[cocoa55]

What

On Linux, fscanf, vfscanf, scanf and vscanf threw NotImplemented on the 16-bit wide conversions (%ls, %lc, %l[, %S, %C), and so did the _s stream forms. sscanf already reads them with the guest scanner (ScanSource), which also has a StreamSource, but its wide branch only ran for contiguous strings (if constexpr (TSource::Contiguous)). The UTF-8 decoding is the existing sscanf behaviour; #12468 replaces it with the byte-by-byte rule measured on titles’ libc.prx.

• The wide branch of ScanSource no longer walks a const char*. Each source gets PeekWide, which decodes the next character without consuming it, and the branch calls Skip for its bytes only once the field takes the character. StringSource::PeekWide is the old ReadWideScanCharacter call. StreamSource::PeekWide reads the UTF-8 sequence with fgetc and pushes it back with ungetc, as the fwscanf stream reader in ScanningWide.cpp already does. A rejected character therefore stays in the stream, and the accept/reject logic is the same code for strings and streams.
• fscanf/vfscanf use the scanner when the format has a wide conversion (same check as sscanf, ScansWideInProcess) and the host vfscanf otherwise. ScanGuestStream (fscanf_s and friends) no longer rejects wide conversions and keeps the positional-argument check.
• RejectGuestWideScan is removed. ScanWideField gets WideScanCharacterLength, which DecodeUtf8 now uses too.
• docs/dev/debt/libc.md: the scanf family and _s stream entries no longer list fscanf as throwing.

Overlap: #9087 and #9340 change the same wide branch of ScanSource (secure-size violations, and surrogate pairs / non-ASCII scansets). Neither reads wide fields from a stream, so they don't duplicate this PR. Agreed order (asked on both PRs): #9087, then #9340, then this PR, which I'll rebase after each one merges. StringSource::Remaining is kept because #12530 uses it. #11726 and #12468 build on this PR.

Tested

Ubuntu 24.04, GCC 13.3, full build. New CheckNarrowFormatWideFields in core/libs/tests/GuestWideStreamScan.cpp (Linux only, like the sscanf wide tests): fscanf %ls/%lc on UTF-8, vfscanf %S with %ln, wide and narrow fields in one call, a scanset stopping before é and the é read by the next call, %3lc with the next character left in the stream, matching failure and EOF, fscanf_s/vfscanf_s sizes, scanf/vscanf on the guest stdin. It fails on main with fscanf 16-bit wide conversion on Linux not implemented. guest_formatting, guest_scan_count, guest_wide_stream_scan and file_streams pass. Full ctest: 2476 of 2491 pass, and the 15 failures (agc_* submit/metadata, guest_net_*, guest_tcp, guest_socket_*) fail the same way on main in the same container (no GPU, sandboxed network). No title tested.

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: fix(libc): follow snprintf_s and sscanf_s constraints as measured on a PS5 #9087, feat(libc): decode non-ASCII scanset members and non-BMP characters in the guest wide scanf #9340
• Fixes: none
• Unblocks: none
• GPUs tested: n/a
• AI-assisted: yes

[github-actions]

Progress report (f2d874c → 0193ac7)

📈 System libraries: 26.26% (+0.01%, +3 implemented functions)

✅ 3 implemented

Library	Name
libc	fscanf_nid_postfix
libc	vfscanf_nid_postfix
libc	vfscanf_s_nid_postfix

Changed libraries (implemented exports)

Library	Before	After	Change
libc	1489/1505 (98.94%)	1492/1505 (99.14%)	+0.20%

[MarcZX10]

I reviewed the scanner changes and the reported build/test results. git diff --check and the current-base conventions check pass. The author reports a full Ubuntu build, the focused scanner tests, and a main-vs-change control for the wider CTest failures; hosted product build jobs are currently skipped.

There is one repository-policy blocker before this can be approved. Both overlapping PRs #9087 and #9340 are still open, and I independently confirmed that merging this head with either branch produces a content conflict in core/libs/prx/libc/src/Formatting.cpp (#9340 also conflicts in docs/dev/debt/libc.md). CONTRIBUTING requires the authors to agree which overlapping PR goes first and declare that order with Depends on. The PR description identifies the overlaps and says the second merge will need a textual rebase, but its checklist still says Depends on: none, and I found no coordination with those authors in the PR discussion.

