fix(shader): record BDA faults in hull programs instead of refusing them by Celegans12 · Pull Request #4222 · boykopovar/AnyPS5

Celegans12 · 2026-10-12T01:45:14Z

What

Stacked on #4201; please review the last commit only.

A hull (tessellation-control) program with a BDA read fails to compile with "BDA fault termination requires a barrier-safe tessellation-control execution protocol". The refusal dates from the first BDA commit and is now redundant: BdaInvocationsMayStop already switches any program that contains s_barrier to the non-stopping fault protocol, and a hull's only barriers come from guest s_barrier (EmitBarrier); the LDS-lock barrier runs before any guest code and wave-LDS barriers are compute-only.

The refusal is removed. A hull with s_barrier keeps every invocation to its barriers; a hull without one may stop a faulting invocation, since no barrier can be stranded.
BdaContracts asserted the refusal and now accepts a hull with s_barrier.

PPSA26344's terrain hull 0x80002fc700 has no s_barrier and reads guest memory through BDA. With this change, its replayed request recompiles completely (388,903 SPIR-V words) once GE_CNTL.BREAK_WAVE_AT_EOI is set in the request; the title clears that bit, which #4201 refuses and #4450 handles.

Tested

Based on main f265cd5 plus #4172 and #4201. Windows 11 (MinGW-w64 GCC 15.2.0) and Linux: configure -DBUILD_TESTING=ON, targeted test builds, ctest.

Windows NVIDIA: agc_driver_tessellation, agc_shader_memory, agc_driver_graphics, agc_tessellation_factor_ring and agc_driver_bda_device pass (5/5)
Linux (upstream's job in WSL, GCC 13.3, llvmpipe), fix(shader): record BDA faults in hull programs instead of refusing them #4222's head merged with current main: the same 5 tests pass (5/5)
Full suite on fix(shader): record BDA faults in hull programs instead of refusing them #4222's head (which contains this commit): Windows NVIDIA 1110/1111, the failure is log_macros, which fails on main too; Linux (upstream's job in WSL) merged with main 5a81f60: 1089/1091, agc_driver_bc_block_copy and agc_driver_depth_compare_copy fail on that main as well

agc_driver_tessellation gains two hulls that scale their output by a float read through BDA, with and without s_barrier: both render correctly, and an unmapped read reports BDA access failed ... stage=7. Both this and agc_driver_bda_device fail on #4201's head with the refusal. The unmapped-read draws are skipped on llvmpipe, which segfaults on them although the loads are guarded by if (physical != 0) and removing just those guarded loads stops the crash; the mapped-read draws run there.

Not tested: console behaviour.

Checklist

blawesom · 2026-10-12T13:18:31Z

Rebase handoff — this PR is the 5th link of the hull/tessellation chain (it contains #4172 and #4201 underneath its own commit), so I stacked the rebase on the already-rebased chain: rebase/4114 → … → rebase/4201 → this. Authorship unchanged; applied cleanly.

Verified on Linux (CachyOS, kernel 7.2.9, GCC 16.2.1, Ninja, Release, BUILD_TESTING=ON, AMD RX 9070 XT / RADV):

builds; tools/check_conventions.py --base origin/main passes
full suite: 952/952 passed

To pick it up (whole chain):

git fetch https://github.com/blawesom/AnyPS5 rebase/4114 rebase/4163 rebase/4172 rebase/4201 rebase/4222

Head is d321aca7b. Worth listing Depends on: #4114, #4163, #4172, #4201 in the description so the chain merges in order.

Celegans12 · 2026-10-14T01:56:42Z

Quick note: I restacked this on the new #4172 and #4201 heads. The only conflict was the order of the test calls in Tessellation.cpp; this commit's own diff is unchanged. On this head, agc_driver_tessellation, agc_shader_memory, agc_driver_graphics, agc_tessellation_factor_ring and agc_driver_bda_device pass on Windows (NVIDIA) and on Linux (llvmpipe, merged with main).

