[MarcZX10]

What

Adds FT_Library_SetLcdFilter and FT_Library_SetLcdFilterWeights to libSceFreeType, forwarding to the bundled FreeType 2.13.3 through the guest library. The bundled build has FT_CONFIG_OPTION_SUBPIXEL_RENDERING off, so both return FT_Err_Unimplemented_Feature as in ftlcdfil.c and no filter takes effect. The five weights are read from the guest buffer through GuestArenaPrepareHostRead_nid_no_patch and copied before the call. Whether the console library enables subpixel rendering and its signatures are unverified; the debt entry says so.

Commits: for each function in turn (FT_Library_SetLcdFilter, then FT_Library_SetLcdFilterWeights), the implementation, its test and its debt entry.

Tested

Linux x86-64, GCC, full CMake build (Ninja, Release). guest_freetype_lcd_filter compares the return code of each call with the native API for every filter value including FT_LCD_FILTER_MAX, -1 and 99, for valid and null weights, and for a null library. The host CPU (Xeon E5-1620 v2) fails the startup CPU-feature gate, so I ran the test with that check disabled in a local, uncommitted edit; it exits 0. tools/check_conventions.py and git diff --check pass. No title or GPU involved.

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
• Fixes: none
• Unblocks: none
• GPUs tested: n/a
• AI-assisted: yes

[github-actions]

Progress report (b11c7bf → e38fcc3)

📈 System libraries: 26.71% (+0.0039%, +2 implemented, +2 declared, -2 removed functions)

✅ 2 implemented

Library	Name
libSceFreeType	FT_Library_SetLcdFilterWeights_nid_postfix
libSceFreeType	FT_Library_SetLcdFilter_nid_postfix

🆕 2 declared as stubs

Library	Name
libSceFreeTypeOt	sceFreeTypeOtLibrarySetLcdFilterWeights_nid_no_patch
libSceFreeTypeOt	sceFreeTypeOtLibrarySetLcdFilter_nid_no_patch

🗑️ 2 removed

Library	Name
libSceFreeTypeOt	sceFreeTypeOtUnknown_GiuNQjoj4fs_nid_no_patch
libSceFreeTypeOt	sceFreeTypeOtUnknown_VW4GD_TaIUc_nid_no_patch

Changed libraries (implemented exports)

Library	Before	After	Change
libSceFreeType	93/99 (93.94%)	95/101 (94.06%)	+0.12%
libSceFreeTypeOt	49/148 (33.11%)	49/148 (33.11%)	+0.00%

[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 9410	core/libs/prx/libSceFreeType/CMakeLists.txt core/libs/prx/libSceFreeType/include/Api.hpp docs/dev/debt/libSceFreeType.md
PR 9419	core/libs/prx/libSceFreeType/CMakeLists.txt core/libs/prx/libSceFreeType/include/Api.hpp docs/dev/debt/libSceFreeType.md
PR 9848	core/libs/prx/libSceFreeType/CMakeLists.txt docs/dev/debt/libSceFreeType.md
PR 9933	core/libs/prx/libSceFreeType/include/Api.hpp docs/dev/debt/libSceFreeType.md
PR 10897	core/libs/prx/libSceFreeType/CMakeLists.txt core/libs/prx/libSceFreeType/include/Api.hpp core/libs/prx/libSceFreeTypeOt/Export.cpp docs/dev/debt/libSceFreeType.md docs/dev/debt/libSceFreeTypeOt.md
PR 12004	core/libs/prx/libSceFreeType/CMakeLists.txt core/libs/prx/libSceFreeType/include/Api.hpp core/libs/prx/libSceFreeTypeOt/Export.cpp docs/dev/debt/libSceFreeType.md
PR 12370	core/libs/prx/libSceFreeTypeOt/Export.cpp docs/dev/debt/libSceFreeTypeOt.md
PR 12373	core/libs/prx/libSceFreeType/CMakeLists.txt core/libs/prx/libSceFreeType/include/Api.hpp core/libs/prx/libSceFreeTypeOt/Export.cpp docs/dev/debt/libSceFreeType.md
PR 12632	core/libs/prx/libSceFreeType/include/Api.hpp core/libs/prx/libSceFreeTypeOt/Export.cpp docs/dev/debt/libSceFreeType.md docs/dev/debt/libSceFreeTypeOt.md
PR 12814	core/libs/prx/libSceFreeTypeOt/Export.cpp docs/dev/debt/libSceFreeTypeOt.md
PR 13486	core/libs/prx/libSceFreeTypeOt/Export.cpp docs/dev/debt/libSceFreeType.md docs/dev/debt/libSceFreeTypeOt.md
PR 14085	docs/dev/debt/libSceFreeTypeOt.md
PR 14212	docs/dev/debt/libSceFreeType.md docs/dev/debt/libSceFreeTypeOt.md
PR 15052	core/libs/prx/libSceFreeTypeOt/Export.cpp docs/dev/debt/libSceFreeTypeOt.md
PR 15053	core/libs/prx/libSceFreeTypeOt/Export.cpp docs/dev/debt/libSceFreeTypeOt.md
PR 15604	core/libs/prx/libSceFreeType/CMakeLists.txt docs/dev/debt/libSceFreeType.md
PR 15687	core/libs/prx/libSceFreeTypeOt/Export.cpp docs/dev/debt/libSceFreeTypeOt.md
PR 15865	core/libs/prx/libSceFreeTypeOt/Export.cpp docs/dev/debt/libSceFreeTypeOt.md