Please coordinate the merge order with the authors of #9087 and #9340, update the dependency line accordingly, and rebase when appropriate. This is a merge-order request; I did not find a separate correctness blocker in the reviewed code.

[cocoa55]

I reviewed the scanner changes and the reported build/test results. git diff --check and the current-base conventions check pass. The author reports a full Ubuntu build, the focused scanner tests, and a main-vs-change control for the wider CTest failures; hosted product build jobs are currently skipped.

There is one repository-policy blocker before this can be approved. Both overlapping PRs #9087 and #9340 are still open, and I independently confirmed that merging this head with either branch produces a content conflict in core/libs/prx/libc/src/Formatting.cpp (#9340 also conflicts in docs/dev/debt/libc.md). CONTRIBUTING requires the authors to agree which overlapping PR goes first and declare that order with Depends on. The PR description identifies the overlaps and says the second merge will need a textual rebase, but its checklist still says Depends on: none, and I found no coordination with those authors in the PR discussion.

Please coordinate the merge order with the authors of #9087 and #9340, update the dependency line accordingly, and rebase when appropriate. This is a merge-order request; I did not find a separate correctness blocker in the reviewed code.

Thanks for the review and for checking the conflicts yourself. I've asked the authors of #9087 and #9340 to go first (#9087 comment, #9340 comment) and changed the checklist to Depends on: #9087, #9340. I'll rebase this PR once they merge and re-request review then. If either author would rather go second, I'll update the dependency line.

[github-actions]

