+                    if (wordCount == 0 || offset + wordCount > inputBinary.size()) {
+                        return false;
+                    }
+
+                    if (opcode == spv::Op::OpTypeImage && wordCount >= kTypeImageMinWordCount) {
+                        if (static_cast<spv::Dim>(inputBinary[offset + kTypeImageDimWordIndex]) == spv::Dim::Rect) {
+                            rectDimWordOffsets.push_back(offset + kTypeImageDimWordIndex);
+                        }
+                    } else if (opcode == spv::Op::OpCapability && wordCount >= 2) {
+                        const auto capability = static_cast<spv::Capability>(inputBinary[offset + 1]);
+                        if (capability == spv::Capability::SampledRect ||
+                            capability == spv::Capability::ImageRect) {
+                            rectCapabilityWordOffsets.push_back(offset + 1);
+                        }
+                    } else {
+                        switch (opcode) {
+                        // Normalized-coordinate lookups whose ESSL form the backend's
+                        // NormalizeRectSamplerCoordinates post-pass cannot repair: the
+                        // coordinate is either fused with something else in a single argument
+                        // (the Dref sample forms carry the compare value in coord.z) or the
+                        // divide would have to happen after a projective divide. Tracing each
+                        // one back to its image type would let a module mix a normalized 2D
+                        // lookup with a rectangle fetch, but the extra reach is not worth the
+                        // risk of getting the trace wrong: decline the whole module instead.
+                        //
+                        // OpImageSampleImplicitLod, OpImageGather and OpImageDrefGather are
+                        // absent because all three become an ESSL call whose argument 1 is the
+                        // bare texel-space coordinate, which the post-pass divides by the
+                        // texture size.
+                        case spv::Op::OpImageSampleExplicitLod:
+                        case spv::Op::OpImageSampleDrefImplicitLod:
+                        case spv::Op::OpImageSampleDrefExplicitLod:
+                        case spv::Op::OpImageSampleProjImplicitLod:
+                        case spv::Op::OpImageSampleProjExplicitLod:
+                        case spv::Op::OpImageSampleProjDrefImplicitLod:
+                        case spv::Op::OpImageSampleProjDrefExplicitLod:
+                        case spv::Op::OpImageSparseSampleImplicitLod:
+                        case spv::Op::OpImageSparseSampleExplicitLod:
+                        case spv::Op::OpImageSparseSampleDrefImplicitLod:
+                        case spv::Op::OpImageSparseSampleDrefExplicitLod:
+                        case spv::Op::OpImageSparseGather:
+                        case spv::Op::OpImageSparseDrefGather:
+                            hasNormalizedCoordinateLookup = true;
+                            break;
+                        default:
+                            break;
+                        }
+                    }
+                    offset += wordCount;
+                }
+
+                if (rectDimWordOffsets.empty() || hasNormalizedCoordinateLookup) {
+                    return false;
+                }
+
+                outputBinary.assign(inputBinary.begin(), inputBinary.end());
+                for (const SizeT dimWordOffset : rectDimWordOffsets) {
+                    outputBinary[dimWordOffset] = static_cast<Uint32>(spv::Dim::Dim2D);
+                }
+                // The rectangle capabilities describe types that no longer exist. Shader is always
+                // declared by a graphics module, so restating it keeps the word count intact
+                // without leaving a capability SPIRV-Cross would key off.
+                for (const SizeT capabilityWordOffset : rectCapabilityWordOffsets) {
+                    outputBinary[capabilityWordOffset] = static_cast<Uint32>(spv::Capability::Shader);
+                }
+                return true;
+            }
+
             bool ShaderCompiler::RebaseInstanceIndexForVulkan(const Vector<Uint32>& inputBinary,
                                                               Vector<uint32_t>& outputBinary) {
                 using namespace spvtools;

MobileGL/MG_Util/ShaderTranspiler/ShaderCompiler.h

@@ -43,6 +43,16 @@ namespace MobileGL {
                 // devices lacking GL_NV_shader_noperspective_interpolation. See EmulateNoPerspectivePass.
                 static bool EmulateNoPerspectiveForEssl(const Vector<Uint32>& inputBinary,
                                                         Vector<uint32_t>& outputBinary);
+                // Rewrites rectangle images (Dim::Rect) to plain 2D so SPIRV-Cross can emit ESSL
+                // for them at all - it refuses outright ("Rectangle textures are not supported on
+                // OpenGL ES"), which left the whole program unlinkable. Only valid while every use
+                // of the image takes integer texel coordinates (texelFetch / textureSize), where a
+                // rectangle target and a 2D target are indistinguishable; a normalized-coordinate
+                // lookup would also need its coordinates divided by the texture size, so the pass
+                // declines those modules instead of emitting something subtly wrong. Returns false
+                // when it changed nothing or cannot safely convert. DirectGLES only.
+                static bool LowerRectImagesForEssl(const Vector<Uint32>& inputBinary,
+                                                   Vector<uint32_t>& outputBinary);
                 // Rebases loads of the InstanceIndex builtin to (InstanceIndex - BaseInstance) so
                 // shaders see GL's zero-based gl_InstanceID. Vertex shaders only; DirectVulkan
                 // backend only (glslang's relaxed mode aliases gl_InstanceID to gl_InstanceIndex,

MobileGL/MG_Util/ShaderTranspiler/ShaderSourceProcessor.cpp

@@ -688,6 +688,10 @@ namespace {
         return info;
     }
 
+    // Stamped onto the normalized directive when a legacy (or absent) desktop
+    // version was rewritten to 330; consumed by RetargetLegacyVersionDirectiveTo460.
+    constexpr const char* kNormalizedLegacyMarker = "/*mobilegl-normalized-legacy*/";
+
     MobileGL::String GetNormalizedVersionDirective(const ShaderLanguageInfo& info) {
         if (info.profile == MobileGL::ShaderProfile::ES) {
             // Preserve the pre-existing behavior for standard lowercase "es" directives. MobileGL's Vulkan
@@ -702,9 +706,23 @@ namespace {
             return "#version 460 compatibility\n";
         }
 
+        // An explicitly declared modern core version keeps its number: the GL CTS
+        // negative-compile cases (reserved names, layout-qualifier forms, missing
+        // overloads) rely on the declared version's rules, and raising it would
+        // silently legalize them. gpu_shader5 opt-ins keep the 460 escalation -
+        // Vulkan glslang's ARB_gpu_shader5 support is not complete enough alone.
+        if (info.hasValidVersionDirective && info.version >= 330 && !info.enablesGpuShader5) {
+            return "#version " + std::to_string(info.version) + " core\n";
+        }
+
         const bool useLegacyDesktopVersion =
             info.version < 400 && !info.enablesGpuShader5;
-        return useLegacyDesktopVersion ? "#version 330 core\n" : "#version 460 core\n";
+        // The trailing marker records that this 330 came from a legacy declaration
+        // (or none at all), so the compile-failure retry may re-raise it to 460.
+        // An application's own "#version 330" never carries it and keeps strict
+        // 3.30 semantics.
+        return useLegacyDesktopVersion ? MobileGL::String("#version 330 core ") + kNormalizedLegacyMarker + "\n"
+                                       : "#version 460 core\n";
     }
 
     void NormalizeVersionDirective(MobileGL::String& source, const ShaderLanguageInfo& info) {
@@ -1306,12 +1324,123 @@ namespace MobileGL {
                 // Only the set NormalizeVersionDirective downgraded: desktop core below 400. ES and
                 // compatibility shaders keep whatever they declared.
                 if (info.profile != ShaderProfile::Core || info.version >= 400) return false;
+                // Only rescue MobileGL's own legacy normalization (marked on the directive line).
+                // An application-declared "#version 330" keeps strict 3.30 semantics: raising it
+                // would re-legalize the CTS negative-compile cases (reserved names, arrays of
+                // arrays, missing overloads).
+                SizeT lineEnd = source.find('\n', info.versionDirectiveStart);
+                if (lineEnd == MobileGL::String::npos) {
+                    lineEnd = source.size();
+                }
+                const SizeT markerPos = source.find(kNormalizedLegacyMarker, info.versionDirectiveStart);
+                if (markerPos == MobileGL::String::npos || markerPos > lineEnd) {
+                    return false;
+                }
 
                 source.replace(info.versionDirectiveStart, info.versionDirectiveEnd - info.versionDirectiveStart,
                                "#version 460 core\n");
                 return true;
             }
 
+            std::optional<String> FindReservedIdentifierViolation(const String& source) {
+                // Reserved anywhere; glslang accepts them as plain identifiers.
+                static constexpr const char* kAlwaysReserved[] = {
+                    "image1DShadow",
+                    "image2DShadow",
+                    "image1DArrayShadow",
+                    "image2DArrayShadow",
+                };
+                // Keywords legal only inside a layout(...) qualifier list.
+                static constexpr const char* kLayoutOnlyKeywords[] = {
+                    "packed",
+                    "row_major",
+                };
+
+                const auto isIdentChar = [](char c) {
+                    return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || (c >= '0' && c <= '9') || c == '_';
+                };
+
+                const SizeT length = source.size();
+                SizeT i = 0;
+                Int layoutParenDepth = 0;   // >0 while inside layout(...)
+                Bool pendingLayoutParen = false; // saw "layout", awaiting its '('
+                while (i < length) {
+                    const char c = source[i];
+                    // Comments.
+                    if (c == '/' && i + 1 < length && source[i + 1] == '/') {
+                        while (i < length && source[i] != '\n') ++i;
+                        continue;
+                    }
+                    if (c == '/' && i + 1 < length && source[i + 1] == '*') {
+                        i += 2;
+                        while (i + 1 < length && !(source[i] == '*' && source[i + 1] == '/')) ++i;
+                        i = (i + 1 < length) ? i + 2 : length;
+                        continue;
+                    }
+                    // Preprocessor lines stay out of scope (macro names may shadow anything).
+                    if (c == '#' && (i == 0 || source[i - 1] == '\n' ||
+                                     source.find_last_not_of(" \t", i - 1) == MobileGL::String::npos ||
+                                     source[source.find_last_not_of(" \t", i - 1)] == '\n')) {
+                        while (i < length && source[i] != '\n') {
+                            if (source[i] == '\\' && i + 1 < length && source[i + 1] == '\n') ++i;
+                            ++i;
+                        }
+                        continue;
+                    }
+                    if (c == '(') {
+                        if (pendingLayoutParen) {
+                            layoutParenDepth = 1;
+                            pendingLayoutParen = false;
+                        } else if (layoutParenDepth > 0) {
+                            ++layoutParenDepth;
+                        }
+                        ++i;
+                        continue;
+                    }
+                    if (c == ')') {
+                        if (layoutParenDepth > 0) --layoutParenDepth;
+                        ++i;
+                        continue;
+                    }
+                    if (c == ' ' || c == '\t' || c == '\r' || c == '\n') {
+                        ++i;
+                        continue;
+                    }
+                    if (isIdentChar(c) && !(c >= '0' && c <= '9')) {
+                        const SizeT start = i;
+                        while (i < length && isIdentChar(source[i])) ++i;
+                        const StringView word(source.data() + start, i - start);
+                        if (word == "layout") {
+                            pendingLayoutParen = true;
+                            continue;
+                        }
+                        pendingLayoutParen = false;
+                        for (const char* reserved : kAlwaysReserved) {
+                            if (word == reserved) {
+                                return String("ERROR: reserved identifier '") + reserved + "' may not be used.";
+                            }
+                        }
+                        if (layoutParenDepth == 0) {
+                            for (const char* keyword : kLayoutOnlyKeywords) {
+                                if (word == keyword) {
+                                    return String("ERROR: '") + keyword +
+                                        "' is a keyword and may not be used as an identifier.";
+                                }
+                            }
+                        }
+                        continue;
+                    }
+                    if (isIdentChar(c)) { // digit-led token: skip the whole number/identifier tail
+                        while (i < length && isIdentChar(source[i])) ++i;
+                        pendingLayoutParen = false;
+                        continue;
+                    }
+                    pendingLayoutParen = false;
+                    ++i;
+                }
+                return std::nullopt;
+            }
+
         } // namespace ShaderTranspiler
     } // namespace MG_Util
 } // namespace MobileGL

MobileGL/MG_Util/ShaderTranspiler/ShaderSourceProcessor.h

@@ -40,6 +40,11 @@ namespace MobileGL {
             // uses 420-era syntax without the matching #extension line, which real drivers tend to
             // accept - can be retried instead of failing to compile.
             Bool RetargetLegacyVersionDirectiveTo460(String& source);
+
+            // GLSL reserves a few names glslang happily accepts as identifiers ("packed",
+            // "row_major" outside a layout(...) list, the image*Shadow family). Returns the
+            // compile-error text for the first violation, or nullopt for a clean source.
+            std::optional<String> FindReservedIdentifierViolation(const String& source);
         } // namespace ShaderTranspiler
     } // namespace MG_Util
 } // namespace MobileGL

MobileGL/MG_Util/Texture/PixelStoreProcessor.cpp

@@ -95,6 +95,7 @@ namespace MobileGL::MG_Util::PixelStoreProcessor {
             Int32,
             Half,
             Float32,
+            UNorm32, // 32-bit fixed-point depth shadow
         };
 
         struct InternalShadowLayout {
@@ -123,6 +124,21 @@ namespace MobileGL::MG_Util::PixelStoreProcessor {
 
         Bool GetInternalShadowLayout(TextureInternalFormat internal, InternalShadowLayout& out) {
             switch (internal) {
+            // Depth shadows follow TextureFormatProcessor::NormalizePixelFormat: 16-bit
+            // unorm for DEPTH_COMPONENT16, 32-bit unorm for the 24/32-bit fixed-point
+            // depths, float for DEPTH_COMPONENT32F.
+            case TextureInternalFormat::DepthComponent16:
+                out = {1, ShadowComponent::UNorm16, false};
+                return true;
+            case TextureInternalFormat::DepthComponent24:
+            case TextureInternalFormat::DepthComponent32:
+            case TextureInternalFormat::DepthComponent:
+                out = {1, ShadowComponent::UNorm32, false};
+                return true;
+            case TextureInternalFormat::DepthComponent32F:
+                out = {1, ShadowComponent::Float32, false};
+                return true;
+
             case TextureInternalFormat::R8:
             case TextureInternalFormat::Red:     out = {1, ShadowComponent::UNorm8, false}; return true;
             case TextureInternalFormat::RG8:
@@ -299,8 +315,10 @@ namespace MobileGL::MG_Util::PixelStoreProcessor {
             case TextureInputFormat::RGBAInteger:  out = {{0, 1, 2, 3}, 4, true};     return true;
             case TextureInputFormat::BGRA:         out = {{2, 1, 0, 3}, 4, false};    return true;
             case TextureInputFormat::BGRAInteger:  out = {{2, 1, 0, 3}, 4, true};     return true;
+            // A depth value converts like a single normalized/float channel.
+            case TextureInputFormat::DepthComponent: out = {{0, -1, -1, -1}, 1, false}; return true;
             default:
-                return false; // depth / stencil / unknown
+                return false; // stencil / packed depth-stencil / unknown
             }
         }
 
@@ -346,8 +364,7 @@ namespace MobileGL::MG_Util::PixelStoreProcessor {
                 out = isInteger ? ShadowComponent::Int16 : ShadowComponent::SNorm16;
                 return true;
             case TexturePixelDataType::UnsignedInt:
-                if (!isInteger) return false; // no 32-bit normalized shadow layout
+                out = isInteger ? ShadowComponent::UInt32 : ShadowComponent::UNorm32;
-                out = ShadowComponent::UInt32;
                 return true;
             case TexturePixelDataType::Int:
                 if (!isInteger) return false;
@@ -617,6 +634,12 @@ namespace MobileGL::MG_Util::PixelStoreProcessor {
             case ShadowComponent::Float32:
                 Memcpy(dst, &v, sizeof(v));
                 break;
+            case ShadowComponent::UNorm32: {
+                const auto out = static_cast<Uint32>(
+                    std::llround(static_cast<double>(std::clamp(v, 0.0f, 1.0f)) * 4294967295.0));
+                Memcpy(dst, &out, sizeof(out));
+                break;
+            }
             default:
                 break; // integer components never reach the float encoder
             }
@@ -771,6 +794,64 @@ namespace MobileGL::MG_Util::PixelStoreProcessor {
         const Int effectiveHeight = (params.ImageHeight > 0) ? params.ImageHeight : height;
         const SizeT inputRowStride = CalculateRowStride(effectiveWidth, pixelSize, params.Alignment);
 
+        // GL_DEPTH_COMPONENT client data may populate a packed depth-stencil internal
+        // format (the stencil half becomes zero); the generic channel converter cannot
+        // express the packed shadow words, so convert here.
+        const Bool packedDepthStencilInternal = targetInternalFormat == TextureInternalFormat::Depth24Stencil8 ||
+            targetInternalFormat == TextureInternalFormat::DepthStencil ||
+            targetInternalFormat == TextureInternalFormat::Depth32FStencil8;
+        if (!isBitmap && packedDepthStencilInternal && textureInputFormat == TextureInputFormat::DepthComponent &&
+            (inputDataType == TexturePixelDataType::Float || inputDataType == TexturePixelDataType::UnsignedInt ||
+             inputDataType == TexturePixelDataType::UnsignedShort)) {
+            const Bool floatShadow = targetInternalFormat == TextureInternalFormat::Depth32FStencil8;
+            const SizeT outPixelSize = floatShadow ? 8 : 4;
+            outSize = static_cast<SizeT>(width) * height * std::max(depth, 1) * outPixelSize;
+            Uint8* outputPixels = static_cast<Uint8*>(malloc(outSize));
+            if (!outputPixels) {
+                outSize = 0;
+                return nullptr;
+            }
+            const Uint8* srcBase = static_cast<const Uint8*>(inputPixels) +
+                static_cast<SizeT>(params.SkipImages) * static_cast<SizeT>(effectiveHeight) * inputRowStride +
+                static_cast<SizeT>(params.SkipRows) * inputRowStride +
+                static_cast<SizeT>(params.SkipPixels) * pixelSize;
+            Uint8* dst = outputPixels;
+            for (Int z = 0; z < std::max(depth, 1); ++z) {
+                for (Int y = 0; y < height; ++y) {
+                    const Uint8* srcRow = srcBase +
+                        static_cast<SizeT>(z) * static_cast<SizeT>(effectiveHeight) * inputRowStride +
+                        static_cast<SizeT>(y) * inputRowStride;
+                    for (Int x = 0; x < width; ++x) {
+                        Float depthValue = 0.0f;
+                        if (inputDataType == TexturePixelDataType::Float) {
+                            Memcpy(&depthValue, srcRow + static_cast<SizeT>(x) * 4, sizeof(depthValue));
+                        } else if (inputDataType == TexturePixelDataType::UnsignedInt) {
+                            Uint32 raw = 0;
+                            Memcpy(&raw, srcRow + static_cast<SizeT>(x) * 4, sizeof(raw));
+                            depthValue = static_cast<Float>(static_cast<double>(raw) / 4294967295.0);
+                        } else {
+                            Uint16 raw = 0;
+                            Memcpy(&raw, srcRow + static_cast<SizeT>(x) * 2, sizeof(raw));
+                            depthValue = static_cast<Float>(raw) / 65535.0f;
+                        }
+                        if (floatShadow) {
+                            const Uint32 stencilWord = 0;
+                            Memcpy(dst, &depthValue, sizeof(depthValue));
+                            Memcpy(dst + 4, &stencilWord, sizeof(stencilWord));
+                            dst += 8;
+                        } else {
+                            const Uint32 depth24 = static_cast<Uint32>(
+                                std::llround(static_cast<double>(std::clamp(depthValue, 0.0f, 1.0f)) * 16777215.0));
+                            const Uint32 word = depth24 << 8;
+                            Memcpy(dst, &word, sizeof(word));
+                            dst += 4;
+                        }
+                    }
+                }
+            }
+            return outputPixels;
+        }
+
         UnpackConversionSpec conversion{};
         const Bool needConversion =
             !isBitmap && GetUnpackConversionSpec(targetInternalFormat, textureInputFormat, inputDataType, conversion);
@@ -972,6 +1053,11 @@ namespace MobileGL::MG_Util::PixelStoreProcessor {
                 Memcpy(&v, p, sizeof(v));
                 return v;
             }
+            case ShadowComponent::UNorm32: {
+                Uint32 v;
+                Memcpy(&v, p, sizeof(v));
+                return static_cast<Float>(static_cast<double>(v) / 4294967295.0);
+            }
             default:
                 return 0.0f;
             }