[oneandonlydean]

@MarcZX10 The wrappers and the native comparison cases look right, but FT_Library_SetLcdFilter_nid_postfix and FT_Library_SetLcdFilterWeights_nid_postfix are declared only in core/libs/tests/GuestFreeTypeLcdFilter.cpp:14-15, so test_public_headers.py fails on them when merged with main. The repo rule is one declaration per export, in its library header. Please add both, with your existing signatures, to core/libs/prx/libSceFreeType/include/Api.hpp, include that header in src/LcdFilter.cpp and the test, and remove the local declarations. Then run public_headers, public_headers_compile_* and guest_freetype_lcd_filter.

[wachi35676]

Tested head 0d93517ee9 merged onto main 14eb2cda22 on Windows 11 (WinLibs GCC 15.2.0, Release, BUILD_TESTING=ON). This adds the Windows run to the Linux one in the description.

• The tree configures with the new test registration, and guest_freetype_lcd_filter builds and passes 3 of 3 runs on Windows.

AI-assisted: yes (Claude Code); results above were run on this machine.

[MarcZX10]

Head moved to ed644ceac8 (rebased onto current main, plus new commits): the sceFreeTypeOtUnknown_* alias whose NID is a symbol this PR implements is renamed after the symbol, as the unknown-nid convention check requires (with the debt counts updated), and the exports of this PR are declared once in libSceFreeType/include/Api.hpp, with the tests including that header instead of their own extern "C" blocks, as the public header audit requires. Behaviour is unchanged. tools/check_conventions.py --base upstream/main and test_public_headers.py --audit-only pass, and the changed sources pass a syntax check; the guest tests were not rerun for this change.

[oneandonlydean]

@MarcZX10 The two LCD wrapper declarations now use the canonical public API, clearing that part of the earlier ask. The same head also renames two helpers in libSceFreeTypeOt/Export.cpp:109-117 to sceFreeTypeOtLibrarySetLcdFilter_nid_no_patch and sceFreeTypeOtLibrarySetLcdFilterWeights_nid_no_patch, but neither name has a public declaration. Could you declare each once with its existing int() import-resolution signature in the shared API reached by libSceFreeTypeOt.hpp and include that header in Export.cpp? Please keep their NIDs and throwing placeholder behaviour unchanged, then run public_headers, public_headers_compile_* and guest_freetype_lcd_filter on the resulting head.

[MarcZX10]

Declared sceFreeTypeOtLibrarySetLcdFilter_nid_no_patch and sceFreeTypeOtLibrarySetLcdFilterWeights_nid_no_patch once each in libSceFreeType/include/Api.hpp with their existing int() signature (Export.cpp already includes that header), on top of a rebase onto current main. NIDs and throwing bodies are unchanged. On 315b087056 (Linux, host CPU gate disabled in a throwaway copy): guest_freetype_lcd_filter passes, public_headers and all 64 public_headers_compile_* batches pass, conventions pass. Not run: Windows.

[babyhuey]

Windows run of this head 315b0870568, merged onto main ea048ce2c35 (clean merge, local only), with the selection CI's skipped Windows product job uses: Windows 11, Ryzen 7 5800X, RTX 3080, WinLibs GCC 15.2, Release, BUILD_TESTING=ON. The tree was configured fresh after the merge, then select_tests.py select / ninja -f select.ninja selected / --target libs / select_tests.py ctest --parallel 4 --timeout 120.

