Commit Graph
84 Commits
Author SHA1 Message Date
swung0x48 07669aacd4 [Fix] (ShaderTranspiler): fan #extension implications out to both new gates and bound the mid-line #version probe to its line 2026-08-27 03:35:58 -04:00
swung0x48 2ae848ca19 [Fix] (ShaderTranspiler): route restored ES extension macros through glslang's custom preamble and detect a mid-line repeated #version 2026-08-27 03:35:55 -04:00
swung0x48 acf86d1fb6 [Feat] (RenderState): implement glMinSampleShading and the GL_SAMPLE_SHADING enable on both backends 2026-08-27 03:35:55 -04:00
swung0x48 7eac33d17b [Feat, Fix, Test] (ShaderTranspiler, DirectGLES): carry the seven normalized image formats as their own codes in rgba16ui 2026-08-22 04:17:46 -04:00
swung0x48 d0f7fb99db [Feat, Test] (ShaderTranspiler, DirectGLES): carry rgb10_a2ui storage images in rgba16ui and split its packed upload 2026-08-22 03:21:48 -04:00
swung0x48 cbb616093b [Refactor, Test] (ShaderTranspiler, GLState): take what the relaxed parse destroys from glslang instead of scanning the source 2026-08-21 13:51:22 -04:00
swung0x48 194c2f189b [Merge] (ShaderTranspiler): land the macro-spelled storage-block binding repair 2026-08-21 12:16:52 -04:00
swung0x48 7de7cfc6eb [Fix, Test] (ShaderTranspiler): read a storage block's macro-spelled binding as declared, not as absent 2026-08-21 12:09:03 -04:00
swung0x48 77c05b151a [Fix, Test] (ShaderTranspiler): lower a single-implementation GLSL subroutine to a forwarding call 2026-08-21 11:45:49 -04:00
swung0x48 2a902ff58c [Fix, Test] (ShaderTranspiler): bound the whole loop nest a fragment-output index marks for unrolling 2026-08-21 07:04:19 -04:00
swung0x48 d9def5c1bb [Fix, Test] (GLImpl, GLState): give a storage block with no binding qualifier GL's default binding of zero 2026-08-21 05:13:58 -04:00
swung0x48 21a4c8aa95 [Fix, Test] (ShaderTranspiler, DirectGLES): widen the offset and gradients of a 1D sampler lookup for ESSL 2026-08-21 04:57:52 -04:00
swung0x48 668f3e90c9 [Fix] (DirectGLES, ShaderTranspiler): widen the formats SPIRV-Cross refuses to print even where GL_NV_image_formats exists 2026-08-21 04:13:01 -04:00
swung0x48 b164692387 [Test] (ShaderTranspiler): retarget the bake-decline pin at the formats no core carrier rescues 2026-08-21 03:48:59 -04:00
swung0x48 c129cdec2d [Fix, Test] (ShaderTranspiler, ProgramState): reject an out-of-range atomic-counter offset at compile 2026-08-21 00:10:53 -04:00
swung0x48 4154f2e941 [Fix, Test] (ShaderTranspiler, DirectGLES): widen a non-arrayed 1D storage image's atomic coordinate 2026-08-20 22:46:32 -04:00
swung0x48 de8e7a4606 [Fix, Test] (ShaderTranspiler): parse layout literals in every GLSL base and key array-of-arrays uniforms per element 2026-08-20 13:24:04 -04:00
swung0x48 e2923a239f [Fix, Test] (ShaderTranspiler): size a non-final unsized storage-block member so the members after it stop aliasing it 2026-08-20 13:05:50 -04:00
swung0x48 01fbe0b4b0 [Fix, Test] (GLState, ShaderTranspiler): reject a storage-block binding at or past GL_MAX_SHADER_STORAGE_BUFFER_BINDINGS 2026-08-20 12:45:56 -04:00
swung0x48 7769156cfc [Fix, Test] (ShaderTranspiler, WGL, TraceReplay): remove subgroup pack quirks and tune iterationRP 2026-08-17 01:55:17 -04:00
swung0x48 0ecfdff4e7 [Fix, Test] (MG_State, MG_Util, DirectVulkan, MG_Test): replay narrow-subgroup reductions correctly 2026-08-16 13:33:01 -04:00
swung0x48 14d3901d30 [Chore, Test] (MG_State, MG_Util, DirectGLES, DirectVulkan): make SPIR-V validation task-local 2026-08-15 22:58:28 -04:00
swung0x48 9d1b280375 [Fix, Test] (MG_Util, MG_Test): rename Photon-conflicting MSL identifiers 2026-08-15 09:40:13 -04:00
swung0x48 43398e33e8 [Fix, Test] (MG_Util, MG_IntegrationTest): review round - anchor a clone only behind a real definition, decline the atomic-image shape, and say what the software lanes cannot falsify 2026-08-12 19:12:39 -04:00
swung0x48 b7557d6615 [Feat, Fix, Test] (MG_Util, MG_Backend/DirectGLES, MG_IntegrationTest): bake the bound image format into the ESSL a format-less image declaration needs 2026-08-12 18:56:31 -04:00
swung0x48 44805bfa07 [Fix, Test] (MG_Util, MG_Backend/DirectVulkan, MG_IntegrationTest): adversarial review - the rewritten 1D-array image collided with a module's own 2D-array one and left an invalid duplicate type; pin the component order, keep the subject kinds in a truncated matrix 2026-08-12 16:49:59 -04:00
swung0x48 257fcbfd0b [Fix, Test] (MG_Util, MG_Backend/DirectVulkan, MG_IntegrationTest): review round - one module parse for the shaders with no 1D-array image, per-kind values the combined case can name, and the invariants two shared buffers rest on 2026-08-12 16:43:29 -04:00
swung0x48 2b46a3db96 [Fix, Test] (MG_Util, MG_Backend/DirectGLES, MG_Backend/DirectVulkan): the two image target kinds a compute dispatch could not read - 1D-array on ES, imageBuffer on Vulkan 2026-08-12 16:11:10 -04:00
swung0x48 a0bf4a83bc [Fix, Test] (MG_Util, MG_Backend/DirectGLES, MG_Test, MG_IntegrationTest): detect buffer-texture support, emit the directive the driver advertises, and name the capability when it is missing 2026-08-12 01:24:40 -04:00
swung0x48 8f3ce5f5b7 [Fix] (MG_Backend/DirectGLES, MG_State, Docs): CopyImageSubData held a registry reference across a re-entrant sync; retarget container rationales at the new erase contract 2026-08-12 00:11:54 -04:00
swung0x48 cef81df73f [Fix, Test] (ShaderTranspiler, DirectGLES): fold or lower dynamically indexed fragment outputs before ESSL emission - GLSL ES requires constant integral indices, so the OIT coefficient shader linked nothing on ANGLE and every translucent draw was a silent no-op 2026-08-11 23:39:03 -04:00
swung0x48 b219992ee3 [Feat, Perf, Test] (MG_Util, MG_Backend, MG_Test): test-lane SPIR-V validation latch; entry-local Private rewrite, in-chain rect lowering and uniform-Location strip 2026-08-10 13:00:20 -04:00
swung0x48 c3412ca394 [Fix, Test] (MG_Util): give inactive vertex inputs a Location - undecorated inputs are invalid SPIR-V and Adreno rejects the whole pipeline 2026-08-10 09:27:01 -04:00
BZLZHH c93e5fa409 [Refactor] (MG_State, MG_Util): join-by-construction link/compile artifacts (P1 stage 2)
Still fully synchronous - EnsureLinkJoined()/EnsureCompileJoined() are empty
inline no-ops (verified to fold away at every one of the ~1200 call sites;
this project builds without LTO) - but every read of link- or compile-produced
state now goes through a private accessor the compiler enforces, so when
stage 4 moves the bodies onto pool workers, 'which reads must join' is a
type-system fact instead of a 400-line audit.