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 3965	docs/dev/debt/libc.md
PR 5021	docs/dev/debt/libc.md
PR 5303	docs/dev/debt/libc.md
PR 6819	docs/dev/debt/libc.md
PR 7447	docs/dev/debt/libc.md
PR 7544	docs/dev/debt/libc.md
PR 7715	docs/dev/debt/libc.md
PR 7851	docs/dev/debt/libc.md
PR 8150	docs/dev/debt/libc.md
PR 8417	docs/dev/debt/libc.md
PR 8965	docs/dev/debt/libc.md
PR 9371	docs/dev/debt/libc.md
PR 9447	docs/dev/debt/libc.md
PR 9466	docs/dev/debt/libc.md
PR 9526	docs/dev/debt/libc.md
PR 9677	docs/dev/debt/libc.md
PR 10002	docs/dev/debt/libc.md
PR 10007	docs/dev/debt/libc.md
PR 10113	docs/dev/debt/libc.md
PR 10226	docs/dev/debt/libc.md
PR 10253	docs/dev/debt/libc.md
PR 10274	docs/dev/debt/libc.md
PR 10321	docs/dev/debt/libc.md
PR 10397	docs/dev/debt/libc.md
PR 10506	docs/dev/debt/libc.md
PR 10510	docs/dev/debt/libc.md
PR 10512	docs/dev/debt/libc.md
PR 10569	docs/dev/debt/libc.md
PR 10780	docs/dev/debt/libc.md
PR 10987	docs/dev/debt/libc.md
PR 11109	docs/dev/debt/libc.md
PR 11172	docs/dev/debt/libc.md
PR 11262	docs/dev/debt/libc.md
PR 11266	docs/dev/debt/libc.md
PR 11332	core/libs/prx/libc/src/Formatting.cpp docs/dev/debt/libc.md
PR 11625	docs/dev/debt/libc.md
PR 11634	docs/dev/debt/libc.md
PR 11661	docs/dev/debt/libc.md
PR 11664	docs/dev/debt/libc.md
PR 11689	docs/dev/debt/libc.md
PR 11698	docs/dev/debt/libc.md
PR 11831	docs/dev/debt/libc.md
PR 11848	docs/dev/debt/libc.md
PR 11870	docs/dev/debt/libc.md
PR 11895	docs/dev/debt/libc.md
PR 12053	docs/dev/debt/libc.md
PR 12057	docs/dev/debt/libc.md
PR 12184	docs/dev/debt/libc.md
PR 12514	docs/dev/debt/libc.md
PR 12518	docs/dev/debt/libc.md
PR 12558	core/libs/prx/libc/src/Formatting.cpp
PR 12574	docs/dev/debt/libc.md
PR 12592	docs/dev/debt/libc.md
PR 12593	docs/dev/debt/libc.md
PR 12594	docs/dev/debt/libc.md
PR 12598	docs/dev/debt/libc.md
PR 12605	docs/dev/debt/libc.md
PR 12622	core/libs/prx/libc/src/Formatting.cpp docs/dev/debt/libc.md
PR 12627	docs/dev/debt/libc.md
PR 12682	docs/dev/debt/libc.md
PR 12688	docs/dev/debt/libc.md
PR 12700	docs/dev/debt/libc.md
PR 12704	docs/dev/debt/libc.md
PR 12849	docs/dev/debt/libc.md
PR 12920	docs/dev/debt/libc.md
PR 12947	docs/dev/debt/libc.md
PR 12965	docs/dev/debt/libc.md
PR 13390	docs/dev/debt/libc.md
PR 13449	docs/dev/debt/libc.md
PR 13486	core/libs/prx/libc/src/Formatting.cpp docs/dev/debt/libc.md
PR 13794	docs/dev/debt/libc.md
PR 13804	docs/dev/debt/libc.md
PR 13809	docs/dev/debt/libc.md
PR 13828	docs/dev/debt/libc.md
PR 14212	docs/dev/debt/libc.md
PR 14298	docs/dev/debt/libc.md
PR 14326	docs/dev/debt/libc.md
PR 14588	docs/dev/debt/libc.md
PR 14595	docs/dev/debt/libc.md
PR 14610	docs/dev/debt/libc.md
PR 14915	docs/dev/debt/libc.md
PR 15095	docs/dev/debt/libc.md
PR 15199	docs/dev/debt/libc.md
PR 15206	docs/dev/debt/libc.md
PR 15236	docs/dev/debt/libc.md
PR 15336	docs/dev/debt/libc.md
PR 15471	docs/dev/debt/libc.md
PR 15484	docs/dev/debt/libc.md
PR 15486	docs/dev/debt/libc.md
PR 15490	docs/dev/debt/libc.md
PR 15501	docs/dev/debt/libc.md
PR 15519	docs/dev/debt/libc.md
PR 15535	docs/dev/debt/libc.md
PR 15552	docs/dev/debt/libc.md
PR 15561	docs/dev/debt/libc.md
PR 15602	docs/dev/debt/libc.md
PR 15614	docs/dev/debt/libc.md
PR 15667	docs/dev/debt/libc.md
PR 15683	docs/dev/debt/libc.md
PR 15713	docs/dev/debt/libc.md
PR 15714	docs/dev/debt/libc.md
PR 15768	docs/dev/debt/libc.md
PR 15773	docs/dev/debt/libc.md
PR 15778	docs/dev/debt/libc.md
PR 15791	docs/dev/debt/libc.md
PR 15796	docs/dev/debt/libc.md
PR 15800	docs/dev/debt/libc.md
PR 15801	core/libs/prx/libc/src/Formatting.cpp core/libs/tests/GuestWideStreamScan.cpp docs/dev/debt/libc.md
PR 15802	core/libs/prx/libc/src/Formatting.cpp core/libs/tests/GuestWideStreamScan.cpp docs/dev/debt/libc.md
PR 15803	core/libs/prx/libc/src/Formatting.cpp core/libs/tests/GuestWideStreamScan.cpp docs/dev/debt/libc.md
PR 15804	core/libs/prx/libc/src/Formatting.cpp core/libs/tests/GuestWideStreamScan.cpp docs/dev/debt/libc.md
PR 15807	core/libs/prx/libc/src/Formatting.cpp core/libs/tests/GuestWideStreamScan.cpp docs/dev/debt/libc.md
PR 15809	docs/dev/debt/libc.md
PR 15834	docs/dev/debt/libc.md
PR 15879	docs/dev/debt/libc.md
PR 15973	docs/dev/debt/libc.md
PR 16004	docs/dev/debt/libc.md
PR 16017	docs/dev/debt/libc.md
PR 16083	docs/dev/debt/libc.md
PR 16084	docs/dev/debt/libc.md
PR 16107	docs/dev/debt/libc.md
PR 16170	docs/dev/debt/libc.md
PR 16178	docs/dev/debt/libc.md
PR 16210	docs/dev/debt/libc.md
PR 16214	docs/dev/debt/libc.md
PR 16216	docs/dev/debt/libc.md
PR 16280	docs/dev/debt/libc.md

