[LeandroLP23]

What

PICO-8 for PS5 (PPSA99808 v0.1.2, RafaelNGP/pico8-ps5, GPL-3.0) stops at start-up on this import once its earlier stops resolve (chain in #1578):

FAIL: unresolved ELF import JZKw5+Wrnaw from libkernel.prx

JZKw5+Wrnaw is pthread_sigmask (tools/nid_names.py). It follows FreeBSD libthr's, which calls sigprocmask and returns the error number instead of setting errno:

• it goes through the existing _sigprocmask, so it shares that function's process-wide mask; FreeBSD keeps one per thread, recorded in TechnicalDebt;
• how outside SIG_BLOCK/SIG_UNBLOCK/SIG_SETMASK with a set returns EINVAL, leaves the mask and the old set untouched and does not change errno; with a null set how is ignored, as in FreeBSD's kern_sigprocmask.

Tested

Windows 11, MinGW-w64 GCC 15.2.0, Release, main d70b899:

• PICO-8, with the earlier stops' pull requests merged locally, gets past this import and stops at the next one, vhtcIgZG-Lk from libSceLibcInternal.prx (realpath, feat(libc): implement realpath #1692).
• guest_signals checks blocking through pthread_sigmask and reading it back through sigprocmask, unblocking, the null-set query, and EINVAL for how 0 and 4 without touching errno.
• tools/check_conventions.py passes.

Fork CI on this head rebased onto main d70b899 (with the build workflow from before 0ab2a04, since the new one needs an upstream pull request number): https://github.com/LeandroLP23/AnyPS5/actions/runs/37921135437, all jobs pass (Windows 507/507, Linux 516/516).

#1916 (sigaction) also appends to GuestSignals.cpp; whichever merges second needs a one-line rebase.

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 in TechnicalDebt
• 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
• AI-assisted: yes (Claude Code)

[github-actions]

Progress report (8b393b6 → 97f6541)

📈 System libraries: 84.7% (+0.01%, +1 implemented functions)

✅ 1 implemented

Library	Name
libkernel	pthread_sigmask_nid_postfix

[adriadam10]

Closing for now: pthread_sigmask isn't a title's next stop on main. In the PICO-8 (PPSA99808) chain from #1578 it comes after pthread_setcanceltype, lstat, sigaction and other imports that are still unresolved, which is the same situation as #1692 (realpath), closed for that reason. The change itself looks right: it goes through _sigprocmask, EINVAL leaves errno alone, and the TechnicalDebt line records the process-wide mask.

Feel free to reopen, rebased, once a title stops on this import on main, with the log line in the description.

[overlordxrz-source]

@LeandroLP23 @adriadam10 pthread_sigmask is now a title's next stop on main. On our Linux rig, built from plain main at 7a7eb369f359aedbea1f60f148ecbff85b0986c7, both of these stop there:

./app.elf: symbol lookup error: ./app.elf: undefined symbol: JZKw5+Wrnaw
./app.elf: symbol lookup error: ./app.elf: undefined symbol: JZKw5+Wrnaw

Yamagi Quake II (PPSA99007) gets past system() (#1859) and now stops at JZKw5+Wrnaw = pthread_sigmask. Ruffle (PPSA68091) reaches the same import on main + #2028 (ModuleList/ModuleInfo). Could #2106 be reopened? We're happy to retest it on the rig against both titles.

AI-assisted: yes (the analysis; the rig run and logs are real).

[overlordxrz-source]

@adriadam10 quick question while I'm here: I've got a Linux rig running the homebrew catalog against main plus my PRs, and I'm looking to point it at whatever helps most. Is there a specific area (titles, subsystems, retests) where you'd appreciate a hand? Happy to take direction.

[babyhuey]

This is now a title's next stop on current main, so here is the log line the close note asked for, plus a check of this branch.

Checked 97f6541 merged onto main c963432, Linux (Ubuntu 24.04, GCC 13) and Windows 11 (MinGW-w64 GCC from WinLibs), Release. The only conflict is in GuestSignals.cpp, where main added the sigaction checks at the end of main(); I kept your pthread_sigmask block right after the sigprocmask lines it builds on, followed by main's block.

• guest_signals passes 5/5 on both (--repeat until-fail:5); no compiler warnings and no CRT assertion dialog on Windows.
• Yamagi Quake II (PPSA99007) stops at pthread_sigmask on main:
  • Windows main: FAIL: unresolved ELF import JZKw5+Wrnaw (pthread_sigmask) from libkernel.prx
  • With this branch, Linux: app.elf: symbol lookup error: app.elf: undefined symbol: oxMp8uPqa+U (pthread_set_name_np)
  • With this branch, Windows: FAIL: unresolved ELF import oxMp8uPqa+U (pthread_set_name_np) from libkernel.prx

Four more titles reach pthread_sigmask once the PR before it in their chain is in, on both Linux and Windows: VITA5 (PPSA39410), PS5 OpenGL Showcase (PPSA99005) and Homebrew UI Lab (PPSA99050) after _umtx_op (#1482), and ScreenTester5 (PPSA69420) after pthread_set_name_np (#1872).

[LeandroLP23]

I can't reopen this myself (GitHub only lets a maintainer reopen it). Yamagi Quake II (PPSA99007) now stops on pthread_sigmask on main, as @overlordxrz-source and @babyhuey showed above (FAIL: unresolved ELF import JZKw5+Wrnaw (pthread_sigmask) from libkernel.prx). @adriadam10, if you reopen it, I'll push the rebase right away: it's ready on main 04d06f0, with the GuestSignals.cpp conflict resolved the same way as @babyhuey's check. guest_signals passes 5/5 and the full non-GPU ctest passes on Windows.

[LeandroLP23]

A second title stops here on Windows: ps5-opengl (PPSA99005 v1.0.1). On main 89a1fb4 with #1482 (_umtx_op) merged, its next stop is

FAIL: unresolved ELF import JZKw5+Wrnaw from libkernel.prx

With the rebase of this PR on top it goes on to pthread_set_name_np (oxMp8uPqa+U). Radeon 860M, Windows 11 (26300), WinLibs GCC 15.2.0 r7, Release.

[LeandroLP23]

Update: in @babyhuey's new table on #1578 (comment, Linux main 197ffac and Windows main cf8b129), pthread_sigmask is now the next stop on main for five titles: Ruffle (PPSA68091), Yamagi Quake II (PPSA99007), Prospero Explorer (PPSA99109), Prospero Multi Tools (PPSA99110) and Prospero Vibrate (PPSA99920). The rebase is still ready and still applies cleanly to current main.

[babyhuey]

@LeandroLP23 pthread_sigmask is now the next stop on main for five catalog titles once each title's own libc.prx is loaded (required since #2605): Ruffle (PPSA68091), Yamagi Quake II (PPSA99007), Prospero Explorer (PPSA99109), Prospero Multi Tools (PPSA99110) and Prospero Vibrate (PPSA99920). Linux undefined symbol: JZKw5+Wrnaw, Windows FAIL: unresolved ELF import JZKw5+Wrnaw (pthread_sigmask) from libkernel.prx (table in #1578). That's what the close note asked for, and the intake check now needs 10 merged PRs, so this can go up again as a new PR.