MobileGL/MG_Util/Texture/TextureFormatProcessor.cpp

@@ -29,11 +29,16 @@ namespace MobileGL::MG_Util::TextureFormatProcessor {
         case GL_RGB12: // stored as RGB16 (see NormalizePixelFormat)
             applicableOptions |= options & PixelFormatNormalizeOptionBit::NoNorm16;
             applicableOptions |= options & PixelFormatNormalizeOptionBit::NoRgb16;
+            applicableOptions |= options & PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget;
             break;
         case GL_RGB16_SNORM:
             applicableOptions |= options & PixelFormatNormalizeOptionBit::NoRGB16Snorm;
             applicableOptions |= options & PixelFormatNormalizeOptionBit::NoNorm16;
             applicableOptions |= options & PixelFormatNormalizeOptionBit::NoSnorm16;
+            applicableOptions |= options & PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget;
+            if (options & PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget) {
+                applicableOptions |= options & PixelFormatNormalizeOptionBit::NoSnorm16RenderTarget;
+            }
             break;
         case GL_RGBA16_SNORM:
         case GL_RG16_SNORM:
@@ -46,6 +51,9 @@ namespace MobileGL::MG_Util::TextureFormatProcessor {
             applicableOptions |= options & PixelFormatNormalizeOptionBit::NoRGBA8Snorm;
             break;
         case GL_RGB8_SNORM:
+            applicableOptions |= options & PixelFormatNormalizeOptionBit::NoSnorm8;
+            applicableOptions |= options & PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget;
+            break;
         case GL_RG8_SNORM:
         case GL_R8_SNORM:
             applicableOptions |= options & PixelFormatNormalizeOptionBit::NoSnorm8;
@@ -66,7 +74,15 @@ namespace MobileGL::MG_Util::TextureFormatProcessor {
             switch (internalFormat) {
             case GL_DEPTH_COMPONENT32:
                 if (options & PixelFormatNormalizeOptionBit::NoDepthComponent32) {
-                    *outInternalFormat = GL_DEPTH_COMPONENT;
+                    // The unsized GL_DEPTH_COMPONENT base format is not a legal
+                    // glTexStorage/glRenderbufferStorage internal format on ES, which left
+                    // the attachment with no storage at all (KHR-GL3x.framebuffer_blit's
+                    // GL_DEPTH_COMPONENT32 config then read an incomplete framebuffer).
+                    // GL_DEPTH_COMPONENT24 is the nearest sized ES format that keeps the
+                    // same fixed-point encoding, so the GL_UNSIGNED_INT transfer type below
+                    // still describes the data; GL_DEPTH_COMPONENT32F would need a float
+                    // conversion the upload path does not apply.
+                    *outInternalFormat = GL_DEPTH_COMPONENT24;
                     break;
                 }
                 *outInternalFormat = internalFormat;
@@ -79,6 +95,13 @@ namespace MobileGL::MG_Util::TextureFormatProcessor {
                 *outInternalFormat = internalFormat;
                 break;
             case GL_RGB16:
+                if (options & PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget) {
+                    // GL_RGB32F is a legal ES texture format but is not colour-renderable, so
+                    // glTexStorage2DMultisample rejects it and the attachment ends up with no
+                    // storage at all.
+                    *outInternalFormat = GL_RGBA32F;
+                    break;
+                }
                 if ((options & PixelFormatNormalizeOptionBit::NoNorm16) ||
                     (options & PixelFormatNormalizeOptionBit::NoRgb16)) {
                     *outInternalFormat = GL_RGB32F;
@@ -109,6 +132,14 @@ namespace MobileGL::MG_Util::TextureFormatProcessor {
                 *outInternalFormat = internalFormat;
                 break;
             case GL_RGB16_SNORM:
+                if (options & PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget) {
+                    // A half float loses the low bits of a 16-bit SNORM channel, so keep the
+                    // signed-normalized encoding whenever the driver can render to it.
+                    *outInternalFormat = (options & PixelFormatNormalizeOptionBit::NoSnorm16RenderTarget)
+                                             ? GL_RGBA16F
+                                             : GL_RGBA16_SNORM;
+                    break;
+                }
                 if ((options & PixelFormatNormalizeOptionBit::NoNorm16) ||
                     (options & PixelFormatNormalizeOptionBit::NoRGB16Snorm) ||
                     (options & PixelFormatNormalizeOptionBit::NoSnorm16)) {
@@ -142,6 +173,10 @@ namespace MobileGL::MG_Util::TextureFormatProcessor {
                 *outInternalFormat = internalFormat;
                 break;
             case GL_RGB8_SNORM:
+                if (options & PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget) {
+                    *outInternalFormat = GL_RGBA16F;
+                    break;
+                }
                 if (options & PixelFormatNormalizeOptionBit::NoSnorm8) {
                     *outInternalFormat = GL_RGB16F;
                     break;
@@ -552,6 +587,8 @@ namespace MobileGL::MG_Util::TextureFormatProcessor {
                 *outType = GL_UNSIGNED_INT;
                 break;
             case GL_DEPTH_COMPONENT32:
+                // Follows the internal-format normalization above: ES only accepts
+                // GL_FLOAT data for a GL_DEPTH_COMPONENT32F store.
                 *outType = GL_UNSIGNED_INT;
                 break;
             case GL_DEPTH_COMPONENT32F:

MobileGL/MG_Util/Texture/TextureFormatProcessor.h

@@ -18,6 +18,17 @@ namespace MobileGL {
         NoDepthComponent32 = 1 << 4,
         NoRGBA8Snorm = 1 << 5,
         NoRGB16Snorm = 1 << 6,
+        // The target must be colour-renderable and ES has no renderable three-channel
+        // form of the requested format, so it has to be widened to the four-channel one.
+        // Only meaningful for multisample textures: those can never be uploaded to, only
+        // rendered into, so the extra alpha comes from the draw (1.0 for an RGB source)
+        // and no transfer path has to expand three-channel client data.
+        NoThreeChannelRenderTarget = 1 << 7,
+        // Pairs with the bit above: the widened four-channel format has to stay renderable AND
+        // keep 16-bit signed-normalized precision, which needs both EXT_texture_norm16 and
+        // EXT_render_snorm. Without them the only renderable widening left is a half float, whose
+        // 11-bit mantissa cannot represent a 16-bit SNORM channel exactly.
+        NoSnorm16RenderTarget = 1 << 8,
         None = 0,
     };
     namespace MG_Util::TextureFormatProcessor {

MobileGL/MG_Util/Types.h

@@ -181,7 +181,19 @@ namespace MobileGL {
         explicit BindingSlotRange1D(TargetEnum target, const Range1D& range = Range1D())
             : BindingSlot<ObjectType>(target), m_range(range) {}
 
-        Range1D GetRange() const { return m_range; }
+        // The range the binding actually covers right now. A whole-buffer binding
+        // (glBindBufferBase) does not freeze anything: GL resolves it against the
+        // object's size at every use, so a glBufferData issued after the bind has to
+        // be visible here - binding an empty buffer and giving it storage afterwards
+        // is ordinary application code. Only glBindBufferRange pins a fixed window.
+        Range1D GetRange() const {
+            if (!m_hasExplicitRange) {
+                if (const auto& object = this->GetBoundObject()) {
+                    return Range1D(0, object->GetSize());
+                }
+            }
+            return m_range;
+        }
 
         Bool HasExplicitRange() const { return m_hasExplicitRange; }
 

android-plugin/app/build.gradle.kts

@@ -55,7 +55,7 @@ val releaseSigningReady = signingStoreFile.exists()
 val debuggableRelease = (findProperty("mobilegl.debuggableRelease") ?: "false").toString().toBoolean()
 
 val mobileGlVersionMajor = 26
-val mobileGlVersionMinor = 7
+val mobileGlVersionMinor = 8
 val mobileGlGitShortHash = runGit("rev-parse", "--short=7", "HEAD") ?: "nogit"
 val mobileGlMonthlyRevision = runGit(
     "rev-list",

docs/superpowers/specs/2026-07-15-mobilegl-post-format-capability-tables-design.md

@@ -1,78 +0,0 @@
-# MobileGL POST Format Capability Tables
-
-## Goal
-
-Expose the format-capability results used during MobileGL backend startup in the Android plugin's driver POST screen. The screen must show the exact `Full`, `Caveat`, or `None` result for every backend, target, internal format, and capability without duplicating the backend's detection rules.
-
-## Existing Architecture
-
-- `DriverPost.cpp` probes the device GLES and Vulkan drivers before `MobileGL::Initialize()` and returns a `BackendPostReport` for each backend.
-- `DriverPostJni.cpp` serializes those reports to JSON for `PostActivity`.
-- `PostActivity` uses platform Android views and already supports collapsible check details and a collapsible raw report.
-- Backend startup fills a `FormatCapabilityCache` in the DirectGLES and DirectVulkan `InitCapabilities()` paths. `FullCaps` takes precedence over `CaveatCaps`; an absent bit means `None`.
-- The capability matrix contains 12 targets, 75 internal formats, and 14 capability columns per backend.
-
-## Selected Approach
-
-Extract callable format-probe entry points from the existing DirectGLES and DirectVulkan implementations. Backend startup and the POST will call these same functions, so their results cannot drift.
-
-The POST will run each probe while its temporary driver resources are still valid:
-
-- DirectGLES: after the GLES function table and capabilities have been populated, while the 1x1 pbuffer context is current.
-- DirectVulkan: after selecting the physical device, while the Vulkan instance and physical device handles remain valid.
-
-The resulting optional `FormatCapabilityCache` will be stored in each `BackendPostReport`. Failure to obtain a format table will not discard the existing POST checks or change their verdict; the UI will instead report that the format table is unavailable.
-
-## JSON Contract
-
-The JNI report will add an optional `formatCapabilities` object to each backend. To avoid repeating tens of thousands of status strings, the object will contain:
-
-- one ordered capability-name array;
-- one entry per target;
-- one compact row per internal format containing the format name, a Full bitmask, and a Caveat bitmask.
-
-Java resolves each cell in this order:
-
-1. Full bit present: `Full`.
-2. Otherwise Caveat bit present: `Caveat`.
-3. Otherwise: `None`.
-
-This preserves the backend's current precedence and keeps the raw JSON reasonably small.
-
-## Android UI
-
-Each backend section keeps its existing verdict, renderer string, and check table. A new `Format capabilities` subsection follows it.
-
-- Each of the 11 texture targets and `Renderbuffer` is a separate, initially collapsed table.
-- Target headers can be expanded independently.
-- Table content is created on expansion and removed when collapsed, preventing the activity from retaining roughly 27,000 status views.
-- Each expanded table is placed in a horizontal scroll container.
-- The first column contains internal-format names. The remaining columns use the ordered capability names from the JSON report.
-- Every status cell displays its status text and uses the conventional color mapping:
-  - `Full`: green background with white text.
-  - `Caveat`: yellow background with black text.
-  - `None`: red background with white text.
-- Header and format-name cells use neutral dark backgrounds consistent with the existing POST theme.
-- The existing raw-report toggle remains available at the end of the screen.
-
-## Performance and Lifecycle
-
-- The existing single-flight native POST and cached JSON behavior remains unchanged.
-- Format tables are lazily materialized and discarded on collapse.
-- The JSON carries bitmasks rather than repeated `Full`, `Caveat`, and `None` strings.
-- Existing configuration-change handling remains unchanged.
-
-## Validation
-
-1. Run focused source checks and `git diff --check`.
-2. Build the Android plugin APK with the repository's current Gradle workflow.
-3. If an Android target is connected, install the APK and open `PostActivity`.
-4. Verify both backend sections, all target toggles, horizontal scrolling, visible cell text, and the green/yellow/red mapping.
-5. Confirm collapsing a table removes its generated content and expanding it recreates the same values.
-
-## Non-Goals
-
-- Changing the meanings of `Full`, `Caveat`, or `None`.
-- Changing POST verdict rules.
-- Displaying sample-count vectors in this iteration.
-- Replacing the existing platform-view UI with Compose, AppCompat, or WebView.

tools/cts/README.md

@@ -1,4 +1,31 @@
-# Running the OpenGL CTS (VK-GL-CTS / KHR-GL33) against MobileGL on Android
+# Running the OpenGL CTS against MobileGL
+
+This directory contains two supported paths:
+
+- Android arm64 / MobileGL EGL: the KHR-GL33 workflow documented below and in
+  `skills/gl-cts-on-mobilegl/SKILL.md`.
+- Windows x64 / MobileGL WGL: the GL30-GL46 pipeline in
+  `scripts/wgl_glcts_pipeline.py`, documented by
+  `skills/wgl-gl-cts-on-mobilegl/SKILL.md`.
+
+Windows prerequisites are Git, Python 3.9+, CMake, Visual Studio 2022's Desktop
+C++ workload, and a Vulkan SDK visible to CMake. DirectVulkan also needs a
+working Vulkan loader plus a GPU-vendor ICD and driver; the SDK alone is not a
+GPU driver.
+
+For Windows, start with:
+
+```powershell
+python tools\cts\scripts\wgl_glcts_pipeline.py --help
+```
+
+The pipeline builds MobileGL as a drop-in `opengl32.dll`, builds or reuses
+`glcts.exe`, checks that WGL loaded MobileGL rather than the system driver,
+resumes individual suites after crashes/timeouts, and writes Markdown plus JSON
+reports below the printed `runs/<first-16-of-run-fingerprint>` directory. Its
+manifest records provenance and the runner settings used to validate a resume.
+
+## Android KHR-GL33 workflow
 
 Goal: measure how much of the OpenGL 3.3 core-profile conformance suite MobileGL
 passes, separately for each backend (`DirectGLES`, `DirectVulkan`).

tools/cts/platform/tcuMobileGLPlatform.cpp

@@ -1,5 +1,5 @@
 /*-------------------------------------------------------------------------
- * dEQP platform port for MobileGL on Android
+ * dEQP platform port for MobileGL (Android and desktop Linux)
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -28,14 +28,17 @@
  *  - A real surface is always created. MobileGL rejects EGL_NO_SURFACE with
  *    EGL_BAD_MATCH, and --deqp-surface-type=fbo asks the platform for
  *    SURFACETYPE_DONT_CARE, so "no surface" is not an option.
- *  - Window surfaces are backed by an AImageReader rather than an Activity,
+ *  - On Android, window surfaces are backed by an AImageReader rather than an
- *    which is what lets the suite run as a plain adb-shell binary. DirectVulkan
+ *    Activity, which is what lets the suite run as a plain adb-shell binary.
- *    needs this: its pbuffer path requires VK_EXT_headless_surface, which
+ *    DirectVulkan needs this: its pbuffer path requires VK_EXT_headless_surface,
- *    Adreno's Android driver does not expose.
+ *    which Adreno's Android driver does not expose.
+ *  - On desktop Linux, only pbuffer surfaces are offered. DirectVulkan's
+ *    pbuffer path works there because desktop Vulkan loaders expose
+ *    VK_EXT_headless_surface.
  *
  * Environment:
  *   MOBILEGL_CTS_LIB      path/soname of the MobileGL library (default libMobileGL.so)
- *   MOBILEGL_CTS_SURFACE  "window" (default) or "pbuffer"
+ *   MOBILEGL_CTS_SURFACE  "window" (Android default) or "pbuffer" (desktop default/only)
  *   MOBILEGL_BACKEND_TYPE read by MobileGL itself; set it before launching
  *//*--------------------------------------------------------------------*/
 
@@ -58,9 +61,11 @@
 #include "tcuPlatform.hpp"
 #include "tcuRenderTarget.hpp"
 
+#if defined(__ANDROID__)
 #include <android/hardware_buffer.h>
 #include <android/native_window.h>
 #include <media/NdkImageReader.h>
+#endif
 
 using std::string;
 using std::vector;
@@ -88,13 +93,24 @@ static string getLibraryName(void)
     return (env && env[0]) ? string(env) : string("libMobileGL.so");
 }
 
-//! Window surfaces default on: they are the only kind DirectVulkan can use.
+#if defined(__ANDROID__)
+//! Window surfaces default on: they are the only kind DirectVulkan can use
+//! on Android (its pbuffer path needs VK_EXT_headless_surface).
 static bool useWindowSurface(void)
 {
     const char *env = std::getenv("MOBILEGL_CTS_SURFACE");
     return !(env && string(env) == "pbuffer");
 }
+#else
+//! Desktop: pbuffer only. VK_EXT_headless_surface is available there and no
+//! Activity-free native window abstraction exists.
+static bool useWindowSurface(void)
+{
+    return false;
+}
+#endif
 