- ProgramObject: the 31 fields ResetLinkArtifacts clears plus the 5 link
  outputs it forgot (infoLog, linkedFragData{Location,Index}, the geometry
  strip-capture pair) move into a nested LinkArtifacts behind Artifacts().
  ResetLinkArtifacts is now a worker-safe pure clear; the link-observable
  version bumps (backendState/link/uboContent) move to a GL-thread-only
  BumpLinkObservableVersions() called once from Link()'s prologue and from
  glProgramBinary's mandated failure - the link body never writes them, so
  a stage-4 worker cannot lose an invalidation against the draw path.
- ShaderObject: compile artifacts (TShader, preprocessed source, side-channel
  maps, status/log, consume-once flag) behind Compiled(); the P0b layer-1
  memo trio deliberately stays outside as the future non-joining
  COMPLETION_STATUS_KHR fast path.
- CompileEnv (new): a GL-thread snapshot of everything the compile pipeline
  used to read live from the backend mid-parse - compute limits (the
  GetIntegeri_v reach-back is gone from the worker path), advertised
  extensions, device quirks, TBuiltInResource inputs. Captured lazily per
  backend activation; the consume-once re-parse now runs against the same
  env as the original parse.
- The GL-thread prologue / worker-body boundary is marked in Link() where
  the stage sort ends; everything below is a pure function of the snapshot.