[maskym]

Windows result for head 6a95925815, merged into main f759e165d9 (the PR's Tested section names no Windows run).

Host: Windows 11 26300, Intel i7-12700K, WinLibs GCC 15.2.0 (x86_64-ucrt-posix-seh, r6), Release, BUILD_TESTING=ON, Ninja. Build clean.

The tree also carried the open PRs #11644, #11639, #11636, #11635, #11630, #11353, #11347, #11221, #10925, #11706, #11693, #11686, #11682, all merged cleanly, so the run is a joint one; the tests of this PR are guest_wide_stream_scan (0.05 s); note wchar_t is 16 bits on this host, so the Windows side of the change is what ran, all passed (the batch's mapped tests: 16/16, plus 43/43 for the Ngs2 and host-import suites). Not run on a PS5.

AI-assisted: yes (Claude Code), the runs and the numbers above are from this machine.

[JJenkx]

@cocoa55 measurements that bear on this PR's decoding: titles' own libc.prx doesn't decode UTF-8 in the wide scanf conversions. I ran the sscanf of PPSA07923, PPSA05916 and PPSA13934 natively on Windows 11 (a scratch loader calls it by NID, 1Pk0qZQGeWo) on about 70 input/format pairs. All three titles agree on every row.

What they do. Each byte goes to mbrtowc on its own, and the field width counts bytes. Since MB_CUR_MAX is 1, any byte ≥ 0x80 ends the conversion:

input (hex)	format	titles	main on Windows (UCRT)
c3 a9 ("é" in UTF-8)	%ls	0, nothing stored	1, 00c3 00a9 0000
61 c3 a9	%ls	0, 0061 stored, no terminator	1, 0061 00c3 00a9 0000
80	%ls	-1	1, 0080 0000
7f 80	%ls	0, 007f stored	1
c3 (end of input)	%ls	-1	1
c3 20 78	%ls	-1	1
61 c3 62	%2ls	-1, 0061 stored	1, 0061 00c3 0000
c3 a9	%lc	-1	1, 00c3
c3 61	%2lc	0	1
35 20 c3 a9	%d %ls	1	2
61 62 63	%C	0, nothing stored	(throws here)
61 62 63	%S	1, as %ls	(throws here)

Rules that fit every row.

• Lead byte (0xC0–0xEF, where mbrtowc returns -2): the byte is held.
  • Any further byte the field would take then fails the conversion as a matching failure.
  • If the field ends (width used up, a space for %ls, a byte outside the set for %l[, or the end of input) while the byte is held, it's an input failure.
• Invalid byte (0x80–0xBF, 0xF0–0xFF): an input failure if it's the field's first byte, a matching failure otherwise.
• Return values: an input failure returns EOF unless an earlier conversion succeeded. A matching failure returns the count so far.
• Stored units: units read before the failure are stored, without a terminator, and the scan stops.
• %C is not a conversion at all: with any width or length it stores nothing and ends the scan.
• %S is %ls.
• %l[ tests membership on the raw byte.

So the UTF-8 decoding on main (and in this PR's PeekWide) differs from the titles for any byte ≥ 0x80. Windows differs too, in another way: main hands these conversions to the UCRT, which stores each byte as one unit.

Proposal. I'd like to make the wide branch follow these rules on both platforms. Your PR already makes the branch source-generic, so the natural order is yours first, with mine stacked on it (Depends on: #11724) changing only the per-character step. #11726 should be unaffected beyond a rebase. If you'd rather fold the rules into this PR yourself, say so and I'll stay out of it. Fields cut by fscanf on a stream are the one part I can't measure with this loader, so the stream position after a failure would follow the string case.