A wave64 hull program narrows EXEC with s_and_saveexec_b64 and restores it later, so its exec words reach the lane-bit test as ANDs with the host ballot, as loop or join phis, and through separate ballots for the two halves. Tessellation lowering refused them as "uses a lane mask other than its own lane's bit". The own-lane bit of a mask pair is now rebuilt from the words' operations (ballots, bitwise ops, selects, phis, constants); both halves must have the same shape and ballot the same predicate. Dead hull cycles such as a saved-exec loop are removed before the remaining-ballot check, so a mask whose other lanes are read (s_bcnt1_i32_b64 exec) is still refused. The words are replaced through identities and block conditions are resolved, so a branch on a compare copied to VCC keeps its condition. Local and domain programs keep their exec words for the half-wave matcher. Block conditions fix and HullCompareBranch test from EagleBearerYS.

… data Hull programs whose relative patch index (v1 bits 7:0) reached anything but a tessellation ring or LDS address threw, because the lowering runs every Vulkan patch as relative patch 0. The quad terrain hull 0x80002fc700 in Ghost of Yotei is one, and so is the domain 0x8068fbd900 (v7). Such hulls and domains now take (PrimitiveID mod GE_CNTL.PRIM_GRP_SIZE) mod VGT_LS_HS_CONFIG.NUM_PATCHES, as if hull threadgroups took NUM_PATCHES patches at a time out of each primitive group; both stages see the same PrimitiveID for a patch. This holds only when every instance starts a new threadgroup, so it needs GE_CNTL.BREAK_WAVE_AT_EOI (bit 18); with the bit clear the instances share threadgroups and such programs still throw. PRIM_GRP_SIZE and BREAK_WAVE_AT_EOI join the tessellation configuration (draw keys, hull and domain cache keys, request version 24). A missing GE_CNTL throws, and so do local programs, whose relative vertex index depends on vertex reuse that Vulkan does not expose.

Every hull program with a BDA access was refused with "BDA fault termination requires a barrier-safe tessellation-control execution protocol". The refusal predates BdaInvocationsMayStop: a program with s_barrier no longer stops a faulting invocation; the fault is recorded, the read returns zero and every invocation reaches the barrier. In a hull, s_barrier is the only control barrier the backend emits after the guest code starts, so that rule already covers it, and a hull without one can stop an invocation without stranding a barrier. The tessellation test draws hulls that scale their levels by a BDA read, with and without s_barrier, and requires an unmapped read to report a tessellation-control fault (not on llvmpipe, which runs hull loads that the unmapped-address check skips). The contract test now accepts a hull with s_barrier.

github-actions Bot added shader core/shader docs agc libSceAgc* labels Oct 12, 2026

Celegans12 force-pushed the fix/agc-tessellation-hull-bda-faults branch from 47446b9 to 22be94e Compare October 12, 2026 02:06

Celegans12 mentioned this pull request Oct 12, 2026

feat(agc): rebuild the hull's relative patch index when it is used as data #4201

Merged

11 tasks done

Celegans12 force-pushed the fix/agc-tessellation-hull-bda-faults branch from 22be94e to d537af1 Compare October 12, 2026 02:56

Celegans12 mentioned this pull request Oct 12, 2026

fix(agc): lower hull exec masks restored through phis to own lane bits #4172

Merged

11 tasks done

Celegans12 force-pushed the fix/agc-tessellation-hull-bda-faults branch from d537af1 to 699733d Compare October 12, 2026 03:45

github-actions Bot mentioned this pull request Oct 12, 2026

fix(agc): export the stencil reference from pixel shaders with VK_EXT_shader_stencil_export #4355

Merged

9 tasks done

github-actions Bot added the needs rebase Conflicts with main, rebase needed label Oct 12, 2026

