swung0x48/MobileGL
1
0
Fork 0
mirror of https://github.com/MobileGL-Dev/MobileGL synced 2026-09-13 14:48:32 +09:00
Code Issues Packages Projects Releases Wiki Activity
2,244 Commits 28 Branches 0 Tags
562076f657b355f876f8bf036a536a4abee28ee0
Commit Graph
479 Commits
Author SHA1 Message Date
Claude 562076f657 [Fix] (ShaderTranspiler, DirectVulkan, MG_IntegrationTest): patch both of iterationRP's under-declared subgroup scratch arrays 
The previous commit's fingerprint was pinned to one array's incidental
dimensions - workgroup exactly 32x16x1, element exactly vec2, length
exactly 32 - which is the auto-exposure reduction and nothing else. The
pack ships the same idiom twice:

  - auto-exposure:  32x16 (512 invocations), shared vec2 prefixSumCache[32]
  - RTW warp:       1024 invocations,        shared float prefixSumCache[64]

so the warp kept writing 128 subgroups into 64 entries on an 8-lane
device and the retrace stayed bit-identically wrong (ssim 0.027902).

Key the fingerprint on the pack's idiom instead of one array's shape: a
workgroup array of 32-bit floats indexed by gl_SubgroupID, fed by a
subgroup scan, whose declared length is below ceil(invocations / native
width). Three properties keep that a targeted repair rather than a
general array resizer:

  - the index must BE gl_SubgroupID (through OpCopyObject, a signedness
    OpBitcast, or a spill whose every store is that id), so an index
    masked or clamped into range is left alone;
  - the >= 16-lane early-out is retained, so every module on the devices
    the pack was written for passes through byte-identical;
  - growth is certified against maxComputeSharedMemorySize using a
    natural-alignment layout model, and declined outright when a
    declaration cannot be sized, so a patched module can never fail
    pipeline creation where the original would not have.

Verified against the shaders the CI trace actually contains: of the 14
compute modules in the fixture exactly these two change, the other
twelve are byte-identical, and all fourteen pass spirv-val. The
integration scenario grows a second case for the 1024-invocation shape;
both abort with heap corruption when the patch is disabled.

Claude-Session: https://claude.ai/code/session_01EXSURVxwp8VVrWrPEQeLCm
 2026-08-19 17:00:33 +00:00
swung0x48 d8576a2ed3 [Fix] (DirectVulkan, ShaderTranspiler, MG_IntegrationTest, SelfTest, TraceReplay): use native subgroups and patch iterationRP's under-declared scratch 
iterationRP's Program 203 declares shared vec2 prefixSumCache[32] for a
512-invocation workgroup indexed by gl_SubgroupID; any device narrower
than 16 lanes partitions into more than 32 subgroups and the pack writes
shared memory out of bounds (heap corruption on lavapipe's CPU
rasterizer, ssim 0.028 on the CI retrace). Fix it where the fault lies -
in the fixture - and keep the GL contract sound everywhere else:

- FixIterationRPSubgroupScratchPass: fingerprint-gated SPIR-V pass that
  grows exactly that array to ceil(invocations/width) entries on sub-16-lane devices; every other module passes through byte-identical.
- DeriveNumSubgroupsPass stays default-on for the Adreno topology bug
  and is made spec-sound: pipelines request REQUIRE_FULL_SUBGROUPS
  whenever the workgroup shape makes the flag legal (computeFullSubgroups
  enabled, local_size_x a multiple of the native width, subgroup count
  within maxComputeWorkgroupSubgroups).
- EmulateSubgroupsPass: 32-lane virtual-subgroup lowering kept in-tree
  as a last resort, enabled only by MOBILEGL_MAGMA_EMULATE_SUBGROUP=1 on
  devices with no native subgroup support; fails closed on extended
  subgroup instructions and on modules whose added scratch would exceed
  maxComputeSharedMemorySize.
- IterationRPFirstReductionScenario skips gracefully outside the pack's
  16..256-lane source domain; the new IterationRPScratchFixScenario runs
  the fixture-shaped reduction on any width and asserts the exact
  width-independent total. DriverPost keeps reporting FAIL on
  out-of-domain devices.
- Program203 -> IterationRP rename throughout; the per-trace
  num_subgroups_quirk plumbing is removed from the trace replayer, JNI
  chain, and CI workflows.
 2026-08-19 09:48:11 -04:00
swung0x48 2b6c2b561c [Fix, Test] (DirectVulkan, ShaderTranspiler, TraceReplay): derive NumSubgroups behind opt-in quirk 2026-08-18 22:31:57 -04:00
swung0x48andClaude Fable 5 12c94111b5 [Fix, Test] (DirectVulkan, SelfTest, MG_IntegrationTest, TraceReplay): snapshot sampler/image feedback and add Program 203 diagnostics 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
 2026-08-18 03:04:09 -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 d4766513e4 [Fix, Test] (DirectVulkan): replay Photon descriptor pressure correctly 2026-08-15 22:35:39 -04:00
swung0x48 8acd885594 [Fix, Test] (DirectVulkan): preserve viewport-index program metadata 2026-08-15 09:40:13 -04:00
swung0x48 10ff5e2b18 [Fix, Test] (DirectGLES, DirectVulkan): advertise indirect draw capabilities accurately 
- advertise GL_ARB_draw_indirect when supported
- gate GL_ARB_base_instance on complete non-zero firstInstance semantics
- synchronize Driver POST reporting
- add capability and extension-advertisement regression tests
 2026-08-15 06:30:44 -04:00
swung0x48 0deff52a1b [Fix, Test] (DirectVulkan, trace-replay): replay quarter-turn surfaces correctly 2026-08-13 06:45:28 -04:00
swung0x48 42ad62b54c [Fix, Test] (MG_Util, MG_IntegrationTest): advertise the GL 4.3 VIEWPORT_BOUNDS_RANGE floor on a GLES driver that has no such query, instead of a range admitting no origin 2026-08-13 04:44:23 -04:00
swung0x48 f41403e227 [Feat, Test] (MG_Backend/DirectVulkan, MG_Test, MG_IntegrationTest): rasterize the viewport gl_ViewportIndex selects, instead of collapsing all sixteen onto viewport 0 2026-08-13 04:44:23 -04:00
swung0x48 5fbb17f6b9 [Feat, Fix, Test] (MG_State, MG_Impl, MG_Backend, MG_Test): give ARB_viewport_array real 16-element indexed state instead of eight stubs and a viewport-0 echo 2026-08-13 04:44:23 -04:00
swung0x48 b8233f9c4e [Fix, Test] (MG_Backend/DirectVulkan, MG_IntegrationTest): the barrier before every attachment transfer only ever moved layer 0, so a copy off a non-zero layer read a layout nothing had transitioned 2026-08-13 04:01:20 -04:00
swung0x48 b62d1f2078 [Fix, Test] (MG_Backend/DirectVulkan, MG_IntegrationTest): map a layered CopyImageSubData onto the axis each endpoint keeps its slices on, instead of copying slice 0 and calling it done 2026-08-13 02:49:57 -04:00
swung0x48 6a80a82dd3 [Fix, Test] (MG_Util, MG_Backend/DirectGLES, MG_Backend/DirectVulkan, MG_Test): a packed texel whose client type already spells its storage word must cross glGetTexImage unencoded, or the RGB9_E5 shared exponent gets canonicalized 2026-08-13 02:41:37 -04:00
swung0x48 f37b511fca [Fix, Test] (MG_Impl/GLImpl, MG_Backend/DirectVulkan, MG_Backend/DirectGLES, MG_Test, MG_IntegrationTest): a glCopyImageSubData to a mipmap level the texture never had is INVALID_VALUE, not a subresource handed to the driver 2026-08-13 02:28:55 -04:00
Swung0x48 dd2a62228f [Fix, Feat, Test] (MG_Util, MG_Backend, MG_Impl, MG_State): correct the log severity ordering and unwind the diagnostics it silenced 2026-08-13 02:02:49 -04:00
swung0x48 96646df12e [Fix] (MG_Backend/DirectVulkan): review round - failure diagnostics take MGLOG_E, the once-per-patch-size success note takes MGLOG_D, and the per-draw refusal stays latched 2026-08-13 01:34:57 -04:00
swung0x48 f20b20e643 [Fix, Feat, Test] (MG_Backend/DirectVulkan, MG_Test): synthesize the pass-through tessellation control stage GL gives an evaluation-only program, and refuse the half-tessellated pipeline Mali dereferences null inside 2026-08-13 01:28:35 -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 6b6623ae72 [Fix, Test] (MG_Backend, MG_Impl, MG_State): advertise the uniform-block and stencil-texturing strings a 4.0 context hides behind, forward DEPTH_STENCIL_TEXTURE_MODE to both backends, and stop a reserved transform feedback name passing for an object 2026-08-12 15:12:01 -04:00
swung0x48 1b05a84928 [Fix, Test] (MG_State, MG_Backend, MG_Impl, MG_Test): review wave - hand the mapping back instead of renewing it, retire an immutable ES store on respecify, and tag glTexStorage2D levels too 2026-08-12 10:32:39 -04:00
