[fdfawef324]

What

A tessellation draw whose domain, partitioning or output topology the reference path does not support was refused with only VGT_SHADER_STAGES_EN:

AGC graphics: VGT_SHADER_STAGES_EN=0x200210d: only triangle and quad domains with integer, fractional-even or fractional-odd partitioning and triangle output are supported by the reference path

The same message covers an isoline domain, power-of-two partitioning, point or line output and the reserved encodings, so a log did not show which configuration was refused. Resident Evil 4 stops on this refusal.

• The refusal now also prints VGT_TF_PARAM with TYPE, PARTITIONING and TOPOLOGY, and VGT_LS_HS_CONFIG with NUM_PATCHES, HS_NUM_INPUT_CP and HS_NUM_OUTPUT_CP, for example:

  AGC graphics: VGT_SHADER_STAGES_EN=0x200210d: VGT_TF_PARAM=0x40045 (TYPE=1 TESS_TRIANGLE, PARTITIONING=1 PART_POW2, TOPOLOGY=2 OUTPUT_TRIANGLE_CW), VGT_LS_HS_CONFIG=0xc355 (NUM_PATCHES=85, HS_NUM_INPUT_CP=3, HS_NUM_OUTPUT_CP=3): only triangle and quad domains with integer, fractional-even or fractional-odd partitioning and triangle output are supported by the reference path
  

• The fields come before the explanation because the APS5_CAPTURE_TRACE log cuts the reason at 256 characters (PacketExecution.cpp, reason=%.256s). Up to the end of VGT_LS_HS_CONFIG the message is at most 240 characters, so the trace keeps every field; the [gpu] driver failure line on stderr prints the whole message.

• Field and enumerator names are those of Mesa's src/amd/registers/gfx103.json (VGT_TF_PARAM, VGT_LS_HS_CONFIG, enums VGT_TESS_TYPE, VGT_TESS_PARTITION, VGT_TESS_TOPOLOGY). A value with no name there (TYPE 3, PARTITIONING or TOPOLOGY 4-7) is printed as a number only.

• The printed values are the fields the decoder already extracts; which configurations are accepted is unchanged. After fix(agc): accept fractional-even tessellation partitioning #5402 the reference path accepts triangle and quad domains with integer, fractional-odd and fractional-even partitioning and clockwise or counter-clockwise triangle output; this PR implements no new mode.

Tested

Windows Server 2025 (10.0.26100), MinGW-w64 GCC 15.2.0, NVIDIA GeForce RTX 3060 (driver 591.86), on main at 6239ceffc.

• agc_driver_graphics: new UnsupportedTessellationFieldTests drives DecodeState into the refusal with VGT_TF_PARAM power-of-two triangle and quad patches, an isoline domain, line and point output, TYPE 3 and PARTITIONING 4 with TOPOLOGY 7, and with VGT_LS_HS_CONFIG 0xc355 and 0x1033f (3 input and 4 output control points, which checks the order), and checks that the message holds VGT_SHADER_STAGES_EN, then the decoded fields, then the explanation. It fails without the change (unexpected failure: AGC graphics: VGT_SHADER_STAGES_EN=0x200210d: only triangle and quad domains ... supported by the reference path) and passes with it (under 15 s for the whole executable; six runs here took 10.0 to 11.5 s).
• agc_driver_tessellation, agc_driver_alpha_to_coverage_state, agc_driver_undithered_alpha_to_coverage_state and agc_sparse_buffer_holes pass on the RTX 3060.
• No title was run.

Checklist

• Based on current main; no other open PR implements the same functions
• One topic per PR; follow-ups go in a new PR
• No comments in code except technical debt
• Unimplemented paths throw (NotImplemented_nid_no_patch); silent stubs are listed under Silent stubs in their library's technical debt file
• No notes, investigation .md files or images added to the repository (attach them to this PR)
• New third-party code is a submodule built from source
• Generic behaviour, not specific to one title
• Depends on: none
• Unblocks: none (Resident Evil 4 still stops on this refusal; the log now names its configuration)
• GPUs tested: NVIDIA GeForce RTX 3060 12 GB, driver 591.86 (32.0.15.9186), Vulkan 1.4.325, Windows Server 2025 10.0.26100
• AI-assisted: yes

🤖 Generated with Claude Code

[coltken95]

Tested on Linux x86-64 (Release build, lavapipe / llvmpipe LLVM 20.1.2), with d42f496 merged into main ab9a476:

• With the PR: all 8 tests built from the changed test file pass (agc_driver_graphics and the others in the same executable).
• With main's Graphics/src/State.cpp: agc_driver_graphics fails with AGC graphics: VGT_SHADER_STAGES_EN=0x200210d: only triangle and quad domains with integer, fractional-even or fractional-odd partitioning and triangle output are supported by the reference path; the other 7 pass.

AI-assisted: yes (Claude Code)

[github-actions]

Other open pull requests touch the same code. Merging one will break the other, or both implement the same thing. If this one builds on them, list them in Depends on.

Pull request	Conflicting files (lines on main)	Same exports	Same new files
PR 6612	core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp L905-925

Some comparisons could not be completed because a pull request conflicts with the base branch. Review these shared files manually before concluding that there is no overlap.

Pull request	Shared files not compared
PR 1191	core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp
PR 1395	core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 4040	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 4650	core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 5441	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 5499	core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 5500	core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 5925	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 5999	core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6188	core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6194	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6222	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6235	core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6297	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6472	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6529	core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6535	core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6548	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6563	core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6741	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6808	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6972	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6977	core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp
PR 6987	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6989	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 6995	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 7002	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp
PR 7426	core/libs/prx/libSceAgcDriver/tests/Graphics.cpp

[adriadam10]

Checked the diff: the refusal condition is unchanged, only the message gains VGT_TF_PARAM and VGT_LS_HS_CONFIG fields (front-loaded so the 256-char trace keeps them). UnsupportedTessellationFieldTests fails against main's State.cpp and covers unnamed encodings and the control-point order; also confirmed by @coltken95 on Linux/lavapipe. Merging once the full Build is green.