Public getter signatures unchanged - MG_Impl and both backends compile
untouched. Unit 476/476, Program suites 117/117, DirectGLES retrace 38/39 on
llvmpipe (the one failure is the known pre-existing non-CI iterationrp case;
the NVIDIA userspace driver was updated out from under the running kernel
module mid-session, so GLX there is down until a reboot).
2026-08-08 07:12:37 -04:00
BZLZHH 9152e88734 [Perf] (MG_State): dedupe shader compiles by source hash
Iris-style packs hand MobileGL the same source text repeatedly: probed across
three shaderpack traces, 28-32% of all glCompileShader work was redundant -
~9% same-object recompiles with byte-identical source, ~21% distinct shader
objects sharing identical source (the same common GLSL chunk glued into many
program stages). Two layers, both keyed by XXH64 + length with a full byte
compare on every hit (correctness never rides on the hash):

- Per-object: a successful (or failed) compile remembers its source hash;
  glShaderSource with byte-identical text keeps the compiled state and
  glCompileShader on unchanged source returns immediately. Deterministic
  (stage, source) pipeline makes the memo observationally identical to
  recompiling; the consume-once TakeShaderForLink re-parse path is untouched.

- Cross-object: a per-context bounded cache (ProgramState-owned, declared to
  outlive every shader object) shares the preprocessed source, both explicit
  side-channel maps, and the validation verdicts between objects with equal
  source; only the glslang parse stays per-object. Single-GL-thread today;
  flagged for a mutex when compiles go async (P1).

Interleaved A/B on the iterationrp trace (the recompile-heavy pack):
5.65s -> 5.46s median total replay, every round faster; BSL/complementary
stay flat (their duplicate sources are the small common shaders, so calls
drop but wall time is parse-bound on unique sources). Full DirectGLES
retrace, 445-test unit suite, and dedupe-semantics tests (no-op recompile,
invalidation on new source, failed-compile memo, cache bounds) all green.
2026-08-08 04:18:32 -04:00
BZLZHH 2406e2d219 [Perf, Fix] (MG_Util): preprocessing cleanups - dead scanners, quote-mask bug, one version inspection per compile
Three scoped changes to ShaderSourceProcessor, none altering any transform's
output (pinned by a byte-stability test across the legacy-shader anchor path):

- Delete BlankBlockComments and RemoveDefineForIdentifier - dead since their
  callers left; the former's newline-terminated quote handling moves into
  MaskCommentsAndQuotedText (below) together with its rationale comment.

- Fix MaskCommentsAndQuotedText treating a quote as running past end-of-line.
  GLSL has no multi-line literals, but a stray apostrophe in a directive or
  comment tail ("#pragma message can't") blanked the REST OF THE FILE for
  every masked consumer - the tokenizer, the version inspection, and the P0a
  explicit-location/binding extractors silently lost everything after it.

- Inspect the shader language once per PreprocessShaderSource run instead of
  up to five times: NormalizeVersionDirective now takes the already-computed
  ShaderLanguageInfo, and the two after-version injections share one
  AfterVersionAnchor instead of re-running a full masked sweep each
  (FindAfterVersionDirective -> InspectShaderLanguage) to find the same spot.

Compile-phase timings hold (BSL 1.848s, complementary-reimagined ~5.7s);
retraces and the 435-test unit suite unchanged.
2026-08-08 03:43:56 -04:00
BZLZHH b228f813c0 [Perf] (MG_Util): replace the builtin-shadowing string scans with one tokenize and a SPIR-V OpName pass
RenameBuiltinShadowingFunction probed the whole source ten times per compile
(5 names x mask + scan, each a full-text pass) and still had two blind spots:
a 5-name list and single-line-definition-only detection. On Complementary-scale
packs (4.5MB of sources) that was ~68% of the compile phase.