swung0x48 0e31c1481b [Fix, Test] (MG_State, MG_Backend/DirectVulkan, MG_Backend/DirectGLES, MG_IntegrationTest): a respecified capture buffer left the transform feedback writing one store and the readback reading another 2026-08-12 08:42:45 -04:00
swung0x48 2fced2241b [Fix, Test] (MG_Util, MG_State, MG_Backend): the GL 4.3 vertex binding model - array vertex inputs, zero binding strides, instance divisors, and formats ES refuses 2026-08-12 05:36:00 -04:00
swung0x48 7311251f30 [Fix] (MG_Util, MG_Backend/DirectVulkan): GL reads gl_BaseVertex as zero on a non-indexed draw where Vulkan's builtin hands over firstVertex 2026-08-12 04:18:07 -04:00
swung0x48 97facf777b [Fix] (MG_Backend/DirectVulkan): act on a failed binding remap, bound the storage-block array count, and correct the decline rationale review found describing the reverted mechanism 2026-08-12 01:09:56 -04:00
swung0x48 5cfbb716c0 Merge remote-tracking branch 'origin/dev' into feat/sampler-array-descriptors 2026-08-12 00:58:20 -04:00
swung0x48 068786e812 [Fix, Test] (MG_Backend/DirectVulkan, MG_IntegrationTest): a declined descriptor must refuse the draw, not just leave the layout - a multi-dimensional sampler array otherwise faulted in the shader 2026-08-12 00:45:36 -04:00
swung0x48 d2a36d65a3 [Fix, Test] (MG_Backend/DirectVulkan, MG_IntegrationTest): a sampler uniform array is one binding with many descriptors too - Magma wrote only element zero, and declines the multi-dimensional shape loudly 2026-08-12 00:31:37 -04:00
swung0x48 574634adfa [Fix, Test] (MG_Backend/DirectVulkan, MG_IntegrationTest): let the binding remap accept the descriptor arrays that now have per-element paths, and reserve image-info scratch for them 2026-08-12 00:14:18 -04:00
swung0x48 e71d715e1a [Fix, Test] (MG_Backend/DirectVulkan, MG_Util, MG_Test): VkProgramObject move dropped stageSpirvDigests - a robin-hood swap would pair one program's modules with another's digests; sanity tests assert key identity, not iteration counts 2026-08-12 00:11:56 -04:00
swung0x48 7b946fd527 [Fix] (MG_Backend/DirectVulkan, MG_Util): RenderPassEntry needs move assignment - robin-hood probing swaps entries, so mapped types must be swappable 2026-08-12 00:11:55 -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 f6849fc0b3 [Fix, Test] (MG_Backend/DirectVulkan, MG_IntegrationTest): an image uniform array is one binding with many descriptors, not one - Magma wrote only element zero and left the rest undefined 2026-08-11 23:49:31 -04:00
swung0x48 43bcd03dca [Fix] (MG_Backend/DirectVulkan, MG_Impl): a malformed graphics pipeline shape skips the draw instead of faulting inside the driver 2026-08-11 20:10:18 -04:00
swung0x48 2ce0595fab [Fix, Test] (MG_State, MG_Impl, MG_Backend): a program pipeline's compute stage is dispatched on its own, and the graphics composite draws its stage programs' uniform values 2026-08-11 20:10:17 -04:00
swung0x48 757b31592d [Fix, Test] (MG_Backend/DirectVulkan): a multisample resolve that must also change orientation resolves through a pooled scratch image, then blits 2026-08-11 10:49:58 -04:00
swung0x48 dbae4eda10 [Fix, Test] (MG_Backend/DirectVulkan): depth blits to or from the default framebuffer convert their rect out of GL's bottom-origin space, like the colour blit does 2026-08-11 10:24:59 -04:00
swung0x48 994ae372f8 [Fix, Test] (MG_Backend/DirectVulkan): an SSBO block instance array is one binding of N descriptors, so bind every element from its own GL binding point 2026-08-11 10:11:23 -04:00
swung0x48 b1fdffd767 [Fix, Test] (MG_Backend/DirectVulkan): read the default framebuffer's depth and stencil back instead of leaving the caller's buffer untouched 2026-08-11 10:11:22 -04:00
swung0x48 18c17ae5ca [Fix, Test] (MG_Backend/DirectVulkan): a blit into the default framebuffer must execute the clear parked before it, not leave it for the readback 2026-08-11 10:08:25 -04:00
swung0x48 5d47698349 [Fix, Test] (MG_Backend/DirectVulkan): materialize the default framebuffer's pending clear before a readback, alpha included 2026-08-11 09:09:31 -04:00
swung0x48 964a7fcc92 [Fix, Test] (MG_State, MG_Backend/DirectVulkan): resolve transform-feedback captures that name a member of an output interface block 2026-08-11 08:39:39 -04:00
swung0x48 5a7bd9942d [Fix] (MG_Backend/DirectVulkan): print VkShaderModule as a 64-bit value - the const void* cast is ill-formed on 32-bit ABIs where the handle is a plain uint64_t 2026-08-11 08:19:31 -04:00
swung0x48 21a43bf6a4 [Fix, Test] (MG_Backend/DirectVulkan): re-land the gl_FragCoord default-framebuffer origin fix - the suspected slowdown was a mismeasurement, paired timings are within 1% 2026-08-11 07:54:24 -04:00
swung0x48 5b6dec2d81 [Revert] (MG_Backend/DirectVulkan): back out the gl_FragCoord default-framebuffer origin fix - correct, but it costs DirectVulkan a large order-dependent slowdown that is not yet root-caused 2026-08-11 06:30:30 -04:00
swung0x48 543c29bf86 [Fix, Test] (MG_Backend/DirectVulkan): gl_FragCoord on the default framebuffer reports GL's window origin - the stored row is not the window row once the viewport rect is converted 2026-08-11 05:25:32 -04:00
swung0x48 a6c362c6ce [Fix, Test] (MG_Backend/DirectVulkan): convert every default-framebuffer rectangle between GL and display Y origins - viewport, scissor, ReadPixels offset, rect-capable readback remap, blit source 2026-08-11 02:51:22 -04:00