+#if defined(__ANDROID__)
 /*--------------------------------------------------------------------*//*!
  * \brief A real ANativeWindow with no Activity behind it.
  *
@@ -160,6 +176,12 @@ private:
     AImageReader *m_reader;
     ANativeWindow *m_window;
 };
+#else
+//! Never instantiated on desktop; keeps EglRenderContext's member deletable.
+class ImageReaderWindow
+{
+};
+#endif
 
 class GetProcFuncLoader : public glw::FunctionLoader
 {
@@ -352,6 +374,7 @@ EglRenderContext::EglRenderContext(const glu::RenderConfig &config, const tcu::C
 
     if (wantWindow)
     {
+#if defined(__ANDROID__)
         m_window = new ImageReaderWindow(width, height);
 
         eglw::EGLint visualId = 0;
@@ -361,6 +384,9 @@ EglRenderContext::EglRenderContext(const glu::RenderConfig &config, const tcu::C
         m_eglSurface = m_egl.createWindowSurface(m_eglDisplay, eglConfig,
                                                  (eglw::EGLNativeWindowType)m_window->getWindow(), nullptr);
         EGLU_CHECK_MSG(m_egl, "eglCreateWindowSurface()");
+#else
+        throw tcu::NotSupportedError("Window surfaces are not supported by the desktop MobileGL platform");
+#endif
     }
     else
     {

tools/cts/scripts/cts_matrix_report.py

@@ -0,0 +1,545 @@
+#!/usr/bin/env python
+"""Build a GL 3.0--3.3 CTS conformance matrix from dEQP QPA logs.
+
+The report deliberately scores against the unique cases in each supplied
+caselist.  A case that has not produced a result therefore cannot disappear
+from the denominator and make a partial run look conformant.
+
+QPA parsing and crash/hang sidecar handling follow :mod:`qpa_report`:
+
+* a later QPA observation of a case wins;
+* ``crashed.txt`` upgrades a missing/incomplete result to ``Crash``;
+* ``hung.txt`` upgrades a missing/incomplete/crash result to ``DeviceHang``.
+
+Example::
+
+    python cts_matrix_report.py \
+      --gl30-caselist gl30-main.txt --gl30-results runs/gl30 \
+      --gl31-caselist gl31-main.txt --gl31-results runs/gl31 \
+      --gl32-caselist gl32-main.txt --gl32-results runs/gl32 \
+      --gl33-caselist gl33-main.txt --gl33-results runs/gl33 \
+      --json runs/cts-matrix.json
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import os
+import re
+import sys
+from collections import Counter, defaultdict
+from datetime import datetime, timezone
+from typing import Iterable, Optional, Sequence
+
+try:  # Works both as a directly executed script and as a package import.
+    from . import qpa_report
+except ImportError:  # pragma: no cover - exercised by the command-line tests
+    import qpa_report
+
+
+VERSIONS = ("gl30", "gl31", "gl32", "gl33")
+ACCEPTED_STATUSES = (
+    "Pass",
+    "NotSupported",
+    "QualityWarning",
+    "CompatibilityWarning",
+    "Waiver",
+)
+ACCEPTED = frozenset(ACCEPTED_STATUSES)
+CHUNK_QPA = re.compile(r"^chunk(\d+)\.qpa$", re.IGNORECASE)
+
+
+class ReportInputError(ValueError):
+    """An input path cannot be used to construct a meaningful report."""
+
+
+def _read_non_comment_lines(path: str) -> list[str]:
+    try:
+        # Match run_cts_windows.py: Khronos lists are UTF-8 and may carry a BOM.
+        with open(path, "r", encoding="utf-8-sig", errors="strict") as fh:
+            return [
+                line.strip()
+                for line in fh
+                if line.strip() and not line.lstrip().startswith("#")
+            ]
+    except (OSError, UnicodeError) as exc:
+        raise ReportInputError(f"cannot read {path}: {exc}") from exc
+
+
+def read_caselist(path: str) -> tuple[list[str], dict[str, int]]:
+    """Return unique cases in file order and repeated caselist entries.
+
+    The mustpass files consumed by glcts and ``run_cts.py`` are one case per
+    non-empty, non-comment line, so this intentionally uses the same syntax.
+    """
+
+    entries = _read_non_comment_lines(path)
+    counts = Counter(entries)
+    unique = list(dict.fromkeys(entries))
+    duplicates = {case: count for case, count in counts.items() if count > 1}
+    return unique, duplicates
+
+
+def _collect_qpa_files(paths: Sequence[str]) -> list[str]:
+    missing = [path for path in paths if not os.path.exists(path)]
+    if missing:
+        raise ReportInputError(
+            "result path(s) do not exist: " + ", ".join(sorted(missing))
+        )
+
+    # qpa_report.collect provides the established directory-recursion rules.
+    # De-duplicate aliases so specifying the same directory twice does not
+    # manufacture duplicate observations.
+    files = qpa_report.collect(paths)
+    by_identity: dict[str, str] = {}
+    for path in files:
+        if not os.path.isfile(path):
+            raise ReportInputError(f"QPA input is not a file: {path}")
+        absolute = os.path.abspath(path)
+        by_identity.setdefault(os.path.normcase(absolute), absolute)
+    def order_key(value: str) -> tuple[str, str, int, str]:
+        absolute = os.path.abspath(value)
+        directory = os.path.normcase(os.path.dirname(absolute))
+        filename = os.path.normcase(os.path.basename(absolute))
+        match = CHUNK_QPA.fullmatch(filename)
+        if match:
+            # run_cts_windows.py uses a minimum width of four digits, not a
+            # fixed width.  Numeric ordering is therefore required once a run
+            # reaches chunk10000; lexical ordering would put it before
+            # chunk9999 and break the later-observation-wins rule.
+            return directory, "chunk", int(match.group(1)), filename
+        # Preserve a deterministic, name-based position for foreign/legacy
+        # QPA files while grouping numeric runner chunks at the lexical
+        # position occupied by the "chunk" basename.
+        return directory, filename, -1, filename
+
+    return sorted(by_identity.values(), key=order_key)
+
+
+def _ratio(numerator: int, denominator: int) -> float:
+    return numerator / denominator if denominator else 0.0
+
+
+def _sidecar(paths: Sequence[str], name: str) -> set[str]:
+    """Load a run_cts.py sidecar with qpa_report-compatible lookup rules."""
+
+    return qpa_report.load_sidecar(paths, name)
+
+
+def build_version_report(
+    version: str, caselist: str, result_paths: Sequence[str]
+) -> dict:
+    """Build the serialisable report for one GL mustpass version."""
+
+    expected_cases, expected_duplicates = read_caselist(caselist)
+    expected = set(expected_cases)
+    qpa_files = _collect_qpa_files(result_paths)
+
+    results: dict[str, str] = {}
+    observation_history: dict[str, list[dict[str, str]]] = defaultdict(list)
+    for qpa_file in qpa_files:
+        for case, status in qpa_report.parse_qpa(qpa_file):
+            observation_history[case].append(
+                {"file": qpa_file, "status": status}
+            )
+            results[case] = status
+
+    crashed = _sidecar(result_paths, "crashed.txt")
+    hung = _sidecar(result_paths, "hung.txt")
+    explicit_unrun = _sidecar(result_paths, "unrun.txt")
+    skipped = _sidecar(result_paths, "skipped.txt")
+
+    # Keep this order and these guards in lock-step with qpa_report.py.
+    for case in crashed:
+        if results.get(case, "Incomplete") == "Incomplete":
+            results[case] = "Crash"
+    for case in hung:
+        if results.get(case, "Incomplete") in ("Incomplete", "Crash"):
+            results[case] = "DeviceHang"
+
+    # A begin/end pair without <Result>, or a QPA truncated mid-case, is not a
+    # completed observation.  Sidecars above may upgrade it to Crash/Hang;
+    # anything still Incomplete must stay in the expected denominator as unrun.
+    incomplete_results = {
+        case for case, status in results.items() if status == "Incomplete"
+    }
+    for case in incomplete_results:
+        del results[case]
+
+    expected_results = {
+        case: status for case, status in results.items() if case in expected
+    }
+    unexpected_results = {
+        case: status for case, status in results.items() if case not in expected
+    }
+
+    # Missing cases are inferred from the caselist even if unrun.txt itself is
+    # missing or stale.  This is the invariant that prevents partial-run rate
+    # inflation.
+    unrun_cases = expected - set(expected_results)
+    declared_not_measured = explicit_unrun | skipped
+    undeclared_unrun = unrun_cases - declared_not_measured
+    stale_unrun = (explicit_unrun | skipped) & set(expected_results)
+
+    counts = Counter(expected_results.values())
+    strict_pass = counts["Pass"]
+    accepted = sum(counts[status] for status in ACCEPTED)
+    result_count = len(expected_results)
+    expected_count = len(expected)
+    crash_count = counts["Crash"]
+    hang_count = counts["DeviceHang"]
+
+    duplicate_cases = {
+        case: {
+            "observations": len(history),
+            "extra_observations": len(history) - 1,
+            "final_status": results.get(case, "Incomplete"),
+            "history": history,
+        }
+        for case, history in sorted(observation_history.items())
+        if len(history) > 1
+    }
+    duplicate_observations = sum(
+        item["extra_observations"] for item in duplicate_cases.values()
+    )
+
+    sidecar_unknown = {
+        name: sorted(cases - expected)
+        for name, cases in (
+            ("crashed.txt", crashed),
+            ("hung.txt", hung),
+            ("unrun.txt", explicit_unrun),
+            ("skipped.txt", skipped),
+        )
+        if cases - expected
+    }
+
+    errors: list[str] = []
+    warnings: list[str] = []
+    if not expected_count:
+        errors.append("caselist has no cases")
+    if expected_duplicates:
+        errors.append(
+            f"caselist has {sum(n - 1 for n in expected_duplicates.values())} "
+            "duplicate entry/entries"
+        )
+    if not qpa_files:
+        errors.append("no .qpa files found")
+    if unexpected_results:
+        errors.append(
+            f"{len(unexpected_results)} result case(s) are absent from the caselist"
+        )
+    if sidecar_unknown:
+        errors.append("one or more sidecars name cases absent from the caselist")
+    if undeclared_unrun:
+        errors.append(
+            f"{len(undeclared_unrun)} missing result case(s) are not declared by "
+            "unrun.txt/skipped.txt"
+        )
+    if stale_unrun:
+        warnings.append(
+            f"{len(stale_unrun)} case(s) declared unrun/skipped also have a result"
+        )
+    if duplicate_observations:
+        warnings.append(
+            f"{duplicate_observations} duplicate QPA observation(s); last result wins"
+        )
+    if incomplete_results:
+        warnings.append(
+            f"{len(incomplete_results)} QPA case(s) ended without a final result and were treated as unrun"
+        )
+
+    if errors:
+        state = "ERROR"
+    elif unrun_cases:
+        state = "INCOMPLETE"
+    else:
+        state = "OK"
+
+    return {
+        "version": version,
+        "inputs": {
+            "caselist": os.path.abspath(caselist),
+            "result_paths": [os.path.abspath(path) for path in result_paths],
+            "qpa_files": qpa_files,
+        },
+        "expected": expected_count,
+        "result": result_count,
+        "pass": strict_pass,
+        "accepted": accepted,
+        "crash": crash_count,
+        "hang": hang_count,
+        "unrun": len(unrun_cases),
+        "duplicate": duplicate_observations,
+        "counts": dict(sorted(counts.items())),
+        "coverage": {
+            "numerator": result_count,
+            "denominator": expected_count,
+            "rate": _ratio(result_count, expected_count),
+        },
+        "rates": {
+            # These are the report's conformance rates.  Expected, not merely
+            # measured results, is the denominator.
+            "denominator": "expected",
+            "strict_pass_only": _ratio(strict_pass, expected_count),
+            "conformance_accepted": _ratio(accepted, expected_count),
+            # Useful for comparison with qpa_report.py, whose denominator is
+            # cases with a result.  Never presented as the conformance rate.
+            "measured_only_strict_pass": _ratio(strict_pass, result_count),
+            "measured_only_conformance_accepted": _ratio(
+                accepted, result_count
+            ),
+        },
+        "strict_pass_rate": _ratio(strict_pass, expected_count),
+        "conformance_accepted_rate": _ratio(accepted, expected_count),
+        "validation": {
+            "state": state,
+            "ok": state == "OK",
+            "errors": errors,
+            "warnings": warnings,
+            "invariant_expected_equals_result_plus_unrun": (
+                expected_count == result_count + len(unrun_cases)
+            ),
+            "undeclared_unrun": sorted(undeclared_unrun),
+            "stale_unrun_or_skipped": sorted(stale_unrun),
+            "sidecar_cases_absent_from_caselist": sidecar_unknown,
+        },
+        "cases": {
+            "results": dict(sorted(expected_results.items())),
+            "unrun": sorted(unrun_cases),
+            "unexpected_results": dict(sorted(unexpected_results.items())),
+            "incomplete_results": sorted(incomplete_results),
+            "duplicate_results": duplicate_cases,
+            "duplicate_caselist_entries": dict(sorted(expected_duplicates.items())),
+        },
+    }
+
+
+def build_matrix(suites: dict[str, tuple[str, Sequence[str]]]) -> dict:
+    """Build all four version reports and their case-weighted aggregate."""
+
+    version_reports = {
+        version: build_version_report(version, *suites[version])
+        for version in VERSIONS
+    }
+
+    totals = {
+        key: sum(report[key] for report in version_reports.values())
+        for key in (
+            "expected",
+            "result",
+            "pass",
+            "accepted",
+            "crash",
+            "hang",
+            "unrun",
+            "duplicate",
+        )
+    }
+    status_counts: Counter[str] = Counter()
+    for report in version_reports.values():
+        status_counts.update(report["counts"])
+
+    states = {report["validation"]["state"] for report in version_reports.values()}
+    if "ERROR" in states:
+        overall_state = "ERROR"
+    elif "INCOMPLETE" in states:
+        overall_state = "INCOMPLETE"
+    else:
+        overall_state = "OK"
+
+    overall = {
+        **totals,
+        "counts": dict(sorted(status_counts.items())),
+        "aggregation": "weighted_by_expected_cases",
+        "coverage": {
+            "numerator": totals["result"],
+            "denominator": totals["expected"],
+            "rate": _ratio(totals["result"], totals["expected"]),
+        },
+        "rates": {
+            "denominator": "expected",
+            "strict_pass_only": _ratio(totals["pass"], totals["expected"]),
+            "conformance_accepted": _ratio(
+                totals["accepted"], totals["expected"]
+            ),
+            "measured_only_strict_pass": _ratio(
+                totals["pass"], totals["result"]
+            ),
+            "measured_only_conformance_accepted": _ratio(
+                totals["accepted"], totals["result"]
+            ),
+        },
+        "strict_pass_rate": _ratio(totals["pass"], totals["expected"]),
+        "conformance_accepted_rate": _ratio(
+            totals["accepted"], totals["expected"]
+        ),
+        "validation": {
+            "state": overall_state,
+            "ok": overall_state == "OK",
+            "invariant_expected_equals_result_plus_unrun": (
+                totals["expected"] == totals["result"] + totals["unrun"]
+            ),
+        },
+    }
+
+    return {
+        "schema_version": 1,
+        "generated_at": datetime.now(timezone.utc).isoformat(),
+        "accepted_statuses": list(ACCEPTED_STATUSES),
+        "rate_policy": {
+            "denominator": "unique expected cases from each caselist",
+            "unrun_cases": "included in the denominator and never accepted",
+            "duplicate_results": "last QPA result wins, matching qpa_report.py",
+        },
+        "versions": version_reports,
+        "overall": overall,
+    }
+
+
+def _percent(numerator: int, denominator: int) -> str:
+    if not denominator:
+        return "n/a"
+    return f"{100.0 * numerator / denominator:.2f}% ({numerator}/{denominator})"
+
+
+def render_markdown(report: dict) -> str:
+    """Render the compact terminal-facing conformance table."""
+
+    header = (
+        "| Suite | Expected | Result | Pass | Accepted | Crash | Hang | Unrun | "
+        "Duplicate | Coverage | Strict Pass-only | Conformance-accepted | Validation |"
+    )
+    separator = (
+        "|---|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|:---:|"
+    )
+    rows = [header, separator]
+    for version in VERSIONS:
+        item = report["versions"][version]
+        rows.append(
+            "| {version} | {expected} | {result} | {pass_count} | {accepted} | "
+            "{crash} | {hang} | {unrun} | {duplicate} | {coverage} | {strict} | "
+            "{accepted_rate} | {state} |".format(
+                version=version.upper(),
+                expected=item["expected"],
+                result=item["result"],
+                pass_count=item["pass"],
+                accepted=item["accepted"],
+                crash=item["crash"],
+                hang=item["hang"],
+                unrun=item["unrun"],
+                duplicate=item["duplicate"],
+                coverage=_percent(item["result"], item["expected"]),
+                strict=_percent(item["pass"], item["expected"]),
+                accepted_rate=_percent(item["accepted"], item["expected"]),
+                state=item["validation"]["state"],
+            )
+        )
+
+    overall = report["overall"]
+    rows.append(
+        "| **Overall (weighted)** | **{expected}** | **{result}** | **{pass_count}** | "
+        "**{accepted}** | **{crash}** | **{hang}** | **{unrun}** | **{duplicate}** | "
+        "**{coverage}** | **{strict}** | **{accepted_rate}** | **{state}** |".format(
+            expected=overall["expected"],
+            result=overall["result"],
+            pass_count=overall["pass"],
+            accepted=overall["accepted"],
+            crash=overall["crash"],
+            hang=overall["hang"],
+            unrun=overall["unrun"],
+            duplicate=overall["duplicate"],
+            coverage=_percent(overall["result"], overall["expected"]),
+            strict=_percent(overall["pass"], overall["expected"]),
+            accepted_rate=_percent(overall["accepted"], overall["expected"]),
+            state=overall["validation"]["state"],
+        )
+    )
+    rows.extend(
+        (
+            "",
+            "Rates use unique **Expected** caselist cases as the denominator; unrun cases "
+            "remain in that denominator and are not accepted.",
+            "Accepted statuses: " + ", ".join(f"`{s}`" for s in ACCEPTED_STATUSES) + ".",
+            "Duplicate is the number of extra QPA observations; the last observation wins.",
+        )
+    )
+
+    details: list[str] = []
+    for version in VERSIONS:
+        validation = report["versions"][version]["validation"]
+        messages = validation["errors"] + validation["warnings"]
+        if messages:
+            details.append(
+                f"- **{version.upper()} {validation['state']}**: " + "; ".join(messages)
+            )
+    if details:
+        rows.extend(("", "Validation details:", "", *details))
+    return "\n".join(rows)
+
+
+def _write_json(path: str, report: dict) -> None:
+    parent = os.path.dirname(os.path.abspath(path))
+    os.makedirs(parent, exist_ok=True)
+    with open(path, "w", encoding="utf-8", newline="\n") as fh:
+        json.dump(report, fh, indent=2, sort_keys=True)
+        fh.write("\n")
+
+
+def _parser() -> argparse.ArgumentParser:
+    parser = argparse.ArgumentParser(description=__doc__)
+    for version in VERSIONS:
+        parser.add_argument(
+            f"--{version}-caselist",
+            f"--{version}-case-list",
+            required=True,
+            help=f"{version.upper()} mustpass caselist",
+        )
+        parser.add_argument(
+            f"--{version}-results",
+            f"--{version}-result-dir",
+            f"--{version}-results-dir",
+            action="append",
+            required=True,
+            help=f"{version.upper()} result directory or QPA file (repeatable)",
+        )
+    parser.add_argument(
+        "--json",
+        dest="json_out",
+        default="cts_matrix_report.json",
+        help="JSON output path (default: ./cts_matrix_report.json)",
+    )
+    parser.add_argument(
+        "--allow-incomplete",
+        action="store_true",
+        help="return success even when validation is ERROR/INCOMPLETE",
+    )
+    return parser
+
+
+def main(argv: Optional[Iterable[str]] = None) -> int:
+    args = _parser().parse_args(argv)
+    suites = {
+        version: (
+            getattr(args, f"{version}_caselist"),
+            getattr(args, f"{version}_results"),
+        )
+        for version in VERSIONS
+    }
+    try:
+        report = build_matrix(suites)
+        _write_json(args.json_out, report)
+    except (OSError, ReportInputError) as exc:
+        print(f"cts_matrix_report: {exc}", file=sys.stderr)
+        return 2
+
+    print(render_markdown(report))
+    print(f"\nJSON: {os.path.abspath(args.json_out)}")
+    if not args.allow_incomplete and not report["overall"]["validation"]["ok"]:
+        return 1
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