Probes and full outputs: widescan*.cpp and widescan*-PPSA*.txt (local, attachable on request).

AI-assisted: yes (Claude Code).

[JJenkx]

@cocoa55 following up on my note above: I've opened the stacked follow-up as #12468 (Depends on: #11724), so it's reviewable alongside yours. Your two commits come first there, unchanged. If you'd still prefer to fold the byte-by-byte rule into this PR, say so and I'll close #12468.

AI-assisted: yes (Claude Code).

[cocoa55]

@cocoa55 following up on my note above: I've opened the stacked follow-up as #12468 (Depends on: #11724), so it's reviewable alongside yours. Your two commits come first there, unchanged. If you'd still prefer to fold the byte-by-byte rule into this PR, say so and I'll close #12468.

AI-assisted: yes (Claude Code).

Thanks a lot for measuring this on the titles’ libc.prx, that’s much better evidence than the UTF-8 assumption both main and this PR carry. I’d prefer to keep them separate: #11724 stays “wide conversions work on streams” with today’s decoding, and #12468 switches both strings and streams to the byte-by-byte rule on top. Happy to rebase or adjust PeekWide if it makes #12468 simpler.

[ryanc2112]

Linux run of head 7f4610561, merged locally onto main 3bc58629fb (clean merge). Arch distrobox on SteamOS, Ryzen 5 2600, GCC 16.2.1, Release, BUILD_TESTING=ON, 3 runs each:

tree	guest_formatting, guest_wide_stream_scan, guest_scan_count, guest_wide_formatting, guest_stdio_variadic, guest_daemon
main	6/6 passed, 3/3 runs
this PR merged onto main	6/6 passed, 3/3 runs
main libc + only this PR's GuestWideStreamScan.cpp	guest_wide_stream_scan failed, 3/3 runs

The control aborts with fscanf 16-bit wide conversion on Linux not implemented, the throw this PR removes. Merged onto main, tools/progress.py --module libc no longer lists fscanf, vfscanf and vfscanf_s as todo.

#11726 contains this PR's commits; my run of #11726 is there.

AI-assisted: yes (Claude Code)

[SnekPolylepis]

@cocoa55 I opened #13696 for the same stream forms before I saw this PR. I'm closing it in favour of this one. Two things from it that may be useful here:

1. Look-ahead without ungetc. feat(libc): read 16-bit wide conversions in the scanf stream forms #13696 decides whether the next character belongs to the field from its lead byte alone, through StreamSource::Peek(), so it never needs to push back more than one byte. That still holds with byte-by-byte rules (fix(libc): read wide scanf fields byte by byte, as titles' libc.prx does #12468): any byte from 0x80 up ends the field, so a one-byte peek is enough to stop the stream right after the last byte used.
2. Null destination in the secure wide branch. In feat(libc): read 16-bit wide conversions in the scanf stream forms #13696, fscanf_s(stream, "%ls", static_cast<char16_t*>(nullptr), 1u) reached target[0] = u'\0' without checking the pointer. %l[ and %2lc with capacity 1 took the same path, and other capacities consumed input and returned 0 instead of EOF (oneandonlydean's review on feat(libc): read 16-bit wide conversions in the scanf stream forms #13696). If this PR's secure stream path has the same shape, it's worth rejecting a null destination before the capacity check, as the narrow branch does, and testing capacities 0, 1 and a larger one.