The rename is now split by FAILURE LAYER, both halves sharing one name table
header so they cannot drift:

- A SPIR-V OpName pass in SanitizeAndOptimizeBinary covers the full ESSL 3.20
  builtin set (~146 names). Renaming a function id is safe by construction:
  builtin calls are GLSL.std.450 instructions and can never resolve to a user
  OpFunction, overloads are distinct ids (a helper overload delegating to the
  real builtin keeps working), dead preprocessor branches never reach SPIR-V,
  and macro-expanded definitions are covered. ESSL 3.x is the only consumer
  that forbids the redefinitions, and this pass runs before its transpile.

- A lexical pass covers only the 5 names whose exact-signature redefinitions
  glslang's relaxed parse rejects outright (never producing SPIR-V for the
  backstop): the historical fma/max3/min3/round/tanh. One TokenizeCode pass;
  definition detection requires brace depth 0, a type-identifier previous
  token that is neither a statement keyword nor a directive tail, and skips
  files whose token-level braces do not balance (preprocessor-asymmetric
  arms) - over-detection is unrecoverable, so every ambiguity falls through
  to the backstop.

Measured on the compile phase (prefix-diff, 3-run medians, Espryt/NVIDIA):
complementary-reimagined 20.0s -> 5.5s, BSL 2.14s -> 1.85s. bliss (the pack
that ships from-scratch fma/tanh helpers) stays at SSIM 0.999962.

Tests: end-to-end ESSL assertions for the multiline-definition and
new-overload shapes, the three adversarial-review reproductions (statement-
keyword call under asymmetric braces, dead-#if compat shim, overload
delegating to the shadowed builtin), and a source-level assertion pinning
the lexical half specifically.
2026-08-08 03:11:38 -04:00
BZLZHH 86c00bdf18 [Test] (MG_Test): catch the unit tests up with three deliberate behaviour changes
ctest -L unit had been failing 13 of its 418 cases, all of them tests left asserting
what the code did before a commit that changed it on purpose:

- "restore target GL version to 3.3" put the advertised target back after the
  experimental 4.6 run, but the two Voxy sanity tests still demanded 4.6. The
  extensions they really care about are all still advertised, so assert 3.3 and drop
  the now-meaningless AtExperimentalCTSVersion from their names.
- "support rectangle textures where the emulation is exact" made every desktop-only
  target supported - rectangle included, stored as a plain 2D - while the texture
  test still expected rectangle to be rejected.
- "keep declared modern GLSL versions strict" changed two things at once: a
  normalized legacy directive now carries a marker on its line, so the ten tests
  matching "#version 330 core\n" whole no longer match; and a version the
  application declared itself is no longer raised to 460, so the sources declaring
  330/400 keep their own number and only MobileGL's own normalization is retargeted.

Test expectations follow, rather than the implementation being bent back: each of
the three changes is the intended behaviour and is argued for where it was made. The
retry test now drives the 460 escalation from a legacy "#version 130" source, which
is the only thing that is still rescued, and gained a case pinning the other half of
that contract - an application-declared "#version 330" stays at 330.