tools/cts/scripts/cts_multi_report.py

@@ -0,0 +1,540 @@
+#!/usr/bin/env python
+"""Summarise any number of GL CTS suites and MobileGL backends.
+
+Each repeatable suite specification consists of four values: backend, label,
+caselist, and result directory.  For example::
+
+    python cts_multi_report.py \
+      --suite DirectGLES gl30 gl30-main.txt runs/gles/gl30 \
+      --suite DirectVulkan gl30 gl30-main.txt runs/vulkan/gl30 \
+      --markdown cts-summary.md --json cts-summary.json
+
+A compact comma form is accepted as well::
+
+    --suite=DirectGLES,gl31,gl31-main.txt,runs/gles/gl31
+
+Per-suite parsing and validation deliberately delegate to
+``cts_matrix_report`` so QPA ordering, sidecar upgrades, accepted statuses,
+unrun handling, and duplicate-result semantics cannot drift between reports.
+All conformance rates use unique expected caselist cases as their denominator.
+Backend subtotals and the overall total are therefore case-weighted, not an
+unweighted average of suite percentages.
+"""
+
+from __future__ import annotations
+
+import argparse
+from collections import Counter
+from dataclasses import dataclass
+from datetime import datetime, timezone
+import hashlib
+import json
+import os
+import sys
+from typing import Iterable, Optional, Sequence
+
+try:  # Direct script execution and package imports are both supported.
+    from . import cts_matrix_report
+except ImportError:  # pragma: no cover - covered through CLI-style tests
+    import cts_matrix_report
+
+
+SUPPORTED_BACKENDS = ("DirectGLES", "DirectVulkan")
+SUM_FIELDS = (
+    "expected",
+    "result",
+    "pass",
+    "accepted",
+    "crash",
+    "hang",
+    "unrun",
+    "duplicate",
+)
+
+
+class MultiReportInputError(ValueError):
+    """Suite specifications cannot produce an unambiguous report."""
+
+
+@dataclass(frozen=True)
+class SuiteSpec:
+    backend: str
+    label: str
+    caselist: str
+    result_dir: str
+
+    @property
+    def suite_id(self) -> str:
+        return f"{self.backend}/{self.label}"
+
+
+def _ratio(numerator: int, denominator: int) -> float:
+    return numerator / denominator if denominator else 0.0
+
+
+def _validation_state(items: Sequence[dict]) -> str:
+    states = {item["validation"]["state"] for item in items}
+    if "ERROR" in states:
+        return "ERROR"
+    if "INCOMPLETE" in states:
+        return "INCOMPLETE"
+    return "OK"
+
+
+def aggregate_reports(items: Sequence[dict], suite_state_keys: Sequence[str]) -> dict:
+    """Return an expected-case-weighted aggregate for suite reports."""
+
+    if len(items) != len(suite_state_keys):
+        raise MultiReportInputError("internal suite/state key count mismatch")
+
+    totals = {
+        field: sum(int(item[field]) for item in items)
+        for field in SUM_FIELDS
+    }
+    status_counts: Counter[str] = Counter()
+    for item in items:
+        status_counts.update(item["counts"])
+
+    state = _validation_state(items)
+    expected = totals["expected"]
+    result = totals["result"]
+    suite_states = {
+        key: item["validation"]["state"]
+        for key, item in zip(suite_state_keys, items)
+    }
+    return {
+        **totals,
+        "suite_count": len(items),
+        "counts": dict(sorted(status_counts.items())),
+        "aggregation": "weighted_by_expected_cases",
+        "coverage": {
+            "numerator": result,
+            "denominator": expected,
+            "rate": _ratio(result, expected),
+        },
+        "rates": {
+            "denominator": "expected",
+            "strict_pass_only": _ratio(totals["pass"], expected),
+            "conformance_accepted": _ratio(totals["accepted"], expected),
+            "measured_only_strict_pass": _ratio(totals["pass"], result),
+            "measured_only_conformance_accepted": _ratio(
+                totals["accepted"], result
+            ),
+        },
+        # Keep the convenient aliases used by cts_matrix_report consumers.
+        "strict_pass_rate": _ratio(totals["pass"], expected),
+        "conformance_accepted_rate": _ratio(totals["accepted"], expected),
+        "validation": {
+            "state": state,
+            "ok": state == "OK",
+            "suite_states": suite_states,
+            "invariant_expected_equals_result_plus_unrun": (
+                expected == result + totals["unrun"]
+            ),
+        },
+    }
+
+
+def _caselist_fingerprint(path: str) -> tuple[str, int]:
+    cases, _duplicates = cts_matrix_report.read_caselist(path)
+    payload = "\n".join(cases).encode("utf-8") + b"\n"
+    return hashlib.sha256(payload).hexdigest(), len(cases)
+
+
+def _read_provenance(
+    spec: SuiteSpec,
+    require_run_state: bool,
+    expected_run_identity: Optional[str],
+) -> dict:
+    path = os.path.join(spec.result_dir, "run_state.json")
+    if not os.path.isfile(path):
+        if require_run_state or expected_run_identity is not None:
+            raise MultiReportInputError(
+                "suite result directory has no run_state.json; invocation provenance "
+                f"cannot be verified: {spec.result_dir}"
+            )
+        return {"state": "UNVERIFIED", "run_state": None}
+    try:
+        with open(path, "r", encoding="utf-8") as handle:
+            state = json.load(handle)
+    except (OSError, UnicodeError, json.JSONDecodeError) as exc:
+        raise MultiReportInputError(f"cannot read suite run identity {path}: {exc}") from exc
+    if not isinstance(state, dict):
+        raise MultiReportInputError(f"suite run identity must be a JSON object: {path}")
+    if state.get("backend") != spec.backend:
+        raise MultiReportInputError(
+            f"suite {spec.suite_id} is labelled {spec.backend}, but run_state.json "
+            f"records {state.get('backend')!r}"
+        )
+    fingerprint, case_count = _caselist_fingerprint(spec.caselist)
+    if state.get("caselist_sha256") != fingerprint or state.get("case_count") != case_count:
+        raise MultiReportInputError(
+            f"suite {spec.suite_id} run_state.json belongs to a different caselist"
+        )
+    invocation_identity = state.get("invocation_identity")
+    if (
+        expected_run_identity is not None
+        and invocation_identity != expected_run_identity
+    ):
+        raise MultiReportInputError(
+            f"suite {spec.suite_id} run_state.json belongs to a different CTS invocation"
+        )
+    return {
+        "state": "VERIFIED",
+        "run_state": os.path.abspath(path),
+        "invocation_identity": invocation_identity,
+    }
+
+
+def _validate_specs(
+    specs: Sequence[SuiteSpec],
+    require_run_state: bool,
+    expected_run_identity: Optional[str],
+) -> dict[str, dict]:
+    if not specs:
+        raise MultiReportInputError("at least one --suite specification is required")
+
+    seen: set[tuple[str, str]] = set()
+    seen_result_dirs: list[tuple[str, str]] = []
+    provenance: dict[str, dict] = {}
+    for spec in specs:
+        if spec.backend not in SUPPORTED_BACKENDS:
+            raise MultiReportInputError(
+                f"unsupported backend {spec.backend!r}; expected one of "
+                + ", ".join(SUPPORTED_BACKENDS)
+            )
+        if not spec.label.strip():
+            raise MultiReportInputError("suite label cannot be empty")
+        identity = (spec.backend, spec.label)
+        if identity in seen:
+            raise MultiReportInputError(
+                f"duplicate suite specification for {spec.backend}/{spec.label}"
+            )
+        seen.add(identity)
+        if not os.path.isdir(spec.result_dir):
+            raise MultiReportInputError(
+                f"suite result directory does not exist: {spec.result_dir}"
+            )
+        physical_result_dir = os.path.normcase(
+            os.path.realpath(os.path.abspath(spec.result_dir))
+        )
+        for previous_dir, previous_suite in seen_result_dirs:
+            try:
+                common_dir = os.path.commonpath(
+                    [previous_dir, physical_result_dir]
+                )
+            except ValueError:
+                continue
+            if common_dir in (previous_dir, physical_result_dir):
+                raise MultiReportInputError(
+                    f"suite {spec.suite_id} uses a result directory which overlaps "
+                    f"{previous_suite}: {spec.result_dir}"
+                )
+        seen_result_dirs.append((physical_result_dir, spec.suite_id))
+        provenance[spec.suite_id] = _read_provenance(
+            spec, require_run_state, expected_run_identity
+        )
+    return provenance
+
+
+def build_report(
+    specs: Sequence[SuiteSpec],
+    require_run_state: bool = True,
+    expected_run_identity: Optional[str] = None,
+) -> dict:
+    """Build suite, per-backend, and overall serialisable reports."""
+
+    provenance = _validate_specs(
+        specs, require_run_state, expected_run_identity
+    )
+    suite_reports: list[dict] = []
+    backend_order: list[str] = []
+
+    for spec in specs:
+        if spec.backend not in backend_order:
+            backend_order.append(spec.backend)
+        item = cts_matrix_report.build_version_report(
+            spec.label, spec.caselist, [spec.result_dir]
+        )
+        # ``version`` is the generic label argument in build_version_report;
+        # expose explicit multi-report terminology while retaining all of its
+        # validation and case-level evidence.
+        item.pop("version", None)
+        item["backend"] = spec.backend
+        item["label"] = spec.label
+        item["suite_id"] = spec.suite_id
+        item["provenance"] = provenance[spec.suite_id]
+        if item["provenance"]["state"] == "UNVERIFIED":
+            item["validation"]["warnings"].append(
+                "result directory has no run_state.json; backend provenance is unverified"
+            )
+        suite_reports.append(item)
+
+    backends: dict[str, dict] = {}
+    for backend in backend_order:
+        backend_items = [
+            item for item in suite_reports if item["backend"] == backend
+        ]
+        labels = [item["label"] for item in backend_items]
+        aggregate = aggregate_reports(backend_items, labels)
+        aggregate["backend"] = backend
+        aggregate["suite_labels"] = labels
+        backends[backend] = aggregate
+
+    overall = aggregate_reports(
+        suite_reports, [item["suite_id"] for item in suite_reports]
+    )
+    overall["backend_count"] = len(backends)
+    overall["backends"] = backend_order
+
+    return {
+        "schema_version": 1,
+        "generated_at": datetime.now(timezone.utc).isoformat(),
+        "accepted_statuses": list(cts_matrix_report.ACCEPTED_STATUSES),
+        "rate_policy": {
+            "denominator": "unique expected cases from each suite caselist",
+            "unrun_cases": "included in the denominator and never accepted",
+            "backend_aggregation": "weighted by expected cases",
+            "overall_aggregation": "weighted by expected cases across backend-suite pairs",
+            "duplicate_results": "last QPA result wins, matching qpa_report.py",
+        },
+        "suites": suite_reports,
+        "backends": backends,
+        "overall": overall,
+    }
+
+
+def _percent(numerator: int, denominator: int) -> str:
+    if not denominator:
+        return "n/a"
+    return f"{100.0 * numerator / denominator:.2f}% ({numerator}/{denominator})"
+
+
+def _markdown_cell(value: object) -> str:
+    return str(value).replace("|", r"\|").replace("\r", " ").replace("\n", " ")
+
+
+def _table_row(backend: str, label: str, item: dict, bold: bool = False) -> str:
+    values = [
+        backend,
+        label,
+        str(item["expected"]),
+        str(item["result"]),
+        str(item["pass"]),
+        str(item["accepted"]),
+        str(item["crash"]),
+        str(item["hang"]),
+        str(item["unrun"]),
+        str(item["duplicate"]),
+        _percent(item["result"], item["expected"]),
+        _percent(item["pass"], item["expected"]),
+        _percent(item["accepted"], item["expected"]),
+        item["validation"]["state"],
+    ]
+    values = [_markdown_cell(value) for value in values]
+    if bold:
+        values = [f"**{value}**" for value in values]
+    return "| " + " | ".join(values) + " |"
+
+
+def render_markdown(report: dict) -> str:
+    lines = [
+        "# GL CTS multi-suite conformance report",
+        "",
+        (
+            "| Backend | Suite | Expected | Result | Pass | Accepted | Crash | Hang | "
+            "Unrun | Duplicate | Coverage | Strict Pass-only | Conformance-accepted | Validation |"
+        ),
+        "|---|---|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|:---:|",
+    ]
+
+    for backend in report["backends"]:
+        for item in report["suites"]:
+            if item["backend"] == backend:
+                lines.append(_table_row(backend, item["label"], item))
+        subtotal = report["backends"][backend]
+        lines.append(
+            _table_row(backend, f"{backend} weighted subtotal", subtotal, bold=True)
+        )
+
+    lines.append(
+        _table_row(
+            "All backends",
+            "Overall weighted",
+            report["overall"],
+            bold=True,
+        )
+    )
+    lines.extend(
+        [
+            "",
+            (
+                "Rates use unique **Expected** caselist cases as the denominator. "
+                "Unrun cases remain in the denominator and are not accepted."
+            ),
+            "Accepted statuses: "
+            + ", ".join(
+                f"`{status}`" for status in report["accepted_statuses"]
+            )
+            + ".",
+            (
+                "Duplicate is the number of extra QPA observations; the final "
+                "observation wins."
+            ),
+        ]
+    )
+
+    details: list[str] = []
+    for item in report["suites"]:
+        validation = item["validation"]
+        messages = validation["errors"] + validation["warnings"]
+        if messages:
+            details.append(
+                f"- **{_markdown_cell(item['suite_id'])} {validation['state']}**: "
+                + "; ".join(_markdown_cell(message) for message in messages)
+            )
+    if details:
+        lines.extend(["", "## Validation details", "", *details])
+    return "\n".join(lines) + "\n"
+
+
+def _write_text(path: str, contents: str) -> None:
+    absolute = os.path.abspath(path)
+    os.makedirs(os.path.dirname(absolute), exist_ok=True)
+    with open(absolute, "w", encoding="utf-8", newline="\n") as handle:
+        handle.write(contents)
+
+
+def _write_json(path: str, report: dict) -> None:
+    _write_text(path, json.dumps(report, indent=2, sort_keys=True) + "\n")
+
+
+def _normalise_compact_suite_args(argv: Sequence[str]) -> list[str]:
+    """Expand ``--suite=b,l,c,r`` into the four-value argparse form."""
+
+    result: list[str] = []
+    index = 0
+    while index < len(argv):
+        token = argv[index]
+        if token.startswith("--suite="):
+            compact = token.split("=", 1)[1]
+            parts = compact.split(",", 3)
+            if len(parts) != 4:
+                raise MultiReportInputError(
+                    "compact --suite expects backend,label,caselist,result-dir"
+                )
+            result.extend(["--suite", *parts])
+            index += 1
+            continue
+        if token == "--suite" and index + 1 < len(argv) and argv[index + 1].count(",") >= 3:
+            parts = argv[index + 1].split(",", 3)
+            result.extend(["--suite", *parts])
+            index += 2
+            continue
+        result.append(token)
+        index += 1
+    return result
+
+
+def _parser() -> argparse.ArgumentParser:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument(
+        "--suite",
+        action="append",
+        nargs=4,
+        required=True,
+        metavar=("BACKEND", "LABEL", "CASELIST", "RESULT_DIR"),
+        help=(
+            "suite specification; repeat for every backend/suite pair "
+            f"(backends: {', '.join(SUPPORTED_BACKENDS)})"
+        ),
+    )
+    parser.add_argument(
+        "--markdown",
+        default="cts_multi_report.md",
+        help="Markdown output path (default: ./cts_multi_report.md)",
+    )
+    parser.add_argument(
+        "--json",
+        dest="json_out",
+        default="cts_multi_report.json",
+        help="JSON output path (default: ./cts_multi_report.json)",
+    )
+    parser.add_argument(
+        "--allow-incomplete",
+        action="store_true",
+        help="return success even when one or more suites are ERROR/INCOMPLETE",
+    )
+    parser.add_argument(
+        "--adopt-legacy",
+        dest="allow_unverified_provenance",
+        action="store_true",
+        help="accept legacy result directories without run_state.json (provenance remains unverified)",
+    )
+    parser.add_argument(
+        "--allow-unverified-provenance",
+        dest="allow_unverified_provenance",
+        action="store_true",
+        help=argparse.SUPPRESS,
+    )
+    parser.add_argument(
+        "--expected-run-identity",
+        help="require every suite run_state.json to contain this controller fingerprint",
+    )
+    return parser
+
+
+def _specs_from_args(values: Sequence[Sequence[str]]) -> list[SuiteSpec]:
+    return [
+        SuiteSpec(
+            backend=backend.strip(),
+            label=label.strip(),
+            caselist=caselist,
+            result_dir=result_dir,
+        )
+        for backend, label, caselist, result_dir in values
+    ]
+
+
+def main(argv: Optional[Iterable[str]] = None) -> int:
+    raw_argv = list(argv) if argv is not None else sys.argv[1:]
+    try:
+        normalised = _normalise_compact_suite_args(raw_argv)
+    except MultiReportInputError as exc:
+        print(f"cts_multi_report: {exc}", file=sys.stderr)
+        return 2
+    args = _parser().parse_args(normalised)
+
+    if os.path.normcase(os.path.abspath(args.markdown)) == os.path.normcase(
+        os.path.abspath(args.json_out)
+    ):
+        print("cts_multi_report: Markdown and JSON paths must differ", file=sys.stderr)
+        return 2
+
+    try:
+        report = build_report(
+            _specs_from_args(args.suite),
+            require_run_state=not args.allow_unverified_provenance,
+            expected_run_identity=args.expected_run_identity,
+        )
+        markdown = render_markdown(report)
+        _write_text(args.markdown, markdown)
+        _write_json(args.json_out, report)
+    except (
+        OSError,
+        MultiReportInputError,
+        cts_matrix_report.ReportInputError,
+    ) as exc:
+        print(f"cts_multi_report: {exc}", file=sys.stderr)
+        return 2
+
+    print(markdown, end="")
+    print(f"\nMarkdown: {os.path.abspath(args.markdown)}")
+    print(f"JSON: {os.path.abspath(args.json_out)}")
+    if not args.allow_incomplete and not report["overall"]["validation"]["ok"]:
+        return 1
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