AI-assisted: yes (Claude Code)

[TechLevelZero]

✅ AMD GPU test: passed

PR head 0193ac7 merged onto main f2d874c · tested 2026-11-01 22:48 UTC · 16 min

GPU	AMD Radeon RX 6300 (RADV NAVI24), driver Mesa 25.2.8-0ubuntu0.24.04.4, Vulkan 1.4.318
Host	Ubuntu 26.04.1 LTS, kernel 7.0.0-42-generic, container ubuntu:24.04
Logs	build and test logs for this run
Build	✅ passed (Release, -DANYPS5_ENABLE_SPIRV_TOOLS=ON)
Tests	✅ 3,053 passed · 15 failed (0 new, 9 also fail on main, 6 unstable on main) · 58 skipped (ANYPS5_REQUIRE_VULKAN=1)
Test setup	CPU-only tests without a GPU; GPU tests (and any CPU-only test that failed or skipped without one) once on the GPU with VK_LAYER_KHRONOS_validation and ANYPS5_NO_SHADER_CACHE=1; main ran the same way
Test selection	3,126 of 3,168 tests run: the ones this change can reach, chosen by tools/select_tests.py from main's build graph
Vulkan validation	✅ no new validation IDs vs main (1232 tests run on the GPU)
Slow tests (>30 s, GPU tests timed with validation layers)	⚠️ agc_driver_pipeline_journal 120.5 s, ⚠️ agc_driver_graphics 117.7 s, ⚠️ agc_driver_graphics_maps_fallback 116.9 s, ⚠️ public_headers 71.4 s, ⚠️ agc_driver_bda_device 56.1 s

Unstable on main (6, these also failed in recent main runs on this machine; not counted against this PR)

agc_driver_array_layer_target_cpu: failed in 1 of the last 5 main runs
agc_driver_job_manager_compute_capacity: failed in 4 of the last 5 main runs
agc_driver_pipeline_journal: failed in 3 of the last 5 main runs
agc_driver_submit_progress: failed in 3 of the last 5 main runs
agc_submit_progress: failed in 4 of the last 5 main runs
agc_submit_progress_no_timeline: failed in 4 of the last 5 main runs

Failing on main too (9, not counted against this PR)

agc_driver_guest_memory_lifetime_direct
agc_driver_image_table
agc_driver_native_depth_bda_clear
agc_driver_native_depth_buffer_clear
agc_driver_native_depth_clears
agc_driver_native_multisample_depth
agc_idle_publish_wait
guest_mmap_executable
guest_socket_options_multicast

Validation IDs already present on main (5 IDs in 711 tests)

VUID-VkDeviceCreateInfo-pNext-pNext (711 tests): agc_compute_pipeline_replay, agc_driver, agc_driver_address_passes, agc_driver_address_passes_draw_transitions, agc_driver_address_passes_targets_split, agc_driver_alias_copy_label, agc_driver_alpha_to_coverage, agc_driver_array_layer_target, and 703 more
VUID-vkCmdCopyImage-srcImage-01548 (11 tests): agc_driver_depth_clear_hle, agc_driver_depth_plane_buffer_memory, agc_driver_depth_plane_cpu_write_after_memo, agc_driver_depth_plane_cpu_write_after_page_unwatch, agc_driver_depth_plane_cpu_write_after_partial_unwatch, agc_driver_depth_plane_partial_store, agc_driver_depth_plane_partial_store_unimported, agc_driver_depth_storage_dispatch, and 3 more
VUID-vkCmdCopyImage-srcImage-01551 (11 tests): agc_driver_depth_clear_hle, agc_driver_depth_plane_buffer_memory, agc_driver_depth_plane_cpu_write_after_memo, agc_driver_depth_plane_cpu_write_after_page_unwatch, agc_driver_depth_plane_cpu_write_after_partial_unwatch, agc_driver_depth_plane_partial_store, agc_driver_depth_plane_partial_store_unimported, agc_driver_depth_storage_dispatch, and 3 more
VUID-VkPipelineColorBlendStateCreateInfo-pAttachments-00605 (2 tests): agc_driver_prepared_exports, agc_driver_srgb8_color_target
VUID-VkShaderModuleCreateInfo-pCode-08737 (1 test): agc_driver_image_sample_derivatives