• Selected 1427 tests: 1418 pass, 1 fails, 8 skip. The new guest_freetype_lcd_filter passes. The one failure is guest_raise_exception timing out at 60 s, which isn't from this PR: on this machine that test went from about 18 s on main 27ad7244d22 to 51-57 s on later main (51 s on ea048ce2c35 itself in this run's baseline), so it sits at the timeout under load. I'm narrowing that down separately; this PR doesn't touch its path.

AI-assisted: yes (Claude Code)

[oneandonlydean]

Thanks, declaring both helpers once in libSceFreeType/include/Api.hpp covers the header point. The rebase moved two lines in docs/dev/debt/libSceFreeTypeOt.md: the updated FT_Outline_Get_Orientation, ... FT_Library_SetLcdFilter and FT_Library_SetLcdFilterWeights entry now sits above ### Unknown function info, and the old "The 138 sceFreeTypeOtUnknown_*_nid_no_patch exports" line came back, though main replaced it with the "import-resolution helpers" entry (and there are 135 of them after this PR). Could you put the updated entry back in place of main's eight-name entry under ### Unknown function info, and drop the 138 line?

[MarcZX10]

Fixed on cb9e37c (rebased onto current main): docs/dev/debt/libSceFreeTypeOt.md is back to main's layout, with the identified-NID entry under ### Unknown function info in place of main's eight-name one (now naming FT_Library_SetLcdFilter and FT_Library_SetLcdFilterWeights too) and the 138 line gone. The file now differs from main only in that entry. Conventions pass. Not run again: the tests, which this commit does not touch.

[maskym]

Windows result for head cb9e37c4e5, merged into main 2ee8e9751a (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 #11172, #10780, #10729, #11614, #10631, #11547, #11584, #11572, all merged cleanly, so the run is a joint one; the tests of this PR are guest_freetype_lcd_filter. Together with the whole libc, CesCs, Font, FreeType, LibreSsl3 and NpUniversalDataSystem suites those tests belong to: 160/160 passed, this PR's in 0.06 s; the AudioOut suite separately 21/21. Not run on a PS5.

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

[oneandonlydean]

@MarcZX10 the code and tests are unchanged since the last review and the earlier ask is met, but the last restack left the debt file with two overlapping entries. At f1984cb the first lines of docs/dev/debt/libSceFreeTypeOt.md are:

1. a new entry listing FT_Outline_Get_Orientation, FT_Request_Size, FT_Sin, FT_Tan, FT_Vector_Unit, FT_New_Size, FT_Outline_Copy, FT_Outline_Decompose, FT_Library_SetLcdFilter and FT_Library_SetLcdFilterWeights as identified-but-unverified NIDs;
2. main's existing entry for the same first eight names plus the eleven FT_Stroker_* names, unchanged.

The same eight NIDs are therefore described twice, with different text, in one file. Please fold the two LCD names into main's existing entry (add "FT_Library_SetLcdFilter and FT_Library_SetLcdFilterWeights" to its list) and drop the new duplicate entry, so each NID is listed once. Nothing else changes.

[oneandonlydean]

@MarcZX10 the existing debt-consolidation request in #10582 (comment) still applies on a8f5479. Please fold FT_Library_SetLcdFilter and FT_Library_SetLcdFilterWeights into main's current identified-NID entry and remove the separate overlapping entry, preserving the other names already on main.

One numerical update to the earlier count advice: main 52c09d5 has 124 Unknown export definitions now, and this exact branch has 122. Where this PR edits the count entry, please leave one entry matching the actual source count rather than retaining the old 129-based value. The adapter behavior and prior header/policy fixes can stay unchanged. These counts should be re-derived after any further restack.

[MarcZX10]

Reviewed — debt-file requests still open at this head

The duplicate debt entry and the 129 count the reviewer asked to change are both still in the file at a8f5479.

What I ran

• Live head a8f5479 (matches batch).
• gh pr view 10582 (headRefOid, mergeable, statusCheckRollup); gh api GET of comments and reviews newer than 2026-10-27T10:47:16Z.
• git show of docs/dev/debt/libSceFreeTypeOt.md at the head; git show of core/libs/prx/libSceFreeTypeOt/Export.cpp at the head, with a grep count.
• Nothing was built or run.