tools/cts/scripts/run_cts_local.py

@@ -0,0 +1,227 @@
+#!/usr/bin/env python
+"""Drive a glcts run on the local host, resuming across crashes.
+
+Local-host counterpart of run_cts.py: MobileGL crashes on some cases and glcts
+takes the whole process down with it, so a single invocation stops at the first
+crash. This runner re-invokes glcts with only the cases that have no result
+yet, records the case that was open when the process died as "Crash" (or
+"Hang" on a timeout), and repeats until the list is exhausted.
+
+Usage:
+  python run_cts_local.py --backend DirectVulkan \\
+      --glcts <path-to-glcts-binary> --lib <path-to-libMobileGL.so> \\
+      --caselist <mustpass.txt> --outdir <dir> [--env K=V ...]
+"""
+
+import argparse
+import glob
+import os
+import re
+import resource
+import subprocess
+import sys
+import time
+
+CASE_START = re.compile(r"^#beginTestCaseResult\s+(\S+)")
+CASE_END = re.compile(r"^#endTestCaseResult")
+CASE_TERM = re.compile(r"^#terminateTestCaseResult")
+
+
+def completed_cases(qpa_path):
+    """Return (finished_case_names, last_started_case_or_None)."""
+    finished = []
+    current = None
+    if not os.path.exists(qpa_path):
+        return finished, None
+    with open(qpa_path, "r", encoding="utf-8", errors="replace") as fh:
+        for line in fh:
+            m = CASE_START.match(line)
+            if m:
+                current = m.group(1)
+                continue
+            if current is not None and (CASE_END.match(line) or CASE_TERM.match(line)):
+                finished.append(current)
+                current = None
+    return finished, current
+
+
+def main():
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--backend", required=True, choices=["DirectGLES", "DirectVulkan"])
+    ap.add_argument("--glcts", required=True, help="path to the glcts binary")
+    ap.add_argument("--lib", required=True, help="path to libMobileGL.so")
+    ap.add_argument("--caselist", required=True)
+    ap.add_argument("--outdir", required=True)
+    ap.add_argument("--surface", default="fbo", help="--deqp-surface-type value")
+    # Without an explicit size, dEQP's FboRenderContext sizes the wrapper FBO to
+    # GL_MAX_RENDERBUFFER_SIZE (16384^2 here) and size-derived test allocations
+    # explode (a 4-sample 16K depth texture alone is 4 GiB).
+    ap.add_argument("--surface-size", type=int, default=256,
+                    help="--deqp-surface-width/height value")
+    # With DONT_CARE depth/stencil bits dEQP's FboRenderContext picks the first entry of
+    # its own format list, GL_DEPTH32F_STENCIL8. framebuffer_blit meanwhile hardcodes
+    # GL_DEPTH24_STENCIL8 for its own buffers whenever it detects an FBO surface, then
+    # blits depth between the two - which the spec forbids for mismatched formats, so a
+    # conformant driver has to fail it. Asking for a config the test agrees with avoids
+    # the contradiction instead of papering over it.
+    ap.add_argument("--gl-config-name", default="rgba8888d24s8",
+                    help="--deqp-gl-config-name value (empty string to leave it unset)")
+    ap.add_argument("--max-rounds", type=int, default=4000)
+    ap.add_argument("--max-empty-streak", type=int, default=64,
+                    help="abort after this many consecutive chunks that produce no log at all")
+    ap.add_argument("--chunk-timeout", type=int, default=1800,
+                    help="seconds before killing one glcts invocation (a wedged case never returns)")
+    # dEQP's watchdog aborts the process when a single case exceeds a hardcoded 30s
+    # (framework/common/tcuApp.hpp), which is not a hang on a CPU rasterizer - some
+    # texture_swizzle cases legitimately take ~17s each and cross it once the process is
+    # warm, so they came back as spurious Timeouts. dEQP's own default is off; the
+    # chunk-timeout above is what actually rescues a genuinely wedged case.
+    ap.add_argument("--watchdog", default="disable", choices=["enable", "disable"],
+                    help="--deqp-watchdog value")
+    ap.add_argument("--skip-file", default=None,
+                    help="file of case names to exclude, e.g. cases known to wedge the host")
+    ap.add_argument("--waiver-file", default=None,
+                    help="--deqp-waiver-file value, e.g. tools/cts/waivers/mobilegl-fbo-harness.xml")
+    ap.add_argument("--env", action="append", default=[], metavar="K=V",
+                    help="extra environment variable for glcts (repeatable)")
+    args = ap.parse_args()
+
+    os.makedirs(args.outdir, exist_ok=True)
+    glcts = os.path.abspath(args.glcts)
+    lib = os.path.abspath(args.lib)
+    # glcts resolves its gl_cts data tree relative to the binary's directory.
+    workdir = os.path.dirname(glcts)
+
+    with open(args.caselist, "r", encoding="utf-8") as fh:
+        remaining = [l.strip() for l in fh if l.strip() and not l.strip().startswith("#")]
+
+    skipped = []
+    if args.skip_file and os.path.isfile(args.skip_file):
+        with open(args.skip_file, "r", encoding="utf-8") as fh:
+            skip = {l.strip() for l in fh if l.strip() and not l.strip().startswith("#")}
+        skipped = [c for c in remaining if c in skip]
+        remaining = [c for c in remaining if c not in skip]
+        print(f"[run_cts_local] skipping {len(skipped)} case(s) from {args.skip_file}")
+
+    total = len(remaining)
+    print(f"[run_cts_local] {args.backend}: {total} cases")
+
+    env = dict(os.environ)
+    env["MOBILEGL_BACKEND_TYPE"] = args.backend
+    env["MOBILEGL_CTS_LIB"] = lib
+    for kv in args.env:
+        k, _, v = kv.partition("=")
+        env[k] = v
+
+    crashed = []
+    hung = []
+    done = set()
+    chunk = 0
+    started = time.time()
+    empty_streak = 0
+
+    # Resume: results in chunk files from an interrupted run still count. The
+    # case that was open when that run died is re-tried rather than assumed bad.
+    prior_chunks = sorted(glob.glob(os.path.join(args.outdir, "chunk*.qpa")))
+    for prior in prior_chunks:
+        finished, _ = completed_cases(prior)
+        done.update(finished)
+    if prior_chunks:
+        chunk = int(re.search(r"chunk(\d+)\.qpa$", prior_chunks[-1]).group(1)) + 1
+        remaining = [c for c in remaining if c not in done]
+        print(f"[run_cts_local] resuming: {len(done)} case(s) already measured, "
+              f"{len(remaining)} to go")
+
+    while remaining and chunk < args.max_rounds:
+        listfile = os.path.abspath(os.path.join(args.outdir, "remaining.txt"))
+        with open(listfile, "w", encoding="utf-8", newline="\n") as fh:
+            fh.write("\n".join(remaining) + "\n")
+
+        qpa = os.path.abspath(os.path.join(args.outdir, f"chunk{chunk:04d}.qpa"))
+        cmd = [
+            glcts,
+            f"--deqp-caselist-file={listfile}",
+            f"--deqp-surface-type={args.surface}",
+            f"--deqp-surface-width={args.surface_size}",
+            f"--deqp-surface-height={args.surface_size}",
+            "--deqp-terminate-on-device-lost=disable",
+            f"--deqp-watchdog={args.watchdog}",
+            "--deqp-log-images=disable",
+            "--deqp-log-shader-sources=disable",
+            f"--deqp-log-filename={qpa}",
+        ]
+        if args.gl_config_name:
+            cmd.append(f"--deqp-gl-config-name={args.gl_config_name}")
+        if args.waiver_file:
+            cmd.append(f"--deqp-waiver-file={os.path.abspath(args.waiver_file)}")
+        timed_out = False
+        try:
+            # RLIMIT_CORE=0: MobileGL asserts abort with a core dump, and writing
+            # a multi-GB glcts core image after every crash dominates wall time.
+            subprocess.run(cmd, cwd=workdir, env=env, timeout=args.chunk_timeout,
+                           stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL,
+                           start_new_session=True,
+                           preexec_fn=lambda: resource.setrlimit(resource.RLIMIT_CORE, (0, 0)))
+        except subprocess.TimeoutExpired:
+            timed_out = True
+            print(f"[run_cts_local] chunk {chunk:04d} timed out after {args.chunk_timeout}s",
+                  file=sys.stderr)
+
+        finished, in_flight = completed_cases(qpa)
+        for c in finished:
+            done.add(c)
+
+        progressed = len(finished)
+        if progressed > 0:
+            empty_streak = 0
+        if in_flight is not None:
+            if timed_out:
+                print(f"[run_cts_local] HANG in {in_flight} - quarantining it")
+                hung.append(in_flight)
+            else:
+                crashed.append(in_flight)
+            done.add(in_flight)
+            progressed += 1
+        elif progressed == 0:
+            empty_streak += 1
+            if empty_streak >= args.max_empty_streak:
+                print(f"[run_cts_local] ABORTING: {empty_streak} consecutive chunks produced no "
+                      f"output. Something systemic is wrong; refusing to label the rest of the "
+                      f"suite as crashes.", file=sys.stderr)
+                break
+            victim = remaining[0]
+            label = "Hang" if timed_out else "Crash"
+            print(f"[run_cts_local] no output at all; recording {victim} as {label}")
+            (hung if timed_out else crashed).append(victim)
+            done.add(victim)
+            progressed = 1
+
+        remaining = [c for c in remaining if c not in done]
+        elapsed = time.time() - started
+        print(
+            f"[run_cts_local] chunk {chunk:04d}: +{progressed} (done {len(done)}/{total}, "
+            f"crashes {len(crashed)}, hangs {len(hung)}, {elapsed / 60:.1f} min)"
+        )
+        chunk += 1
+
+    with open(os.path.join(args.outdir, "crashed.txt"), "w", encoding="utf-8", newline="\n") as fh:
+        fh.write("\n".join(crashed) + ("\n" if crashed else ""))
+    with open(os.path.join(args.outdir, "hung.txt"), "w", encoding="utf-8", newline="\n") as fh:
+        fh.write("\n".join(hung) + ("\n" if hung else ""))
+    with open(os.path.join(args.outdir, "unrun.txt"), "w", encoding="utf-8", newline="\n") as fh:
+        fh.write("\n".join(remaining) + ("\n" if remaining else ""))
+    if skipped:
+        with open(os.path.join(args.outdir, "skipped.txt"), "w", encoding="utf-8", newline="\n") as fh:
+            fh.write("\n".join(skipped) + "\n")
+
+    if remaining:
+        print(f"[run_cts_local] WARNING: {len(remaining)} cases were never run (see unrun.txt)",
+              file=sys.stderr)
+    print(f"[run_cts_local] finished: {len(done)}/{total} cases, {len(crashed)} crashes, "
+          f"{len(hung)} hangs, {chunk} invocations")
+    print(f"[run_cts_local] qpa chunks in {args.outdir}")
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