418/418 unit tests pass.
2026-08-02 09:04:16 -04:00
swung0x48 3049c4b82b [Fix] (ShaderTranspiler): stop blanking block comments in the source handed to glslang
- BlankBlockComments replaced comment chars with spaces but preserved interior newlines, so a block comment spanning a newline inside a #define truncated the macro body (VALUE became empty)
- glslang has a conformant preprocessor and collapses a block comment to one space across newlines, so the delivered source now keeps comments intact and lets glslang handle them
- Fixes KHR-GL3x.shaders.preprocessor multiline_comment_define / redefine_object_multiline_comment / function_redefinition_3 (6 cases, both devices)
- FilterUnsupportedGpuShaderInt64 relied on the blanking to skip commented-out #extension lines; it now masks comments locally (MaskCommentsAndQuotedText) like the sibling passes, collecting edits and applying them back-to-front
- conditional_inclusion.basic_2 (defined() via macro expansion) stays failing by design: glslang rejects it as UB and working around it would mean re-running preprocessing MobileGL defers to glslang
2026-07-21 05:27:01 -04:00
swung0x48 79feeffd25 [Feat] (ShaderTranspiler, DirectGLES): emulate noperspective on GLES devices lacking GL_NV_shader_noperspective_interpolation - EmulateNoPerspectivePass pre-multiplies each NoPerspective output by gl_Position.w in the vertex stage and recovers each input via gl_FragCoord.w in the fragment stage (exact screen-linear L = P(a*w)*gl_FragCoord.w, handling whole-variable and component/access-chain reads, scalar and vector varyings), forces highp on emulated varyings, and strips what it cannot emulate; replaces the smooth-strip fallback so no NV extension is ever required. Restricts the vertex pre-multiply to the entry function so a non-inlined helper cannot double-scale 2026-07-20 23:48:48 -04:00
swung0x48 bce9c48c8e [Feat] (ShaderTranspiler, DirectGLES): support noperspective conformantly instead of stripping it - let the qualifier reach glslang as the core SPIR-V NoPerspective decoration (native on DirectVulkan; SPIRV-Cross emits ESSL noperspective + GL_NV_shader_noperspective_interpolation on DirectGLES), and for GLES devices lacking that extension add StripNoPerspectivePass to drop the decoration and fall back to smooth; the old naked substring erase discarded the interpolation shader packs need and mangled identifiers containing the word 2026-07-20 22:59:43 -04:00
swung0x48 b6a7807a3a [Fix] (MG_Util/ShaderTranspiler): reject malformed #version directives instead of legalizing them - an unrecognized version number (329/331), a bad profile keyword, a float or trailing token used to be rewritten to "#version 330 core" (or rescued to 460 by the retry); now InspectShaderLanguage marks such directives invalid so NormalizeVersionDirective and RetargetLegacyVersionDirectiveTo460 leave them for glslang to reject, while every valid version still normalizes as before 2026-07-20 22:03:36 -04:00
swung0x48 48ba622387 [Fix] (MG_Util/ShaderTranspiler): keep #line directives instead of deleting them, dropping only the GLSL-illegal quoted filename and the ones that precede #version, so __LINE__ and compiler diagnostics follow the application's own numbering 2026-07-20 21:06:38 -04:00
swung0x48 05260d1262 [Fix] (MG_Util/ShaderTranspiler): blank block comments lexically instead of erasing them - a '//*** banner ***' line opened a comment the old scanner never closed, so it deleted the rest of the shader, and a commented-out builtin definition renamed every genuine call to a name nothing defines 2026-07-20 21:06:37 -04:00
swung0x48 bf7b5755cc [Refactor] (ShaderTranspiler, MG_Backend, MG_Util): gate the subgroup prefix-scan rewrite behind a generic device-quirk registry with GPU vendor detection and MOBILEGL_QUIRK_SUBGROUP_PREFIX_SCAN override, warn on template mismatch, and block ARB/NV subgroup spellings 2026-07-20 02:36:26 -04:00
swung0x48 68e13705c8 [Fix] (MG_Backend/DirectVulkan, ShaderTranspiler, MG_Test, TraceReplay): make iterationRP retrace pass on 64-lane Vulkan devices with subgroup-width emulation and format-aware readback 2026-07-20 02:36:23 -04:00
swung0x48 e5388c0e7e [Fix] (MG_Backend, MG_Impl, ShaderTranspiler, MG_Test): support iterationRP custom images and storage format reinterpretation 2026-07-20 02:35:30 -04:00
swung0x48 e2f873c95c [Fix] (MG_Util/ShaderTranspiler): retry a legacy shader at 460 when it fails to parse as normalized 330 core, so sources using 420-era syntax without the matching #extension line keep compiling as they did on real drivers 2026-07-16 22:33:30 -04:00
swung0x48 b8db509581 [Fix] (MG_Util/ShaderTranspiler): normalize legacy desktop shaders to GLSL 330 2026-07-16 21:39:39 -04:00
swung0x48 254cf1dc21 [Fix] (MG_Util/ShaderTranspiler, MG_State, MG_Impl/GLImpl, MG_Backend/DirectGLES): GL CTS uniform_block - coerce packed/shared block layouts to std140 at source preprocess (glslang rejects them when targeting SPIR-V; std140 is the only UBO layout the pipeline emits), GL-style block reflection (array "[0]" names, per-element struct-array expansion, unused members and declared-but-unread blocks stay active), vec4-padded GL_UNIFORM_BLOCK_DATA_SIZE, std140 array strides for struct-nested arrays (glslang reflects tight strides there), arrayed-block instances share the first instance member set, glDeleteShader-flagged names stay usable while attached, and backend ESSL emits against highp default precision so relaxed block members match across stages (KHR-GL33.shaders.uniform_block on llvmpipe: 659 Fail -> 828/828 Pass) 2026-07-16 12:09:11 -04:00