Results

Check	Result	Note
eligibility, conventions, labels, report	SUCCESS	hosted
windows	SUCCESS	hosted
relinker, relinker-asan, register-context, write-watch	SKIPPED	hosted
linux	SKIPPED	hosted
merge state	CLEAN	hosted

Findings

• oneandonlydean (2026-10-28T14:57:41Z, and 2026-10-29T05:51:29Z): fold FT_Library_SetLcdFilter and FT_Library_SetLcdFilterWeights into main's identified-NID entry and remove the duplicate.
  • Not done. Debt doc line 3 is a separate entry that now lists the LCD names. Line 4 is main's entry, with the eight shared names and the FT_Stroker names, and no LCD names. The eight shared names appear twice.
  • Verified by reading.
• Count (2026-10-29T05:51:29Z): asks that the count match the source. Line 5 still says 129. Export.cpp has 122 sceFreeTypeOtUnknown_* definitions at this head (grep count of definition lines). Not done.

Sources

• AnyPS5/docs/dev/debt/libSceFreeTypeOt.md

  Lines 3 to 5 in a8f5479

  	- The FT_Outline_Get_Orientation, FT_Request_Size, FT_Sin, FT_Tan, FT_Vector_Unit, FT_New_Size, FT_Outline_Copy, FT_Outline_Decompose, FT_Library_SetLcdFilter and FT_Library_SetLcdFilterWeights NIDs in [Export.cpp](../../../core/libs/prx/libSceFreeTypeOt/Export.cpp) are identified, but their provider-specific prototypes and behavior remain unverified. Their named helpers retain the explicit inventory aliases and throw; the separate libSceFreeType implementations do not supply this provider.
  	- The FT_Outline_Get_Orientation, FT_Request_Size, FT_Sin, FT_Tan, FT_Vector_Unit, FT_New_Size, FT_Outline_Copy, FT_Outline_Decompose, FT_Stroker_New, FT_Stroker_Done, FT_Stroker_GetCounts, FT_Stroker_BeginSubPath, FT_Stroker_LineTo, FT_Stroker_EndSubPath, FT_Stroker_Set, FT_Stroker_Rewind, FT_Stroker_GetBorderCounts, FT_Stroker_ConicTo and FT_Stroker_CubicTo NIDs in [Export.cpp](../../../core/libs/prx/libSceFreeTypeOt/Export.cpp) are identified, but their provider-specific prototypes and behavior remain unverified. Their named helpers retain the explicit inventory aliases and throw; the separate libSceFreeType implementations do not supply this provider.
  	- The 129 `sceFreeTypeOtUnknown_*_nid_no_patch` exports in [Export.cpp](../../../core/libs/prx/libSceFreeTypeOt/Export.cpp) are present in the PS5 3.20 export inventory from [PS5-3.20_Libs](https://github.com/DNNDHH/PS5-3.20_Libs). Their prototypes and behavior are unverified. Each uses the no-parameter import-resolution convention and throws through `NotImplemented_nid_no_patch`; these declarations make no ABI or behavior claim.

• AnyPS5/core/libs/prx/libSceFreeTypeOt/Export.cpp

  Lines 1 to 10 in a8f5479

  	#include "prx/libSceFreeType/include/Api.hpp"
  	#include <cstddef>
  	#include <cstdint>
  	#include "SceTypes.hpp"
  	#include "prx/libSceFreeType/include/Api.hpp"
  	#include "prx/libc/include/General.hpp"
  	extern "C" {

Not verified

• The count of 122 is from a grep of the file; not compared with main 52c09d5.
• No build.

[oneandonlydean]

Hi MarcZX10, thanks for the rebase and for the debt change. Two things are still open in docs/dev/debt/libSceFreeTypeOt.md at e38fcc3256:

1. Line 3 is a new entry listing FT_Outline_Get_Orientation, FT_Request_Size, FT_Sin, FT_Tan, FT_Vector_Unit, FT_New_Size, FT_Outline_Copy, FT_Outline_Decompose, FT_Library_SetLcdFilter and FT_Library_SetLcdFilterWeights, and line 4 is main's entry for the same eight names plus FT_Glyph_Copy and the FT_Stroker_* names. The same NIDs are described twice. Please add the two LCD names to main's entry (line 4) and delete line 3.
2. Line 5 says 68 sceFreeTypeOtUnknown_*_nid_no_patch exports. Export.cpp at this head has 66 distinct names (68 on main; this PR renames two), so the count should be 66.