tools/cts/scripts/run_cts_windows.py

@@ -0,0 +1,878 @@
+#!/usr/bin/env python
+"""Run a local Windows glcts executable and resume across process failures.
+
+The desktop CTS normally runs a complete caselist in one process.  That is a
+poor fit for testing a developing OpenGL implementation: one access violation
+or GPU hang prevents every later case from running.  This driver gives each
+invocation the cases which have not produced a result yet, preserves one QPA
+and stdout/stderr pair per invocation, and starts another process after a
+crash.
+
+Existing ``chunkNNNN.qpa`` files and ``crashed.txt``/``hung.txt`` sidecars are
+read on startup, so invoking the same command and output directory resumes an
+interrupted run.  A timeout is based on *idle QPA time*, not total process wall
+time: a healthy invocation may legitimately run thousands of cases for hours.
+
+Example (values beginning with ``--`` use argparse's ``=`` spelling)::
+
+    py run_cts_windows.py \
+      --exe D:\\glcts\\glcts.exe --workdir D:\\glcts \
+      --caselist D:\\glcts\\mustpass\\gl30.txt --outdir D:\\results\\gl30 \
+      --backend DirectVulkan \
+      --deqp-arg=--deqp-surface-type=window
+"""
+
+from __future__ import annotations
+
+import argparse
+import hashlib
+import json
+import os
+from pathlib import Path
+import re
+import signal
+import subprocess
+import sys
+import tempfile
+import time
+from dataclasses import dataclass
+from datetime import datetime, timezone
+from typing import Iterable, Optional, Sequence
+
+
+CASE_START = re.compile(r"^#beginTestCaseResult\s+(\S+)")
+CASE_END = re.compile(r"^#endTestCaseResult(?:\s|$)")
+CASE_TERM = re.compile(r"^#terminateTestCaseResult(?:\s|$)")
+CASE_RESULT = re.compile(r'<Result\s+StatusCode="[^"]+"')
+CHUNK_ARTIFACT = re.compile(r"^chunk(\d+)(?:\.|$)", re.IGNORECASE)
+CHUNK_META = re.compile(r"^chunk(\d+)\.meta\.json$", re.IGNORECASE)
+RECOVERY_SIDECAR_NAMES = frozenset(
+    {"crashed.txt", "hung.txt", "unrun.txt", "skipped.txt", "remaining.txt"}
+)
+CONTROLLED_DEQP_OPTIONS = {
+    "--deqp-caselist-file",
+    "--deqp-log-filename",
+}
+ATOMIC_REPLACE_ATTEMPTS = 8
+ATOMIC_REPLACE_INITIAL_BACKOFF_SECONDS = 0.025
+ATOMIC_REPLACE_MAX_BACKOFF_SECONDS = 0.2
+
+
+class RunnerError(Exception):
+    """A user/configuration error which should not be attributed to a case."""
+
+
+@dataclass
+class QpaProgress:
+    """Cases recorded by a QPA and its unterminated tail, if any."""
+
+    recorded: list[str]
+    in_flight: Optional[str]
+    begin_count: int
+
+
+@dataclass
+class ProcessOutcome:
+    returncode: Optional[int]
+    duration_seconds: float
+    timed_out: bool = False
+    timeout_reason: Optional[str] = None
+    interrupted: bool = False
+    launch_error: Optional[str] = None
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat(timespec="seconds")
+
+
+def read_caselist(path: Path) -> list[str]:
+    """Read a dEQP text caselist, preserving order and removing duplicates."""
+
+    try:
+        lines = path.read_text(encoding="utf-8-sig", errors="strict").splitlines()
+    except (OSError, UnicodeError) as exc:
+        raise RunnerError(f"cannot read caselist {path}: {exc}") from exc
+
+    cases: list[str] = []
+    seen: set[str] = set()
+    for raw in lines:
+        case = raw.strip()
+        if not case or case.startswith("#") or case in seen:
+            continue
+        cases.append(case)
+        seen.add(case)
+    if not cases:
+        raise RunnerError(f"caselist contains no test cases: {path}")
+    return cases
+
+
+def read_name_set(path: Path) -> set[str]:
+    if not path.is_file():
+        return set()
+    try:
+        return {
+            line.strip()
+            for line in path.read_text(encoding="utf-8-sig", errors="replace").splitlines()
+            if line.strip() and not line.lstrip().startswith("#")
+        }
+    except OSError as exc:
+        raise RunnerError(f"cannot read recovery file {path}: {exc}") from exc
+
+
+def _replace_with_retry(source: Path, destination: Path) -> None:
+    """Replace a state file, tolerating brief Windows access-denied races.
+
+    Antivirus/indexing tools can momentarily open ``remaining.txt`` without
+    delete sharing.  Windows then reports either ``PermissionError`` or a
+    generic ``OSError`` carrying ``winerror == 5``.  Retry only those cases;
+    disk, path, and programming errors remain immediately visible.
+    """
+
+    for attempt in range(ATOMIC_REPLACE_ATTEMPTS):
+        try:
+            os.replace(source, destination)
+            return
+        except OSError as exc:
+            retryable = isinstance(exc, PermissionError) or getattr(exc, "winerror", None) == 5
+            if not retryable or attempt + 1 >= ATOMIC_REPLACE_ATTEMPTS:
+                raise
+            delay = min(
+                ATOMIC_REPLACE_INITIAL_BACKOFF_SECONDS * (2**attempt),
+                ATOMIC_REPLACE_MAX_BACKOFF_SECONDS,
+            )
+            time.sleep(delay)
+
+
+def atomic_write_text(path: Path, text: str) -> None:
+    """Replace a small state file without exposing a partially-written copy."""
+
+    path.parent.mkdir(parents=True, exist_ok=True)
+    fd, temporary = tempfile.mkstemp(prefix=f".{path.name}.", suffix=".tmp", dir=str(path.parent))
+    temporary_path = Path(temporary)
+    try:
+        with os.fdopen(fd, "w", encoding="utf-8", newline="\n") as handle:
+            handle.write(text)
+            handle.flush()
+            os.fsync(handle.fileno())
+        _replace_with_retry(temporary_path, path)
+    finally:
+        try:
+            temporary_path.unlink()
+        except FileNotFoundError:
+            pass
+
+
+def atomic_write_json(path: Path, value: object) -> None:
+    atomic_write_text(path, json.dumps(value, indent=2, sort_keys=True) + "\n")
+
+
+def write_case_file(path: Path, cases: Iterable[str]) -> None:
+    values = list(cases)
+    atomic_write_text(path, "\n".join(values) + ("\n" if values else ""))
+
+
+def scan_qpa(path: Path) -> QpaProgress:
+    """Return cases with a final result and the unfinished tail, if any.
+
+    ``#terminateTestCaseResult`` is a completed result (usually Crash or
+    Timeout). ``#endTestCaseResult`` only completes a case when its XML carried
+    a ``<Result StatusCode=...>``. A truncated case that already wrote Result is
+    also recoverable; a case with no Result remains eligible for a retry.
+    """
+
+    if not path.is_file():
+        return QpaProgress([], None, 0)
+
+    recorded: list[str] = []
+    current: Optional[str] = None
+    has_result = False
+    begin_count = 0
+    try:
+        with path.open("r", encoding="utf-8", errors="replace") as handle:
+            for raw_line in handle:
+                line = raw_line.lstrip("\ufeff")
+                match = CASE_START.match(line)
+                if match:
+                    if current is not None and has_result:
+                        recorded.append(current)
+                    current = match.group(1)
+                    has_result = False
+                    begin_count += 1
+                    continue
+                if current is not None and CASE_RESULT.search(line):
+                    has_result = True
+                    continue
+                if current is not None and CASE_TERM.match(line):
+                    recorded.append(current)
+                    current = None
+                    has_result = False
+                    continue
+                if current is not None and CASE_END.match(line):
+                    if has_result:
+                        recorded.append(current)
+                    current = None
+                    has_result = False
+    except OSError as exc:
+        raise RunnerError(f"cannot read QPA {path}: {exc}") from exc
+    if current is not None and has_result:
+        recorded.append(current)
+        current = None
+    return QpaProgress(recorded, current, begin_count)
+
+
+def numbered_files(outdir: Path, pattern: re.Pattern[str]) -> list[tuple[int, Path]]:
+    found: list[tuple[int, Path]] = []
+    try:
+        children = list(outdir.iterdir())
+    except OSError as exc:
+        raise RunnerError(f"cannot list output directory {outdir}: {exc}") from exc
+    for path in children:
+        match = pattern.match(path.name)
+        if match:
+            found.append((int(match.group(1)), path))
+    found.sort(key=lambda item: item[0])
+    return found
+
+
+def next_chunk_number(outdir: Path) -> int:
+    numbers = [number for number, _path in numbered_files(outdir, CHUNK_ARTIFACT)]
+    return max(numbers, default=-1) + 1
+
+
+def load_meta_classifications(outdir: Path, expected: set[str]) -> tuple[set[str], set[str]]:
+    """Recover an atomic classification written just before sidecar updates."""
+
+    crashed: set[str] = set()
+    hung: set[str] = set()
+    for _number, path in numbered_files(outdir, CHUNK_META):
+        try:
+            value = json.loads(path.read_text(encoding="utf-8"))
+        except (OSError, UnicodeError, json.JSONDecodeError):
+            # A damaged metadata file is diagnostic only.  QPA and sidecars are
+            # authoritative and must still allow recovery.
+            continue
+        if not isinstance(value, dict):
+            continue
+        case = value.get("classified_case")
+        classification = value.get("classification")
+        if not isinstance(case, str) or case not in expected:
+            continue
+        if classification == "DeviceHang":
+            hung.add(case)
+        elif classification == "Crash":
+            crashed.add(case)
+    crashed.difference_update(hung)
+    return crashed, hung
+
+
+def result_qpa_files(outdir: Path) -> list[Path]:
+    """Return every QPA a directory-based report would consume."""
+
+    found: list[Path] = []
+    try:
+        for root, directories, names in os.walk(outdir):
+            directories.sort(key=str.casefold)
+            for name in sorted(names, key=str.casefold):
+                if name.casefold().endswith(".qpa"):
+                    found.append(Path(root) / name)
+    except OSError as exc:
+        raise RunnerError(f"cannot scan output directory {outdir}: {exc}") from exc
+    return found
+
+
+def recover_results(outdir: Path, expected: set[str]) -> tuple[set[str], set[str], set[str]]:
+    recorded: set[str] = set()
+    for path in result_qpa_files(outdir):
+        progress = scan_qpa(path)
+        recorded.update(case for case in progress.recorded if case in expected)
+
+    crashed = read_name_set(outdir / "crashed.txt") & expected
+    hung = read_name_set(outdir / "hung.txt") & expected
+    meta_crashed, meta_hung = load_meta_classifications(outdir, expected)
+    crashed.update(meta_crashed)
+    hung.update(meta_hung)
+    crashed.difference_update(hung)
+    return recorded, crashed, hung
+
+
+def caselist_fingerprint(cases: Sequence[str]) -> str:
+    payload = "\n".join(cases).encode("utf-8") + b"\n"
+    return hashlib.sha256(payload).hexdigest()
+
+
+def recovery_artifacts(outdir: Path) -> list[Path]:
+    """Return prior-run evidence which must not be adopted implicitly."""
+
+    found = set(result_qpa_files(outdir))
+    try:
+        children = list(outdir.iterdir())
+    except OSError as exc:
+        raise RunnerError(f"cannot list output directory {outdir}: {exc}") from exc
+    found.update(
+        path
+        for path in children
+        if CHUNK_ARTIFACT.match(path.name)
+        or path.name.casefold() in RECOVERY_SIDECAR_NAMES
+    )
+    return sorted(
+        found,
+        key=lambda path: str(path.relative_to(outdir)).casefold(),
+    )
+
+
+def check_run_identity(
+    outdir: Path,
+    backend: str,
+    cases: Sequence[str],
+    invocation_identity: Optional[str] = None,
+    adopt_legacy: bool = False,
+) -> None:
+    """Refuse to silently mix different suites/backends in one result dir."""
+
+    path = outdir / "run_state.json"
+    fingerprint = caselist_fingerprint(cases)
+    if path.is_file():
+        try:
+            state = json.loads(path.read_text(encoding="utf-8"))
+        except (OSError, UnicodeError, json.JSONDecodeError) as exc:
+            raise RunnerError(f"cannot read run identity {path}: {exc}") from exc
+        if not isinstance(state, dict):
+            raise RunnerError(f"run identity must be a JSON object: {path}")
+        if state.get("backend") != backend:
+            raise RunnerError(
+                f"output directory belongs to backend {state.get('backend')!r}, not {backend!r}: {outdir}"
+            )
+        if state.get("caselist_sha256") != fingerprint:
+            raise RunnerError(f"output directory belongs to a different caselist: {outdir}")
+        stored_invocation_identity = state.get("invocation_identity")
+        if (
+            stored_invocation_identity is not None
+            or invocation_identity is not None
+        ) and stored_invocation_identity != invocation_identity:
+            raise RunnerError(f"output directory belongs to a different CTS invocation: {outdir}")
+        return
+
+    legacy_artifacts = recovery_artifacts(outdir)
+    if legacy_artifacts and not adopt_legacy:
+        examples = ", ".join(path.name for path in legacy_artifacts[:3])
+        raise RunnerError(
+            "output directory contains CTS recovery artifacts but no run_state.json; "
+            f"refusing to adopt unverified legacy results ({examples}). Re-run with "
+            "--adopt-legacy only after verifying the backend, caselist, and invocation."
+        )
+
+    atomic_write_json(
+        path,
+        {
+            "version": 1,
+            "backend": backend,
+            "case_count": len(cases),
+            "caselist_sha256": fingerprint,
+            "invocation_identity": invocation_identity,
+            "adopted_legacy": bool(legacy_artifacts),
+            "created_utc": utc_now(),
+        },
+    )
+
+
+def persist_sidecars(
+    outdir: Path,
+    ordered_cases: Sequence[str],
+    crashed: set[str],
+    hung: set[str],
+    remaining: Sequence[str],
+) -> None:
+    write_case_file(outdir / "crashed.txt", (case for case in ordered_cases if case in crashed))
+    write_case_file(outdir / "hung.txt", (case for case in ordered_cases if case in hung))
+    write_case_file(outdir / "unrun.txt", remaining)
+    write_case_file(outdir / "remaining.txt", remaining)
+
+
+def parse_environment(values: Sequence[str]) -> dict[str, str]:
+    result: dict[str, str] = {}
+    for value in values:
+        if "=" not in value:
+            raise RunnerError(f"--env expects NAME=VALUE, got {value!r}")
+        name, contents = value.split("=", 1)
+        if not name or "\x00" in name or "=" in name:
+            raise RunnerError(f"invalid environment variable name in {value!r}")
+        result[name] = contents
+    return result
+
+
+def validate_deqp_args(values: Sequence[str]) -> None:
+    for value in values:
+        option = value.split("=", 1)[0].lower()
+        if option in CONTROLLED_DEQP_OPTIONS:
+            raise RunnerError(f"{option} is controlled by this runner and cannot be supplied via --deqp-arg")
+
+
+def resolve_paths(
+    exe_value: str,
+    workdir_value: Optional[str],
+    caselist_value: str,
+    outdir_value: str,
+) -> tuple[Path, Path, Path, Path]:
+    launch_dir = Path.cwd()
+    requested_exe = Path(exe_value).expanduser()
+
+    if workdir_value:
+        workdir = Path(workdir_value).expanduser().resolve()
+    elif requested_exe.is_absolute():
+        workdir = requested_exe.resolve().parent
+    else:
+        workdir = launch_dir
+
+    if requested_exe.is_absolute():
+        exe = requested_exe.resolve()
+    else:
+        in_workdir = (workdir / requested_exe).resolve()
+        in_launch_dir = (launch_dir / requested_exe).resolve()
+        exe = in_workdir if in_workdir.is_file() else in_launch_dir
+
+    caselist = Path(caselist_value).expanduser().resolve()
+    outdir = Path(outdir_value).expanduser().resolve()
+    if not exe.is_file():
+        raise RunnerError(f"glcts executable does not exist: {exe}")
+    if not workdir.is_dir():
+        raise RunnerError(f"working directory does not exist: {workdir}")
+    if not caselist.is_file():
+        raise RunnerError(f"caselist does not exist: {caselist}")
+    return exe, workdir, caselist, outdir
+
+
+def qpa_signature(path: Path) -> Optional[tuple[int, int]]:
+    try:
+        stat = path.stat()
+    except FileNotFoundError:
+        return None
+    except OSError:
+        # A transient sharing violation must not kill a healthy process.  The
+        # next poll will retry and the idle clock retains its previous value.
+        return None
+    return stat.st_size, stat.st_mtime_ns
+
+
+def kill_process_tree(process: subprocess.Popen[bytes]) -> None:
+    """Force-stop the process and descendants, with a parent-only fallback."""
+
+    if process.poll() is not None:
+        return
+
+    if os.name == "nt":
+        # /T is essential: CTS/platform helpers can outlive the top-level
+        # process, retain the QPA/DLL, and poison the next continuation round.
+        taskkill = Path(os.environ.get("SystemRoot", r"C:\Windows")) / "System32" / "taskkill.exe"
+        command = [str(taskkill), "/PID", str(process.pid), "/T", "/F"]
+        try:
+            subprocess.run(
+                command,
+                stdin=subprocess.DEVNULL,
+                stdout=subprocess.DEVNULL,
+                stderr=subprocess.DEVNULL,
+                timeout=20,
+                check=False,
+                creationflags=getattr(subprocess, "CREATE_NO_WINDOW", 0),
+            )
+        except (OSError, subprocess.TimeoutExpired):
+            pass
+    else:
+        try:
+            os.killpg(process.pid, signal.SIGKILL)
+        except (ProcessLookupError, PermissionError, OSError):
+            pass
+
+    try:
+        process.wait(timeout=10)
+        return
+    except subprocess.TimeoutExpired:
+        pass
+
+    try:
+        process.kill()
+    except OSError:
+        pass
+    try:
+        process.wait(timeout=10)
+    except subprocess.TimeoutExpired:
+        pass
+
+
+def run_process(
+    command: Sequence[str],
+    workdir: Path,
+    environment: dict[str, str],
+    qpa_path: Path,
+    stdout_path: Path,
+    stderr_path: Path,
+    idle_timeout: float,
+    max_round_seconds: float,
+    poll_seconds: float,
+) -> ProcessOutcome:
+    """Run one CTS chunk, killing its tree only after QPA progress stalls."""
+
+    started = time.monotonic()
+    with stdout_path.open("wb") as stdout_handle, stderr_path.open("wb") as stderr_handle:
+        popen_options: dict[str, object] = {
+            "cwd": str(workdir),
+            "env": environment,
+            "stdin": subprocess.DEVNULL,
+            "stdout": stdout_handle,
+            "stderr": stderr_handle,
+        }
+        if os.name == "nt":
+            popen_options["creationflags"] = getattr(subprocess, "CREATE_NEW_PROCESS_GROUP", 0)
+        else:
+            popen_options["start_new_session"] = True
+
+        try:
+            process = subprocess.Popen(list(command), **popen_options)  # type: ignore[arg-type]
+        except OSError as exc:
+            message = f"failed to launch {command[0]}: {exc}\n"
+            stderr_handle.write(message.encode("utf-8", errors="replace"))
+            stderr_handle.flush()
+            return ProcessOutcome(None, time.monotonic() - started, launch_error=str(exc))
+
+        last_signature = qpa_signature(qpa_path)
+        last_progress = time.monotonic()
+        timed_out = False
+        timeout_reason: Optional[str] = None
+        interrupted = False
+
+        try:
+            while True:
+                try:
+                    returncode = process.wait(timeout=poll_seconds)
+                    break
+                except subprocess.TimeoutExpired:
+                    pass
+
+                now = time.monotonic()
+                signature = qpa_signature(qpa_path)
+                if signature is not None and signature != last_signature:
+                    last_signature = signature
+                    last_progress = now
+
+                if idle_timeout > 0 and now - last_progress >= idle_timeout:
+                    timed_out = True
+                    timeout_reason = "qpa-idle"
+                    kill_process_tree(process)
+                    returncode = process.poll()
+                    break
+                if max_round_seconds > 0 and now - started >= max_round_seconds:
+                    timed_out = True
+                    timeout_reason = "max-round"
+                    kill_process_tree(process)
+                    returncode = process.poll()
+                    break
+        except KeyboardInterrupt:
+            interrupted = True
+            kill_process_tree(process)
+            returncode = process.poll()
+
+    return ProcessOutcome(
+        returncode=returncode,
+        duration_seconds=time.monotonic() - started,
+        timed_out=timed_out,
+        timeout_reason=timeout_reason,
+        interrupted=interrupted,
+    )
+
+
+def build_parser() -> argparse.ArgumentParser:
+    parser = argparse.ArgumentParser(
+        description="Run Windows glcts against MobileGL, resuming across crashes and GPU hangs."
+    )
+    parser.add_argument("--exe", required=True, help="path to glcts.exe")
+    parser.add_argument(
+        "--workdir",
+        help="glcts working directory (default: executable directory, or current directory for a relative exe)",
+    )
+    parser.add_argument("--caselist", required=True, help="mustpass/caselist text file")
+    parser.add_argument("--outdir", required=True, help="persistent result directory")
+    parser.add_argument("--backend", required=True, choices=("DirectGLES", "DirectVulkan"))
+    parser.add_argument(
+        "--run-identity",
+        help="controller fingerprint for executable, data, arguments, and environment",
+    )
+    parser.add_argument(
+        "--adopt-legacy",
+        action="store_true",
+        help=(
+            "adopt existing chunk/sidecar results which predate run_state.json; "
+            "disabled by default because their provenance cannot be verified"
+        ),
+    )
+    parser.add_argument(
+        "--env",
+        action="append",
+        default=[],
+        metavar="NAME=VALUE",
+        help="extra child environment variable (repeatable)",
+    )
+    parser.add_argument(
+        "--deqp-arg",
+        action="append",
+        default=[],
+        metavar="ARG",
+        help="extra glcts argument; repeat and use --deqp-arg=--option=value for leading dashes",
+    )
+    parser.add_argument(
+        "--idle-timeout",
+        type=float,
+        default=300.0,
+        metavar="SECONDS",
+        help="kill a chunk after this many seconds with no QPA size/mtime change (0 disables; default: 300)",
+    )
+    parser.add_argument(
+        "--max-round-seconds",
+        type=float,
+        default=0.0,
+        metavar="SECONDS",
+        help="optional total wall limit for one invocation (0 disables; default: 0)",
+    )
+    parser.add_argument("--poll-seconds", type=float, default=1.0, help=argparse.SUPPRESS)
+    parser.add_argument(
+        "--max-rounds",
+        type=int,
+        default=10000,
+        help="maximum glcts invocations in this runner process (default: 10000)",
+    )
+    parser.add_argument(
+        "--max-empty-streak",
+        type=int,
+        default=3,
+        help="abort after this many invocations record no case at all; no case is blamed (default: 3)",
+    )
+    return parser
+
+
+def execute(args: argparse.Namespace) -> int:
+    if args.idle_timeout < 0 or args.max_round_seconds < 0:
+        raise RunnerError("timeout values must be non-negative")
+    if args.poll_seconds <= 0:
+        raise RunnerError("--poll-seconds must be greater than zero")
+    if args.max_rounds <= 0 or args.max_empty_streak <= 0:
+        raise RunnerError("--max-rounds and --max-empty-streak must be greater than zero")
+    validate_deqp_args(args.deqp_arg)
+    extra_environment = parse_environment(args.env)
+    exe, workdir, caselist_path, outdir = resolve_paths(
+        args.exe, args.workdir, args.caselist, args.outdir
+    )
+    outdir.mkdir(parents=True, exist_ok=True)
+
+    cases = read_caselist(caselist_path)
+    expected = set(cases)
+    check_run_identity(
+        outdir,
+        args.backend,
+        cases,
+        args.run_identity,
+        adopt_legacy=args.adopt_legacy,
+    )
+    recorded, crashed, hung = recover_results(outdir, expected)
+    accounted = recorded | crashed | hung
+    remaining = [case for case in cases if case not in accounted]
+    persist_sidecars(outdir, cases, crashed, hung, remaining)
+
+    existing_qpas = len(result_qpa_files(outdir))
+    print(
+        f"[run_cts_windows] {args.backend}: expected {len(cases)}, recovered {len(accounted)} "
+        f"({existing_qpas} QPA chunk(s), {len(crashed)} crash, {len(hung)} hang)"
+    )
+    if not remaining:
+        print(f"[run_cts_windows] complete: all {len(cases)} expected cases are accounted")
+        return 0
+
+    environment = os.environ.copy()
+    environment.update(extra_environment)
+    # --backend is authoritative even if the inherited or extra environment
+    # already contains a different value.
+    environment["MOBILEGL_BACKEND_TYPE"] = args.backend
+
+    common_arguments = [
+        "--deqp-terminate-on-device-lost=disable",
+        "--deqp-log-images=disable",
+        "--deqp-log-shader-sources=disable",
+    ]
+
+    chunk_number = next_chunk_number(outdir)
+    rounds = 0
+    empty_streak = 0
+    interrupted = False
+    fatal_launch_error = False
+    started_all = time.monotonic()
+
+    while remaining and rounds < args.max_rounds:
+        prefix = f"chunk{chunk_number:04d}"
+        remaining_path = outdir / "remaining.txt"
+        qpa_path = outdir / f"{prefix}.qpa"
+        stdout_path = outdir / f"{prefix}.stdout.log"
+        stderr_path = outdir / f"{prefix}.stderr.log"
+        meta_path = outdir / f"{prefix}.meta.json"
+
+        # The number allocator considers every chunk artifact, so these should
+        # be new.  Refuse to truncate evidence if a foreign file races us.
+        for artifact in (qpa_path, stdout_path, stderr_path, meta_path):
+            if artifact.exists():
+                raise RunnerError(f"refusing to overwrite existing chunk artifact: {artifact}")
+        write_case_file(remaining_path, remaining)
+
+        command = [
+            str(exe),
+            f"--deqp-caselist-file={remaining_path}",
+            f"--deqp-log-filename={qpa_path}",
+            *common_arguments,
+            *args.deqp_arg,
+        ]
+        print(
+            f"[run_cts_windows] {prefix}: launching {len(remaining)} remaining case(s); "
+            f"idle timeout {args.idle_timeout:g}s"
+        )
+        chunk_started_utc = utc_now()
+        outcome = run_process(
+            command,
+            workdir,
+            environment,
+            qpa_path,
+            stdout_path,
+            stderr_path,
+            args.idle_timeout,
+            args.max_round_seconds,
+            args.poll_seconds,
+        )
+        progress = scan_qpa(qpa_path)
+
+        before = set(accounted)
+        for case in progress.recorded:
+            if case in expected:
+                recorded.add(case)
+                accounted.add(case)
+
+        classification: Optional[str] = None
+        classified_case: Optional[str] = None
+        in_flight = progress.in_flight if progress.in_flight in expected else None
+        if not outcome.interrupted and in_flight is not None and in_flight not in accounted:
+            classified_case = in_flight
+            if outcome.timed_out:
+                classification = "DeviceHang"
+                hung.add(in_flight)
+                crashed.discard(in_flight)
+            else:
+                classification = "Crash"
+                crashed.add(in_flight)
+            accounted.add(in_flight)
+
+        new_accounted = len(accounted - before)
+        if new_accounted:
+            empty_streak = 0
+        elif progress.begin_count == 0:
+            # No #begin marker means there is no evidence that the first
+            # remaining case was reached.  Retry the identical caselist, then
+            # abort rather than manufacturing a string of false Crash results.
+            empty_streak += 1
+        else:
+            # A log containing only already-accounted cases is also no forward
+            # progress, but it is a different failure mode.  Bound it with the
+            # same guard while retaining the QPA evidence.
+            empty_streak += 1
+
+        remaining = [case for case in cases if case not in accounted]
+        metadata = {
+            "version": 1,
+            "chunk": chunk_number,
+            "started_utc": chunk_started_utc,
+            "finished_utc": utc_now(),
+            "duration_seconds": round(outcome.duration_seconds, 3),
+            "returncode": outcome.returncode,
+            "timed_out": outcome.timed_out,
+            "timeout_reason": outcome.timeout_reason,
+            "interrupted": outcome.interrupted,
+            "launch_error": outcome.launch_error,
+            "qpa_begin_count": progress.begin_count,
+            "qpa_recorded_count": len(progress.recorded),
+            "in_flight": progress.in_flight,
+            "classification": classification,
+            "classified_case": classified_case,
+            "new_accounted": new_accounted,
+            "remaining": len(remaining),
+        }
+        # Metadata is committed first.  If the runner itself dies between this
+        # write and the sidecars, recovery can reconstruct the classification.
+        atomic_write_json(meta_path, metadata)
+        persist_sidecars(outdir, cases, crashed, hung, remaining)
+
+        rounds += 1
+        elapsed_minutes = (time.monotonic() - started_all) / 60.0
+        detail = ""
+        if classification:
+            detail = f", {classification}={classified_case}"
+        if outcome.timed_out:
+            detail += f", timeout={outcome.timeout_reason}"
+        print(
+            f"[run_cts_windows] {prefix}: +{new_accounted}, accounted "
+            f"{len(accounted)}/{len(cases)}, remaining {len(remaining)}{detail} "
+            f"({elapsed_minutes:.1f} min)"
+        )
+
+        chunk_number += 1
+        if outcome.interrupted:
+            interrupted = True
+            print("[run_cts_windows] interrupted; process tree stopped and state preserved", file=sys.stderr)
+            break
+        if outcome.launch_error:
+            fatal_launch_error = True
+            print(
+                f"[run_cts_windows] launch failed; see {stderr_path.name}: {outcome.launch_error}",
+                file=sys.stderr,
+            )
+            break
+        if empty_streak >= args.max_empty_streak:
+            print(
+                f"[run_cts_windows] aborting after {empty_streak} consecutive chunks made no "
+                "case progress; no unobserved case was labelled Crash/Hang",
+                file=sys.stderr,
+            )
+            break
+
+    # Recompute from the persisted evidence so the final completeness claim is
+    # subject to the exact same recovery path as a later invocation.
+    final_recorded, final_crashed, final_hung = recover_results(outdir, expected)
+    final_accounted = final_recorded | final_crashed | final_hung
+    final_remaining = [case for case in cases if case not in final_accounted]
+    persist_sidecars(outdir, cases, final_crashed, final_hung, final_remaining)
+
+    if not final_remaining and final_accounted == expected:
+        print(
+            f"[run_cts_windows] complete: all {len(cases)} expected cases are accounted "
+            f"({len(final_crashed)} crash, {len(final_hung)} hang, {rounds} new invocation(s))"
+        )
+        return 0
+
+    print(
+        f"[run_cts_windows] INCOMPLETE: {len(final_accounted)}/{len(cases)} accounted; "
+        f"{len(final_remaining)} listed in {outdir / 'unrun.txt'}",
+        file=sys.stderr,
+    )
+    if interrupted:
+        return 130
+    if fatal_launch_error:
+        return 3
+    return 4
+
+
+def main(argv: Optional[Sequence[str]] = None) -> int:
+    parser = build_parser()
+    args = parser.parse_args(argv)
+    try:
+        return execute(args)
+    except RunnerError as exc:
+        print(f"[run_cts_windows] ERROR: {exc}", file=sys.stderr)
+        return 2
+    except OSError as exc:
+        print(f"[run_cts_windows] ERROR: filesystem/process operation failed: {exc}", file=sys.stderr)
+        return 2
+
+
+if __name__ == "__main__":
+    sys.exit(main())

tools/cts/scripts/sync_to_cts.py

@@ -21,6 +21,7 @@ CTS_TOOLS = os.path.dirname(HERE)
 COPIES = [
     (os.path.join(CTS_TOOLS, "platform"), "framework/platform/mobilegl", None),
     (os.path.join(CTS_TOOLS, "targets"), "targets/mobilegl", ["mobilegl.cmake", "ndk-modern.cmake"]),
+    (os.path.join(CTS_TOOLS, "targets"), "targets/mobilegl-desktop", ["mobilegl-desktop.cmake"]),
 ]
 
