[savvaniss]

What

main (07da04f947) merged into ps4/agc-gcn (26d651d3e0), for boykopovar#9431. It is one merge commit, 4e8243ac48, with no rebase, so you can fast-forward ps4/agc-gcn to it or merge it. Two files conflicted; every other file merged automatically.

1. Graphics/include/GuestTextureResource.hpp: the struct keeps this branch's gcnPitch, gcnPow2Pad and gcnLinearGeneral, and main's fmaskCmaskAddress (the FMASK series from feat(agc): expose expanded FMASK for tracked native owners boykopovar/AnyPS5#7948).
   • DecodeTextureResource zeroes words 6 and 7 of a PS4 T# before it reads the metadata bits.
   • So fmaskCmaskAddress stays 0 for PS4 textures.
2. Graphics/src/DepthSurface.cpp, layeredDepthStorage: main's fix(agc): place a layered depth storage view at the guest's layer pitch boykopovar/AnyPS5#8109 removed the refusal of a layer pitch other than the depth slice size. Its LayeredDepthStorage now places layer n at n * GuestLayerOffset(1), so that refusal is gone here too.
   • This branch's Require(!geometry.levelMajor, ...) stays in front of it.
   • A level-major PS4 surface's layers are mip.tiledSize apart, not GuestLayerOffset apart, so main's new placement would be wrong for it.
   • This is the same guard as the branch's other two depth-plane sites (DepthPlaneCopy::Refresh and the sampled depth-plane slices).

Not in this merge: the Linux journal size check that oneandonlydean and ntamas94 raised on boykopovar#9431. On main, PipelineJournal.cpp now checks sizeof(State) == 840 and sizeof(ShaderStages) == 112. The merge doesn't touch them, and what the journal does with legacyVertex is still your call.

Untested after this merge: PS4 DB depth copies.

• main's feat(agc): draw DB depth copies with a driver pixel program boykopovar/AnyPS5#8999 replaced the DB depth copy compute pass (RunDepthCopyPass) with a draw through the normal path.
• That draw uses a driver-authored pixel program and a PS5 T# built by DepthCopySourceWords.
• With PS4 decoding on, that program goes through the GCN decoder. Its source plane comes from decodeDepthPlane, which now holds this branch's PS4 depth decode.
• This branch had no PS4-specific handling of DB depth copies before the merge either, and no test covers one with PS4 decoding on.

Tested

Windows 11 Pro 26300, WinLibs GCC 15.2.0, Release, AMD Radeon RX 9070 XT (driver 32.0.31041.1004), with ANYPS5_GPU=AMD, ANYPS5_REQUIRE_VULKAN=1, ANYPS5_NO_SHADER_CACHE=1 and VK_INSTANCE_LAYERS=VK_LAYER_KHRONOS_validation, ctest -V, on main 07da04f947 and on this merge.

Tests run: 50 in total:

• agc_driver, agc_driver_pm4, agc_driver_graphics (texture tiling, detiler, texture resource, color target layout), agc_driver_register_key, agc_driver_gcn_detile and agc_driver_pattern_buffer_fill
• agc_driver_flip and the 16 video_out_flip_* modes
• agc_gcn_decode, texture_format_lookup, agc_driver_resource_cache_replaced_texture and agc_depth_copy_kept_bytes
• the 23 agc_driver_depth_* tests, among them agc_driver_depth_storage, which covers fix(agc): place a layered depth storage view at the guest's layer pitch boykopovar/AnyPS5#8109's layered storage path

Results:

	passed	failed
main 07da04f947	48 of 49 (agc_driver_gcn_detile is not on main)	agc_driver_depth_plane_partial_store
this merge	49 of 50	agc_driver_depth_plane_partial_store

• agc_driver_depth_plane_partial_store fails the same way on both: "repeated lookups of the first layer of an unchanged array proved no refresh". It fails on every run on this machine, so it is not from the merge.
• Neither log has a VUID- line or was not found but was requested by env var VK_INSTANCE_LAYERS. Every Selected GPU line names the RX 9070 XT.
• tools/check_conventions.py --base origin/main reports no code findings.
• tools/pr_overlap.py --branch merge/9431-main (against main 88f99678bf) finds trial-merge conflicts with feat(agc): draw tessellation followed by an NGG geometry stage boykopovar/AnyPS5#6612, perf(agc): store whole words of 1- and 2-byte texels in the detile shader boykopovar/AnyPS5#6905, feat(agc): back extra guest mip levels with storage images boykopovar/AnyPS5#7085, fix(agc): raise no end-of-pipe event from flips and carry the release's interrupt context boykopovar/AnyPS5#7978, perf(agc): GPU descriptor heap on VK_EXT_descriptor_heap (APS5_GPU_HEAP) boykopovar/AnyPS5#9508, perf(agc): reuse user-data callee links and check a cached raw shader's range once boykopovar/AnyPS5#9899, fix(agc): follow RDNA 2's range checks for strided vertex fetches with OOB_SELECT 2 and 3 boykopovar/AnyPS5#9915, fix(agc): build a partially committed sampled 3D view from its committed slices boykopovar/AnyPS5#10064, fix(agc): clamp an exported pixel depth to the viewport depth range boykopovar/AnyPS5#10348, fix(shader): decode a T# with an unset format field as null boykopovar/AnyPS5#10556, perf(agc): take the texture detiler's Vulkan entry points from the device table boykopovar/AnyPS5#10764 and fix(agc): select the first level of a single-level depth view boykopovar/AnyPS5#11098.
  • All of them are in files that feat(libSceAgcDriver): draw PS4 (GCN) command buffers boykopovar/AnyPS5#9431 itself changes.
  • None is in GuestTextureResource.hpp or DepthSurface.cpp, the two files this merge resolves.
  • The pull requests stacked on feat(libSceAgcDriver): draw PS4 (GCN) command buffers boykopovar/AnyPS5#9431 share its new files and exports, as before.

Checklist

• Based on current main; no other open PR implements the same functions (main 07da04f947 merged into ps4/agc-gcn)
• 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
• Fixes: none
• Unblocks: feat(libSceAgcDriver): draw PS4 (GCN) command buffers boykopovar/AnyPS5#9431 and the pull requests stacked on it (fix(agc): decode PS4 texture bases, linear-aligned targets and thin-tiled 3D textures boykopovar/AnyPS5#9891, feat(agc): draw PS4 multisampled color targets as host-resident targets boykopovar/AnyPS5#9892, feat(agc): write RELEASE_MEM DATA_SEL 4 for PS4 guests as an 800 MHz count boykopovar/AnyPS5#9893, fix(agc): keep PS4 depth planes coherent with buffer reads of their memory boykopovar/AnyPS5#9895)
• GPUs tested: AMD Radeon RX 9070 XT, driver 32.0.31041.1004, Windows 11 Pro 26300
• AI-assisted: yes

🤖 Generated with Claude Code

[savvaniss]

Closing this: boykopovar#9431 merged into boykopovar/AnyPS5 main on 29 Oct, so this merge is no longer needed. Thanks. AI-assisted: yes (Claude Code)