This was referenced Oct 12, 2026

feat(agc): implement image tables loaded through 64-bit pointers #758

Merged

fix(shader): number an instanced draw's hull patches across its instances #4450

Merged

gjonn mentioned this pull request Oct 12, 2026

feat(shader): read hull inputs at a control-point index computed from data #4727

Merged

10 tasks done

Celegans12 force-pushed the fix/agc-tessellation-hull-bda-faults branch from 699733d to 3f165b6 Compare October 13, 2026 01:44

This was referenced Oct 13, 2026

feat(agc): emulate default eight-sample dithered alpha coverage #4451

Merged

fix(shader): place hull input components by their offset in the control-point slot #4723

Merged

github-actions Bot added needs rebase Conflicts with main, rebase needed and removed needs rebase Conflicts with main, rebase needed labels Oct 13, 2026

Celegans12 force-pushed the fix/agc-tessellation-hull-bda-faults branch from 3f165b6 to c4b02f0 Compare October 13, 2026 09:42

github-actions Bot added needs rebase Conflicts with main, rebase needed and removed needs rebase Conflicts with main, rebase needed labels Oct 13, 2026

Celegans12 force-pushed the fix/agc-tessellation-hull-bda-faults branch from c4b02f0 to 3a72fa3 Compare October 13, 2026 12:17

github-actions Bot mentioned this pull request Oct 13, 2026

fix(shader): read the ancillary render target index from Layer when the draw exports it #5097

Merged

10 tasks done

github-actions Bot removed the needs rebase Conflicts with main, rebase needed label Oct 13, 2026

This was referenced Oct 13, 2026

fix(shader): lower hulls with one input or one output control point (fork CI) EagleBearerYS/AnyPS5#2

Closed

fix(shader): lower hulls with one input or one output control point #5457

Merged

Celegans12 force-pushed the fix/agc-tessellation-hull-bda-faults branch from 3a72fa3 to f33872b Compare October 14, 2026 01:56

github-actions Bot mentioned this pull request Oct 14, 2026

feat(shader): link callees of calls whose target is loaded from a table #4706

Merged

10 tasks done

RINNECODER mentioned this pull request Oct 14, 2026

Upstream PR batch 1 U7: tessellation and draw-state keys RINNECODER/AnyPS5#483

Merged

github-actions Bot added the needs rebase Conflicts with main, rebase needed label Oct 14, 2026

Celegans12 added 3 commits October 14, 2026 04:57

Celegans12 force-pushed the fix/agc-tessellation-hull-bda-faults branch from f33872b to c1e4a0b Compare October 14, 2026 03:06

github-actions Bot removed the needs rebase Conflicts with main, rebase needed label Oct 14, 2026

oneandonlydean self-assigned this Oct 14, 2026

nikowolf123 mentioned this pull request Oct 14, 2026

feat(agc): blend into color targets with a reversed component order #5200

Merged

11 tasks done

oneandonlydean merged commit c698dae into boykopovar:main Oct 14, 2026
8 of 9 checks passed

Pipodi mentioned this pull request Oct 14, 2026

perf(recompiler): read the buffer stride and presence at run time in compute modules #5353

Closed

11 tasks done

Amoenus mentioned this pull request Oct 16, 2026

shader: faulting BDA reads in programs with s_barrier still update guest output buffers #6689

Open

Navigation Menu

Repository navigation

fix(shader): record BDA faults in hull programs instead of refusing them - #4222

fix(shader): record BDA faults in hull programs instead of refusing them#4222
oneandonlydean merged 3 commits into
boykopovar:mainfrom
Celegans12:fix/agc-tessellation-hull-bda-faults

Celegans12 commented Oct 12, 2026 •

edited

Loading

Uh oh!

blawesom commented Oct 12, 2026

Uh oh!

Celegans12 commented Oct 14, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

Celegans12 commented Oct 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What

Tested

Checklist

Uh oh!

blawesom commented Oct 12, 2026

Uh oh!

Celegans12 commented Oct 14, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Celegans12 commented Oct 12, 2026 •

edited

Loading