 

tools/cts/scripts/test_cts_matrix_report.py

@@ -0,0 +1,251 @@
+import contextlib
+import io
+import json
+import tempfile
+import unittest
+from pathlib import Path
+
+try:
+    from . import cts_matrix_report as report
+except ImportError:  # Allows `python test_cts_matrix_report.py`.
+    import cts_matrix_report as report
+
+
+def qpa_case(case, status):
+    return (
+        f"#beginTestCaseResult {case}\n"
+        f'<Result StatusCode="{status}"/>\n'
+        "#endTestCaseResult\n"
+    )
+
+
+class MatrixReportTests(unittest.TestCase):
+    def test_utf8_bom_caselist_matches_runner_semantics(self):
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            caselist = root / "cases.txt"
+            caselist.write_bytes(b"\xef\xbb\xbfcase.a\n")
+            results = root / "results"
+            results.mkdir()
+            (results / "run.qpa").write_text(
+                qpa_case("case.a", "Pass"), encoding="utf-8"
+            )
+
+            item = report.build_version_report("gl30", str(caselist), [str(results)])
+
+            self.assertEqual(1, item["expected"])
+            self.assertEqual({"case.a": "Pass"}, item["cases"]["results"])
+            self.assertEqual("OK", item["validation"]["state"])
+
+    def test_incomplete_qpa_is_unrun_not_a_completed_result(self):
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            caselist = root / "cases.txt"
+            caselist.write_text("case.a\n", encoding="utf-8")
+            results = root / "results"
+            results.mkdir()
+            (results / "run.qpa").write_text(
+                "#beginTestCaseResult case.a\n#endTestCaseResult\n",
+                encoding="utf-8",
+            )
+            (results / "unrun.txt").write_text("case.a\n", encoding="utf-8")
+
+            item = report.build_version_report("gl46", str(caselist), [str(results)])
+
+            self.assertEqual(0, item["result"])
+            self.assertEqual(1, item["unrun"])
+            self.assertEqual(["case.a"], item["cases"]["incomplete_results"])
+            self.assertEqual("INCOMPLETE", item["validation"]["state"])
+
+    def test_incomplete_qpa_is_upgraded_by_crash_sidecar(self):
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            caselist = root / "cases.txt"
+            caselist.write_text("case.a\n", encoding="utf-8")
+            results = root / "results"
+            results.mkdir()
+            (results / "run.qpa").write_text(
+                "#beginTestCaseResult case.a\n", encoding="utf-8"
+            )
+            (results / "crashed.txt").write_text("case.a\n", encoding="utf-8")
+
+            item = report.build_version_report("gl46", str(caselist), [str(results)])
+
+            self.assertEqual("Crash", item["cases"]["results"]["case.a"])
+            self.assertEqual([], item["cases"]["incomplete_results"])
+            self.assertEqual("OK", item["validation"]["state"])
+
+    def test_chunk_numbers_above_four_digits_use_numeric_order(self):
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            caselist = root / "cases.txt"
+            caselist.write_text("case.a\n", encoding="utf-8")
+            results = root / "results"
+            results.mkdir()
+            (results / "alpha.qpa").write_text("# no results\n", encoding="utf-8")
+            (results / "chunk9999.qpa").write_text(
+                qpa_case("case.a", "Fail"), encoding="utf-8"
+            )
+            (results / "chunk10000.qpa").write_text(
+                qpa_case("case.a", "Pass"), encoding="utf-8"
+            )
+            (results / "zeta.qpa").write_text("# no results\n", encoding="utf-8")
+
+            item = report.build_version_report(
+                "gl46", str(caselist), [str(results)]
+            )
+
+            self.assertEqual(
+                ["alpha.qpa", "chunk9999.qpa", "chunk10000.qpa", "zeta.qpa"],
+                [Path(path).name for path in item["inputs"]["qpa_files"]],
+            )
+            self.assertEqual("Pass", item["cases"]["results"]["case.a"])
+            self.assertEqual(1, item["duplicate"])
+
+    def test_qpa_sidecars_duplicates_and_expected_denominator(self):
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            caselist = root / "gl30.txt"
+            caselist.write_text("\n".join("abcdefg") + "\n", encoding="utf-8")
+            results = root / "results"
+            results.mkdir()
+            (results / "chunk0000.qpa").write_text(
+                qpa_case("a", "Fail") + "#beginTestCaseResult e\n",
+                encoding="utf-8",
+            )
+            (results / "chunk0001.qpa").write_text(
+                qpa_case("a", "Pass")
+                + qpa_case("b", "NotSupported")
+                + qpa_case("c", "QualityWarning")
+                + qpa_case("d", "Fail"),
+                encoding="utf-8",
+            )
+            (results / "crashed.txt").write_text("e\n", encoding="utf-8")
+            (results / "hung.txt").write_text("f\n", encoding="utf-8")
+            (results / "unrun.txt").write_text("g\n", encoding="utf-8")
+
+            item = report.build_version_report(
+                "gl30", str(caselist), [str(results)]
+            )
+
+            self.assertEqual(item["expected"], 7)
+            self.assertEqual(item["result"], 6)
+            self.assertEqual(item["pass"], 1)
+            self.assertEqual(item["accepted"], 3)
+            self.assertEqual(item["crash"], 1)
+            self.assertEqual(item["hang"], 1)
+            self.assertEqual(item["unrun"], 1)
+            self.assertEqual(item["duplicate"], 1)
+            self.assertEqual(item["cases"]["results"]["a"], "Pass")
+            self.assertEqual(item["cases"]["results"]["e"], "Crash")
+            self.assertEqual(item["cases"]["results"]["f"], "DeviceHang")
+            self.assertAlmostEqual(item["strict_pass_rate"], 1 / 7)
+            self.assertAlmostEqual(item["conformance_accepted_rate"], 3 / 7)
+            self.assertAlmostEqual(
+                item["rates"]["measured_only_conformance_accepted"], 3 / 6
+            )
+            self.assertEqual(item["validation"]["state"], "INCOMPLETE")
+            self.assertEqual(item["validation"]["errors"], [])
+            self.assertTrue(
+                item["validation"]["invariant_expected_equals_result_plus_unrun"]
+            )
+
+    def test_missing_result_is_inferred_and_rejected_when_not_declared(self):
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            caselist = root / "cases.txt"
+            caselist.write_text("a\nb\n", encoding="utf-8")
+            results = root / "results"
+            results.mkdir()
+            (results / "run.qpa").write_text(qpa_case("a", "Pass"), encoding="utf-8")
+
+            item = report.build_version_report(
+                "gl31", str(caselist), [str(results)]
+            )
+
+            self.assertEqual(item["unrun"], 1)
+            self.assertEqual(item["cases"]["unrun"], ["b"])
+            self.assertEqual(item["validation"]["state"], "ERROR")
+            self.assertEqual(item["validation"]["undeclared_unrun"], ["b"])
+            self.assertIn("not declared", item["validation"]["errors"][0])
+            self.assertEqual(item["strict_pass_rate"], 0.5)
+
+    def test_cli_emits_markdown_json_and_weighted_overall(self):
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            statuses = {
+                "gl30": "Pass",
+                "gl31": "Fail",
+                "gl32": "NotSupported",
+                "gl33": None,
+            }
+            argv = []
+            for version, status in statuses.items():
+                caselist = root / f"{version}.txt"
+                caselist.write_text(f"{version}.case\n", encoding="utf-8")
+                result_dir = root / f"{version}-results"
+                result_dir.mkdir()
+                qpa = result_dir / "run.qpa"
+                qpa.write_text(
+                    qpa_case(f"{version}.case", status) if status else "# empty run\n",
+                    encoding="utf-8",
+                )
+                if status is None:
+                    (result_dir / "unrun.txt").write_text(
+                        f"{version}.case\n", encoding="utf-8"
+                    )
+                argv.extend(
+                    [
+                        f"--{version}-caselist",
+                        str(caselist),
+                        f"--{version}-results",
+                        str(result_dir),
+                    ]
+                )
+            json_path = root / "matrix.json"
+            argv.extend(["--json", str(json_path)])
+
+            stdout = io.StringIO()
+            with contextlib.redirect_stdout(stdout):
+                rc = report.main(argv)
+
+            self.assertEqual(rc, 1)  # GL33 is explicitly incomplete.
+            markdown = stdout.getvalue()
+            self.assertIn("| Suite | Expected | Result", markdown)
+            self.assertIn("| **Overall (weighted)**", markdown)
+            payload = json.loads(json_path.read_text(encoding="utf-8"))
+            overall = payload["overall"]
+            self.assertEqual(overall["expected"], 4)
+            self.assertEqual(overall["result"], 3)
+            self.assertEqual(overall["pass"], 1)
+            self.assertEqual(overall["accepted"], 2)
+            self.assertEqual(overall["unrun"], 1)
+            self.assertEqual(overall["strict_pass_rate"], 0.25)
+            self.assertEqual(overall["conformance_accepted_rate"], 0.5)
+            self.assertEqual(overall["aggregation"], "weighted_by_expected_cases")
+            self.assertEqual(overall["validation"]["state"], "INCOMPLETE")
+
+    def test_duplicate_caselist_and_unexpected_result_are_validation_errors(self):
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            caselist = root / "cases.txt"
+            caselist.write_text("a\na\n", encoding="utf-8")
+            results = root / "results"
+            results.mkdir()
+            (results / "run.qpa").write_text(
+                qpa_case("a", "Pass") + qpa_case("outside", "Pass"),
+                encoding="utf-8",
+            )
+
+            item = report.build_version_report(
+                "gl32", str(caselist), [str(results)]
+            )
+
+            self.assertEqual(item["cases"]["duplicate_caselist_entries"], {"a": 2})
+            self.assertEqual(item["cases"]["unexpected_results"], {"outside": "Pass"})
+            self.assertEqual(item["validation"]["state"], "ERROR")
+            self.assertEqual(len(item["validation"]["errors"]), 2)
+
+
+if __name__ == "__main__":
+    unittest.main()

tools/cts/scripts/test_cts_multi_report.py

@@ -0,0 +1,342 @@
+import contextlib
+import io
+import json
+import tempfile
+import unittest
+from pathlib import Path
+
+try:
+    from . import cts_multi_report as report
+except ImportError:  # Allows `python test_cts_multi_report.py`.
+    import cts_multi_report as report
+
+
+def qpa_case(case: str, status: str) -> str:
+    return (
+        f"#beginTestCaseResult {case}\n"
+        f'<Result StatusCode="{status}"/>\n'
+        "#endTestCaseResult\n"
+    )
+
+
+def write_run_state(
+    caselist: Path,
+    result_dir: Path,
+    backend: str,
+    invocation_identity=None,
+) -> None:
+    fingerprint, case_count = report._caselist_fingerprint(str(caselist))
+    (result_dir / "run_state.json").write_text(
+        json.dumps(
+            {
+                "version": 1,
+                "backend": backend,
+                "case_count": case_count,
+                "caselist_sha256": fingerprint,
+                "invocation_identity": invocation_identity,
+            }
+        ),
+        encoding="utf-8",
+    )
+
+
+def make_inputs(
+    root: Path,
+    name: str,
+    cases: list[str],
+    qpa: str,
+    backend: str = "DirectGLES",
+):
+    caselist = root / f"{name}.txt"
+    caselist.write_text("\n".join(cases) + "\n", encoding="utf-8")
+    result_dir = root / f"{name}-results"
+    result_dir.mkdir()
+    (result_dir / "chunk0000.qpa").write_text(qpa, encoding="utf-8")
+    write_run_state(caselist, result_dir, backend)
+    return caselist, result_dir
+
+
+class MultiReportTests(unittest.TestCase):
+    def test_backend_aggregate_is_weighted_by_expected_cases(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            small_cases, small_results = make_inputs(
+                root, "small", ["small.pass"], qpa_case("small.pass", "Pass")
+            )
+            large_cases, large_results = make_inputs(
+                root,
+                "large",
+                ["large.pass", "large.crash", "large.hang"],
+                qpa_case("large.pass", "Pass"),
+            )
+            (large_results / "crashed.txt").write_text(
+                "large.crash\n", encoding="utf-8"
+            )
+            (large_results / "hung.txt").write_text(
+                "large.hang\n", encoding="utf-8"
+            )
+
+            payload = report.build_report(
+                [
+                    report.SuiteSpec(
+                        "DirectGLES", "small", str(small_cases), str(small_results)
+                    ),
+                    report.SuiteSpec(
+                        "DirectGLES", "large", str(large_cases), str(large_results)
+                    ),
+                ]
+            )
+
+            aggregate = payload["backends"]["DirectGLES"]
+            self.assertEqual(4, aggregate["expected"])
+            self.assertEqual(4, aggregate["result"])
+            self.assertEqual(2, aggregate["pass"])
+            self.assertEqual(2, aggregate["accepted"])
+            self.assertEqual(1, aggregate["crash"])
+            self.assertEqual(1, aggregate["hang"])
+            self.assertEqual(0, aggregate["unrun"])
+            # (1 accepted + 1 accepted) / (1 expected + 3 expected), not
+            # the unweighted mean of 100% and 33.3%.
+            self.assertEqual(0.5, aggregate["conformance_accepted_rate"])
+            self.assertEqual("weighted_by_expected_cases", aggregate["aggregation"])
+            self.assertEqual("OK", aggregate["validation"]["state"])
+
+    def test_declared_unrun_is_incomplete_and_cli_returns_nonzero(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            caselist, result_dir = make_inputs(
+                root, "missing", ["case.a", "case.b"], qpa_case("case.a", "Pass")
+            )
+            (result_dir / "unrun.txt").write_text("case.b\n", encoding="utf-8")
+            markdown_path = root / "report.md"
+            json_path = root / "report.json"
+            argv = [
+                "--suite",
+                "DirectGLES",
+                "gl30",
+                str(caselist),
+                str(result_dir),
+                "--markdown",
+                str(markdown_path),
+                "--json",
+                str(json_path),
+            ]
+
+            with contextlib.redirect_stdout(io.StringIO()):
+                returncode = report.main(argv)
+
+            self.assertEqual(1, returncode)
+            self.assertTrue(markdown_path.is_file())
+            payload = json.loads(json_path.read_text(encoding="utf-8"))
+            suite = payload["suites"][0]
+            self.assertEqual(2, suite["expected"])
+            self.assertEqual(1, suite["result"])
+            self.assertEqual(1, suite["unrun"])
+            self.assertEqual("INCOMPLETE", suite["validation"]["state"])
+            self.assertEqual("INCOMPLETE", payload["overall"]["validation"]["state"])
+            self.assertEqual(0.5, payload["overall"]["conformance_accepted_rate"])
+
+    def test_dual_backend_cli_outputs_markdown_and_json(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            caselist = root / "gl30.txt"
+            caselist.write_text("gl30.case\n", encoding="utf-8")
+            gles = root / "gles"
+            vulkan = root / "vulkan"
+            gles.mkdir()
+            vulkan.mkdir()
+            (gles / "run.qpa").write_text(
+                qpa_case("gl30.case", "Pass"), encoding="utf-8"
+            )
+            (vulkan / "run.qpa").write_text(
+                qpa_case("gl30.case", "Fail"), encoding="utf-8"
+            )
+            write_run_state(caselist, gles, "DirectGLES")
+            write_run_state(caselist, vulkan, "DirectVulkan")
+            markdown_path = root / "dual.md"
+            json_path = root / "dual.json"
+            argv = [
+                f"--suite=DirectGLES,gl30,{caselist},{gles}",
+                "--suite",
+                "DirectVulkan",
+                "gl30",
+                str(caselist),
+                str(vulkan),
+                "--markdown",
+                str(markdown_path),
+                "--json",
+                str(json_path),
+            ]
+
+            stdout = io.StringIO()
+            with contextlib.redirect_stdout(stdout):
+                returncode = report.main(argv)
+
+            self.assertEqual(0, returncode)
+            payload = json.loads(json_path.read_text(encoding="utf-8"))
+            self.assertEqual({"DirectGLES", "DirectVulkan"}, set(payload["backends"]))
+            self.assertEqual(1, payload["backends"]["DirectGLES"]["accepted"])
+            self.assertEqual(0, payload["backends"]["DirectVulkan"]["accepted"])
+            self.assertEqual(2, payload["overall"]["expected"])
+            self.assertEqual(1, payload["overall"]["accepted"])
+            self.assertEqual(0.5, payload["overall"]["conformance_accepted_rate"])
+            markdown = markdown_path.read_text(encoding="utf-8")
+            self.assertIn("DirectGLES weighted subtotal", markdown)
+            self.assertIn("DirectVulkan weighted subtotal", markdown)
+            self.assertIn("Overall weighted", markdown)
+            self.assertIn("Markdown:", stdout.getvalue())
+
+    def test_duplicate_qpa_result_uses_last_observation(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            caselist, result_dir = make_inputs(
+                root,
+                "duplicate",
+                ["case.a"],
+                qpa_case("case.a", "Fail"),
+                backend="DirectVulkan",
+            )
+            (result_dir / "chunk0001.qpa").write_text(
+                qpa_case("case.a", "Pass"), encoding="utf-8"
+            )
+
+            payload = report.build_report(
+                [
+                    report.SuiteSpec(
+                        "DirectVulkan", "gl33", str(caselist), str(result_dir)
+                    )
+                ]
+            )
+
+            suite = payload["suites"][0]
+            self.assertEqual("Pass", suite["cases"]["results"]["case.a"])
+            self.assertEqual(1, suite["duplicate"])
+            self.assertEqual(1, payload["overall"]["duplicate"])
+            self.assertEqual(1.0, payload["overall"]["strict_pass_rate"])
+            self.assertEqual("OK", suite["validation"]["state"])
+            self.assertIn("last result wins", suite["validation"]["warnings"][0])
+
+    def test_backend_provenance_mismatch_is_rejected(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            caselist, result_dir = make_inputs(
+                root,
+                "provenance",
+                ["case.a"],
+                qpa_case("case.a", "Pass"),
+                backend="DirectVulkan",
+            )
+            with self.assertRaises(report.MultiReportInputError):
+                report.build_report(
+                    [report.SuiteSpec("DirectGLES", "gl30", str(caselist), str(result_dir))]
+                )
+
+    def test_missing_provenance_requires_explicit_legacy_opt_in(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            caselist, result_dir = make_inputs(
+                root, "legacy", ["case.a"], qpa_case("case.a", "Pass")
+            )
+            (result_dir / "run_state.json").unlink()
+            spec = report.SuiteSpec("DirectGLES", "gl30", str(caselist), str(result_dir))
+            with self.assertRaises(report.MultiReportInputError):
+                report.build_report([spec])
+            payload = report.build_report([spec], require_run_state=False)
+            self.assertEqual("UNVERIFIED", payload["suites"][0]["provenance"]["state"])
+
+    def test_expected_run_identity_accepts_match_and_rejects_mismatch(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            caselist, result_dir = make_inputs(
+                root, "identity", ["case.a"], qpa_case("case.a", "Pass")
+            )
+            write_run_state(
+                caselist, result_dir, "DirectGLES", invocation_identity="identity-a"
+            )
+            spec = report.SuiteSpec(
+                "DirectGLES", "gl30", str(caselist), str(result_dir)
+            )
+
+            payload = report.build_report(
+                [spec], expected_run_identity="identity-a"
+            )
+            self.assertEqual(
+                "identity-a",
+                payload["suites"][0]["provenance"]["invocation_identity"],
+            )
+            with self.assertRaises(report.MultiReportInputError):
+                report.build_report([spec], expected_run_identity="identity-b")
+
+    def test_expected_identity_rejects_legacy_state_and_missing_state(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            caselist, result_dir = make_inputs(
+                root, "legacy-identity", ["case.a"], qpa_case("case.a", "Pass")
+            )
+            spec = report.SuiteSpec(
+                "DirectGLES", "gl30", str(caselist), str(result_dir)
+            )
+            with self.assertRaises(report.MultiReportInputError):
+                report.build_report([spec], expected_run_identity="identity-a")
+
+            (result_dir / "run_state.json").unlink()
+            with self.assertRaises(report.MultiReportInputError):
+                report.build_report(
+                    [spec],
+                    require_run_state=False,
+                    expected_run_identity="identity-a",
+                )
+
+    def test_duplicate_physical_result_directory_is_rejected(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            caselist, result_dir = make_inputs(
+                root, "duplicate-dir", ["case.a"], qpa_case("case.a", "Pass")
+            )
+            with self.assertRaises(report.MultiReportInputError):
+                report.build_report(
+                    [
+                        report.SuiteSpec(
+                            "DirectGLES", "gl30", str(caselist), str(result_dir)
+                        ),
+                        report.SuiteSpec(
+                            "DirectGLES",
+                            "gl31",
+                            str(caselist),
+                            str(result_dir / "."),
+                        ),
+                    ]
+                )
+
+    def test_ancestor_and_descendant_result_directories_are_rejected(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            caselist = root / "cases.txt"
+            caselist.write_text("case.a\n", encoding="utf-8")
+            parent = root / "results"
+            child = parent / "nested"
+            child.mkdir(parents=True)
+            (parent / "chunk0000.qpa").write_text(
+                qpa_case("case.a", "Pass"), encoding="utf-8"
+            )
+            (child / "chunk0000.qpa").write_text(
+                qpa_case("case.a", "Fail"), encoding="utf-8"
+            )
+            write_run_state(caselist, parent, "DirectGLES")
+            write_run_state(caselist, child, "DirectVulkan")
+
+            with self.assertRaises(report.MultiReportInputError):
+                report.build_report(
+                    [
+                        report.SuiteSpec(
+                            "DirectGLES", "gl30", str(caselist), str(parent)
+                        ),
+                        report.SuiteSpec(
+                            "DirectVulkan", "gl30", str(caselist), str(child)
+                        ),
+                    ]
+                )
+
+
+if __name__ == "__main__":
+    unittest.main()