Skipped (58)

agc_driver_bda_reads_shadowed_results: skipped, the device reports watched host memory written when it imports or reads it
agc_driver_bda_reads_shadowed_results_per_build: skipped, the device reports watched host memory written when it imports or reads it
agc_driver_copied_writer_label_order: skipped, the device reports a minStorageBufferOffsetAlignment of 4, so the dispatch binds its record at +0x1004 in place and nothing is written back; the test needs an alignment above 4
agc_driver_copied_writer_label_order_drain: skipped, the device reports a minStorageBufferOffsetAlignment of 4, so the dispatch binds its record at +0x1004 in place and nothing is written back; the test needs an alignment above 4
agc_driver_image_table_vouched_lookups: skipped, the device reports 4 of 4 host-imported pages written after a GPU read, so a watched imported table cannot be vouched for
agc_driver_linear_target_direct_write_completion: skipped, the device does not keep host imports write-watched, so no completion stamp over the target is tracked
agc_driver_ngg_vertex_wave_reduction: skipped, s3 describes a 32-lane wave only on 32-lane subgroups, and the device has 64 lanes
agc_driver_sparse_dma_buf_import: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_driver_unit_shadow: skipped, the device does not import watched host memory for unit shadows
agc_driver_unit_shadow_budget: skipped, the device does not import watched host memory for unit shadows
agc_driver_vertex_elected_atomic_max: skipped, the device runs vertex programs in subgroups, so they keep their cross-lane reductions
agc_host_import_sparse: skipped, host imports are unavailable: /dev/udmabuf is not accessible
agc_import_settle_initial_dcc: skipped, the device has no deferred settle of host imports: imports under 1 MiB wait for one only on Windows
agc_pipeline_binary_vulkan: skipped, the device lacks VK_KHR_pipeline_binary
agc_shadowed_key_stores: skipped, the device imports no watched host memory under a unit shadow
agc_small_tiled_shadow_label: skipped, host imports are unavailable: the test block is not imported and watched
agc_storage_clear_hand_over: skipped, the device imports no watched host memory for the clear-key hand-over
agc_unit_shadow: skipped, the device does not import watched host memory for unit shadows
agc_unit_shadow_completion_store: skipped, the device reports watched host memory written when it imports or reads it
guest_http_uri_locale: 
guest_net_ctl: skipped, the host has no IPv4 address
guest_video_out_color_space: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_hdr: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_high_frame_rate: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_output_mode: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_resolution_status: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
guest_video_out_wqhd_detection: skipped, no display or Vulkan device: SDL_InitSubSystem(VIDEO | GAMECONTROLLER) failed: wayland,x11 not available
video_out_flip_aftergpu: skipped, no display or Vulkan device: DisplayWindow: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or platform
video_out_flip_backtoback: skipped, no display or Vulkan device: DisplayWindow: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or platform
video_out_flip_controls: skipped, no display or Vulkan device: DisplayWindow: Vulkan support is either not configured in SDL or not available in current SDL video driver (offscreen) or platform
video_out_flip_cpuflip: skipp

Reproduce

git fetch origin 0193ac78e6f2c3f6356e40298e7c3928a21cc3ad f2d874c1fdccf781c0fa5d86619bb5dd8894b958
git checkout f2d874c1fdccf781c0fa5d86619bb5dd8894b958 && git merge --no-ff 0193ac78e6f2c3f6356e40298e7c3928a21cc3ad
git submodule update --init --recursive
cmake -S . -B build -G Ninja -DCMAKE_BUILD_TYPE=Release -DBUILD_TESTING=ON -DANYPS5_ENABLE_SPIRV_TOOLS=ON
cmake --build build && cmake --build build --target libs
VK_DRIVER_FILES=/nonexistent/icd.json ctest --test-dir build -E '^(agc|shader|recompiler|videoout|video_out|guest_video_out|vulkan)|gpu' --timeout 120 --parallel 12 --output-on-failure
ANYPS5_REQUIRE_VULKAN=1 ANYPS5_NO_SHADER_CACHE=1 VK_INSTANCE_LAYERS=VK_LAYER_KHRONOS_validation \
  ctest --test-dir build -R '^(agc|shader|recompiler|videoout|video_out|guest_video_out|vulkan)|gpu' --timeout 120 --parallel 16 --output-on-failure
# failures are then re-run alone: --timeout 360 --parallel 2; new validation IDs: the test alone, twice, --parallel 1

_{Automated build and test on a contributor's AMD machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs.}