tools/cts/scripts/test_run_cts_windows.py

@@ -0,0 +1,401 @@
+import sys
+import tempfile
+import time
+import unittest
+from pathlib import Path
+from unittest import mock
+
+import run_cts_windows as runner
+
+
+def qpa_closed(case: str, status: str = "Pass") -> str:
+    return (
+        f"#beginTestCaseResult {case}\n"
+        f'<Result StatusCode="{status}">ok</Result>\n'
+        "#endTestCaseResult\n"
+    )
+
+
+def command_path(command, option):
+    prefix = option + "="
+    return Path(next(value[len(prefix) :] for value in command if value.startswith(prefix)))
+
+
+class AtomicWriteTests(unittest.TestCase):
+    def test_access_denied_retries_then_replace_succeeds(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            target = Path(temporary) / "remaining.txt"
+            real_replace = runner.os.replace
+            attempts = 0
+
+            def flaky_replace(source, destination):
+                nonlocal attempts
+                attempts += 1
+                if attempts == 1:
+                    raise PermissionError(13, "temporarily denied", str(destination))
+                if attempts == 2:
+                    error = OSError("temporary WinError 5")
+                    error.winerror = 5
+                    raise error
+                real_replace(source, destination)
+
+            with mock.patch.object(runner.os, "replace", side_effect=flaky_replace), mock.patch.object(
+                runner.time, "sleep"
+            ) as sleep:
+                runner.atomic_write_text(target, "case.a\n")
+
+            self.assertEqual(3, attempts)
+            self.assertEqual("case.a\n", target.read_text(encoding="utf-8"))
+            self.assertEqual(2, sleep.call_count)
+            self.assertEqual(
+                [
+                    mock.call(runner.ATOMIC_REPLACE_INITIAL_BACKOFF_SECONDS),
+                    mock.call(runner.ATOMIC_REPLACE_INITIAL_BACKOFF_SECONDS * 2),
+                ],
+                sleep.call_args_list,
+            )
+            self.assertEqual([], list(target.parent.glob(".remaining.txt.*.tmp")))
+
+    def test_permanent_access_denied_stops_after_bounded_attempts(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            target = Path(temporary) / "remaining.txt"
+
+            def always_denied(_source, _destination):
+                error = OSError("persistent WinError 5")
+                error.winerror = 5
+                raise error
+
+            with mock.patch.object(
+                runner.os, "replace", side_effect=always_denied
+            ) as replace, mock.patch.object(runner.time, "sleep") as sleep:
+                with self.assertRaises(OSError) as raised:
+                    runner.atomic_write_text(target, "case.a\n")
+
+            self.assertEqual(5, raised.exception.winerror)
+            self.assertEqual(runner.ATOMIC_REPLACE_ATTEMPTS, replace.call_count)
+            self.assertEqual(runner.ATOMIC_REPLACE_ATTEMPTS - 1, sleep.call_count)
+            self.assertFalse(target.exists())
+            self.assertEqual([], list(target.parent.glob(".remaining.txt.*.tmp")))
+
+    def test_non_access_error_is_not_retried(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            target = Path(temporary) / "remaining.txt"
+            error = OSError(28, "disk full")
+            with mock.patch.object(
+                runner.os, "replace", side_effect=error
+            ) as replace, mock.patch.object(runner.time, "sleep") as sleep:
+                with self.assertRaises(OSError):
+                    runner.atomic_write_text(target, "case.a\n")
+
+            self.assertEqual(1, replace.call_count)
+            sleep.assert_not_called()
+
+
+class QpaParsingTests(unittest.TestCase):
+    def test_terminate_is_a_completed_result(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            path = Path(temporary) / "chunk0000.qpa"
+            path.write_text(
+                "#beginTestCaseResult KHR-GL30.a\n"
+                "#terminateTestCaseResult Crash\n"
+                "#beginTestCaseResult KHR-GL30.b\n",
+                encoding="utf-8",
+            )
+            progress = runner.scan_qpa(path)
+            self.assertEqual(["KHR-GL30.a"], progress.recorded)
+            self.assertEqual("KHR-GL30.b", progress.in_flight)
+            self.assertEqual(2, progress.begin_count)
+
+    def test_end_without_result_is_not_accounted(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            path = Path(temporary) / "chunk0000.qpa"
+            path.write_text(
+                "#beginTestCaseResult KHR-GL46.incomplete\n"
+                "#endTestCaseResult\n"
+                + qpa_closed("KHR-GL46.complete"),
+                encoding="utf-8",
+            )
+            progress = runner.scan_qpa(path)
+            self.assertEqual(["KHR-GL46.complete"], progress.recorded)
+            self.assertIsNone(progress.in_flight)
+
+    def test_result_written_before_truncated_eof_is_recovered(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            path = Path(temporary) / "chunk0000.qpa"
+            path.write_text(
+                "#beginTestCaseResult KHR-GL46.complete\n"
+                '<Result StatusCode="Pass">ok</Result>\n',
+                encoding="utf-8",
+            )
+            progress = runner.scan_qpa(path)
+            self.assertEqual(["KHR-GL46.complete"], progress.recorded)
+            self.assertIsNone(progress.in_flight)
+
+
+class RunIdentityTests(unittest.TestCase):
+    def test_non_object_run_state_is_a_controlled_error(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            outdir = Path(temporary)
+            (outdir / "run_state.json").write_text("null\n", encoding="utf-8")
+            with self.assertRaises(runner.RunnerError):
+                runner.check_run_identity(outdir, "DirectVulkan", ["case.a"])
+
+    def test_controller_identity_prevents_mixed_invocations(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            outdir = Path(temporary)
+            runner.check_run_identity(
+                outdir, "DirectVulkan", ["case.a"], invocation_identity="identity-a"
+            )
+            runner.check_run_identity(
+                outdir, "DirectVulkan", ["case.a"], invocation_identity="identity-a"
+            )
+            with self.assertRaises(runner.RunnerError):
+                runner.check_run_identity(
+                    outdir, "DirectVulkan", ["case.a"], invocation_identity="identity-b"
+                )
+
+    def test_controller_identity_cannot_be_downgraded_by_omission(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            outdir = Path(temporary)
+            runner.check_run_identity(
+                outdir, "DirectVulkan", ["case.a"], invocation_identity="identity-a"
+            )
+            with self.assertRaises(runner.RunnerError):
+                runner.check_run_identity(outdir, "DirectVulkan", ["case.a"])
+
+    def test_legacy_artifacts_require_explicit_adoption(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            outdir = Path(temporary)
+            (outdir / "chunk0000.qpa").write_text(
+                qpa_closed("case.a"), encoding="utf-8"
+            )
+            with self.assertRaises(runner.RunnerError):
+                runner.check_run_identity(
+                    outdir, "DirectVulkan", ["case.a"], invocation_identity="identity-a"
+                )
+            self.assertFalse((outdir / "run_state.json").exists())
+
+            runner.check_run_identity(
+                outdir,
+                "DirectVulkan",
+                ["case.a"],
+                invocation_identity="identity-a",
+                adopt_legacy=True,
+            )
+            state = runner.json.loads(
+                (outdir / "run_state.json").read_text(encoding="utf-8")
+            )
+            self.assertTrue(state["adopted_legacy"])
+
+    def test_foreign_nested_qpa_and_skipped_sidecar_are_legacy_evidence(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            outdir = Path(temporary)
+            nested = outdir / "old"
+            nested.mkdir()
+            qpa = nested / "legacy.qpa"
+            qpa.write_text(qpa_closed("case.a"), encoding="utf-8")
+            skipped = outdir / "skipped.txt"
+            skipped.write_text("case.b\n", encoding="utf-8")
+
+            self.assertEqual(
+                {qpa, skipped}, set(runner.recovery_artifacts(outdir))
+            )
+            with self.assertRaises(runner.RunnerError):
+                runner.check_run_identity(
+                    outdir, "DirectVulkan", ["case.a", "case.b"]
+                )
+            recorded, crashed, hung = runner.recover_results(
+                outdir, {"case.a", "case.b"}
+            )
+            self.assertEqual({"case.a"}, recorded)
+            self.assertEqual(set(), crashed)
+            self.assertEqual(set(), hung)
+
+    def test_non_object_or_non_string_meta_classification_is_ignored(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            outdir = Path(temporary)
+            (outdir / "chunk0000.meta.json").write_text("null\n", encoding="utf-8")
+            (outdir / "chunk0001.meta.json").write_text("[]\n", encoding="utf-8")
+            (outdir / "chunk0002.meta.json").write_text(
+                '{"classified_case": [], "classification": "Crash"}\n', encoding="utf-8"
+            )
+            self.assertEqual(
+                (set(), set()), runner.load_meta_classifications(outdir, {"case.a"})
+            )
+
+
+class RunnerRecoveryTests(unittest.TestCase):
+    def run_args(self, root: Path, caselist: Path, outdir: Path, *extra: str):
+        return [
+            "--exe",
+            sys.executable,
+            "--workdir",
+            str(root),
+            "--caselist",
+            str(caselist),
+            "--outdir",
+            str(outdir),
+            "--backend",
+            "DirectVulkan",
+            *extra,
+        ]
+
+    def test_crash_tail_is_quarantined_and_next_chunk_resumes(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            caselist = root / "cases.txt"
+            outdir = root / "results"
+            caselist.write_text("KHR-GL30.a\nKHR-GL30.b\nKHR-GL30.c\n", encoding="utf-8")
+            seen_remaining = []
+
+            def fake_run(command, workdir, environment, qpa_path, stdout_path, stderr_path, *timeouts):
+                del workdir, timeouts
+                seen_remaining.append(
+                    command_path(command, "--deqp-caselist-file")
+                    .read_text(encoding="utf-8")
+                    .splitlines()
+                )
+                stdout_path.write_text("fake stdout\n", encoding="utf-8")
+                stderr_path.write_text("fake stderr\n", encoding="utf-8")
+                self.assertEqual("DirectVulkan", environment["MOBILEGL_BACKEND_TYPE"])
+                if len(seen_remaining) == 1:
+                    qpa_path.write_text(
+                        qpa_closed("KHR-GL30.a") + "#beginTestCaseResult KHR-GL30.b\n",
+                        encoding="utf-8",
+                    )
+                    return runner.ProcessOutcome(0xC0000005, 0.1)
+                qpa_path.write_text(qpa_closed("KHR-GL30.c"), encoding="utf-8")
+                return runner.ProcessOutcome(0, 0.1)
+
+            with mock.patch.object(runner, "run_process", side_effect=fake_run):
+                result = runner.main(self.run_args(root, caselist, outdir))
+
+            self.assertEqual(0, result)
+            self.assertEqual(
+                [
+                    ["KHR-GL30.a", "KHR-GL30.b", "KHR-GL30.c"],
+                    ["KHR-GL30.c"],
+                ],
+                seen_remaining,
+            )
+            self.assertEqual("KHR-GL30.b\n", (outdir / "crashed.txt").read_text(encoding="utf-8"))
+            self.assertEqual("", (outdir / "hung.txt").read_text(encoding="utf-8"))
+            self.assertEqual("", (outdir / "unrun.txt").read_text(encoding="utf-8"))
+
+    def test_existing_qpa_and_sidecar_are_recovered(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            caselist = root / "cases.txt"
+            outdir = root / "results"
+            outdir.mkdir()
+            caselist.write_text("KHR-GL31.a\nKHR-GL31.b\nKHR-GL31.c\n", encoding="utf-8")
+            (outdir / "chunk0000.qpa").write_text(qpa_closed("KHR-GL31.a"), encoding="utf-8")
+            (outdir / "crashed.txt").write_text("KHR-GL31.b\n", encoding="utf-8")
+            seen_remaining = []
+
+            def fake_run(command, workdir, environment, qpa_path, stdout_path, stderr_path, *timeouts):
+                del workdir, environment, stdout_path, stderr_path, timeouts
+                seen_remaining.extend(
+                    command_path(command, "--deqp-caselist-file")
+                    .read_text(encoding="utf-8")
+                    .splitlines()
+                )
+                qpa_path.write_text(qpa_closed("KHR-GL31.c"), encoding="utf-8")
+                return runner.ProcessOutcome(0, 0.1)
+
+            with mock.patch.object(runner, "run_process", side_effect=fake_run):
+                result = runner.main(
+                    self.run_args(root, caselist, outdir, "--adopt-legacy")
+                )
+
+            self.assertEqual(0, result)
+            self.assertEqual(["KHR-GL31.c"], seen_remaining)
+            self.assertTrue((outdir / "chunk0001.qpa").is_file())
+
+    def test_repeated_no_output_aborts_without_false_case_blame(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            caselist = root / "cases.txt"
+            outdir = root / "results"
+            caselist.write_text("KHR-GL32.a\nKHR-GL32.b\n", encoding="utf-8")
+            seen_remaining = []
+
+            def fake_run(command, workdir, environment, qpa_path, stdout_path, stderr_path, *timeouts):
+                del workdir, environment, stdout_path, stderr_path, timeouts
+                seen_remaining.append(
+                    command_path(command, "--deqp-caselist-file")
+                    .read_text(encoding="utf-8")
+                    .splitlines()
+                )
+                qpa_path.write_text("#sessionInfo releaseName fake\n", encoding="utf-8")
+                return runner.ProcessOutcome(
+                    1, 0.1, timed_out=True, timeout_reason="qpa-idle"
+                )
+
+            with mock.patch.object(runner, "run_process", side_effect=fake_run):
+                result = runner.main(
+                    self.run_args(root, caselist, outdir, "--max-empty-streak", "2")
+                )
+
+            self.assertEqual(4, result)
+            self.assertEqual(
+                [["KHR-GL32.a", "KHR-GL32.b"], ["KHR-GL32.a", "KHR-GL32.b"]],
+                seen_remaining,
+            )
+            self.assertEqual("", (outdir / "crashed.txt").read_text(encoding="utf-8"))
+            self.assertEqual("", (outdir / "hung.txt").read_text(encoding="utf-8"))
+            self.assertEqual(
+                "KHR-GL32.a\nKHR-GL32.b\n",
+                (outdir / "unrun.txt").read_text(encoding="utf-8"),
+            )
+
+
+class ProcessTimeoutTests(unittest.TestCase):
+    def test_qpa_activity_prevents_idle_timeout(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            qpa = root / "active.qpa"
+            helper = (
+                "import pathlib,sys,time\n"
+                "path=pathlib.Path(sys.argv[1])\n"
+                "for size in range(1, 9):\n"
+                " path.write_text('x' * size, encoding='utf-8')\n"
+                " time.sleep(0.08)\n"
+            )
+            outcome = runner.run_process(
+                [sys.executable, "-c", helper, str(qpa)],
+                root,
+                dict(runner.os.environ),
+                qpa,
+                root / "stdout.log",
+                root / "stderr.log",
+                idle_timeout=0.2,
+                max_round_seconds=0,
+                poll_seconds=0.03,
+            )
+            self.assertFalse(outcome.timed_out)
+            self.assertEqual(0, outcome.returncode)
+
+    def test_idle_timeout_really_stops_process(self):
+        with tempfile.TemporaryDirectory() as temporary:
+            root = Path(temporary)
+            started = time.monotonic()
+            outcome = runner.run_process(
+                [sys.executable, "-c", "import time; time.sleep(30)"],
+                root,
+                dict(runner.os.environ),
+                root / "never-created.qpa",
+                root / "stdout.log",
+                root / "stderr.log",
+                idle_timeout=0.2,
+                max_round_seconds=0,
+                poll_seconds=0.05,
+            )
+            elapsed = time.monotonic() - started
+            self.assertTrue(outcome.timed_out)
+            self.assertEqual("qpa-idle", outcome.timeout_reason)
+            self.assertLess(elapsed, 10)
+
+
+if __name__ == "__main__":
+    unittest.main()