[MarrasFederico]

What

scePthreadAttrGetscope, scePthreadAttrSetscope, scePthreadAtfork and scePthreadSetconcurrency were libkernel_sys discovery stubs that threw. libkernel already implements their POSIX forms (pthread_attr_getscope, pthread_attr_setscope, pthread_atfork, pthread_setconcurrency). The four sce forms now call them and turn a POSIX error into 0x80020000 | errno, as #8987 does for the barrier, cpuclockid and concurrency forms; this builds on its SceResult helper.

shadPS4 exports scePthreadAttrGetscope (+7B2AEKKns8) as ORBIS(posix_pthread_attr_getscope), the same wrapping. It has no export for the other three. Their names and NIDs (YdZfEZfRnPk, i1kREW2pchs, gdnv6wF6hwI) are in the symbol database, and their signatures follow the same POSIX forms.

What they do is libkernel's behaviour:

• Scope: 0 (process) or 2 (system), with system as the default; any other scope and null pointers return EINVAL.
• Concurrency: a negative level returns EINVAL; otherwise the level is stored and read back by scePthreadGetconcurrency.
• Atfork: handlers are recorded in order, null handlers included, since no process forks.

The debt entry records that none of this is measured on a console.

Tested

Ubuntu 26.04.1, kernel 7.0.0-38, GCC 15.2.0, Release, full build with BUILD_TESTING=ON. Windows was not compiled. No title was run.

• New guest_kernel_sys_pthread, registered next to guest_kernel_sys_barrier (feat(libkernel_sys): implement the scePthreadBarrier, cpuclockid, concurrency and msync forms #8987, now merged):
  • Scope: a new attribute reads scope 2 and switches to 0 and back. Scopes 1, 3 and -1 return 0x80020016 without changing it. Null attribute or output pointers return 0x80020016.
  • Concurrency: level 3 reads back as 3. -1 returns 0x80020016 and keeps 3. 0 reads back as 0.
  • Atfork: two registrations, one with null handlers, append exactly those handlers in order.
• The test fails on this branch's base, where the four calls throw. Returning the raw errno instead of the sce code, or registering the atfork handlers in the wrong order, makes it fail.
• guest_kernel_sys and guest_kernel_sys_barrier still pass.
• tools/check_conventions.py --base origin/main passes. tools/pr_overlap.py finds no shared export. It reports neighbouring lines in docs/dev/debt/libkernel_sys.md with fix(libkernel_sys): return EPERM for unsupported fchown ownership #8457, fix(libkernel_sys): implement fchown #8531, feat(libkernel_sys): implement sceKernelTitleWorkaroundIsEnabled and sceKernelGetProcessType #9040 and feat(libkernel_sys): implement chown, lchown, fchownat and linkat #9103 (fchown, chown and title-workaround calls), and could not compare feat(aot): consolidate native ARM64 compiler, runtime and consumers #5021, which conflicts with main; none touches these four calls.

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 (0a0ed72 → de981a2)

📈 System libraries: 45.86% (+0.03%, +4 implemented functions)

✅ 4 implemented

Library	Name
libkernel_sys	scePthreadAtfork
libkernel_sys	scePthreadAttrGetscope
libkernel_sys	scePthreadAttrSetscope
libkernel_sys	scePthreadSetconcurrency

[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 5021	core/libs/CMakeLists.txt
PR 6044	core/libs/CMakeLists.txt
PR 6525	docs/dev/debt/libkernel_sys.md
PR 7201	core/libs/CMakeLists.txt
PR 7275	core/libs/CMakeLists.txt
PR 7622	core/libs/CMakeLists.txt
PR 7780	core/libs/CMakeLists.txt
PR 8285	core/libs/CMakeLists.txt
PR 8457	docs/dev/debt/libkernel_sys.md
PR 8650	core/libs/CMakeLists.txt
PR 8658	core/libs/CMakeLists.txt
PR 8701	docs/dev/debt/libkernel_sys.md
PR 9019	core/libs/CMakeLists.txt core/libs/prx/libkernel_sys/Export.Discovery.cpp docs/dev/debt/libkernel_sys.md
PR 9063	core/libs/CMakeLists.txt
PR 9097	core/libs/prx/libkernel_sys/Export.Discovery.cpp
PR 9103	core/libs/prx/libkernel_sys/Export.Discovery.cpp docs/dev/debt/libkernel_sys.md
PR 9161	core/libs/CMakeLists.txt core/libs/prx/libkernel_sys/Export.Discovery.cpp docs/dev/debt/libkernel_sys.md
PR 9162	core/libs/CMakeLists.txt core/libs/prx/libkernel_sys/Export.Discovery.cpp docs/dev/debt/libkernel_sys.md
PR 9172	core/libs/CMakeLists.txt core/libs/prx/libkernel_sys/Export.Discovery.cpp
PR 9173	core/libs/CMakeLists.txt
PR 9187	core/libs/prx/libkernel_sys/Export.Discovery.cpp
PR 9284	core/libs/CMakeLists.txt core/libs/prx/libkernel_sys/Export.Discovery.cpp
PR 9293	core/libs/CMakeLists.txt core/libs/prx/libkernel_sys/Export.Discovery.cpp
PR 9297	core/libs/CMakeLists.txt core/libs/prx/libkernel_sys/Export.Discovery.cpp docs/dev/debt/libkernel_sys.md
PR 9302	core/libs/CMakeLists.txt core/libs/prx/libkernel_sys/Export.Discovery.cpp docs/dev/debt/libkernel_sys.md
PR 9442	core/libs/prx/libkernel_sys/Export.Discovery.cpp
PR 9571	core/libs/CMakeLists.txt
PR 9693	core/libs/prx/libkernel_sys/Export.Discovery.cpp
PR 9802	core/libs/CMakeLists.txt
PR 9896	core/libs/CMakeLists.txt
PR 9924	core/libs/CMakeLists.txt
PR 9933	core/libs/CMakeLists.txt
PR 9942	core/libs/CMakeLists.txt
PR 10060	docs/dev/debt/libkernel_sys.md

[ycbuzi]

Windows run of d6b3846 merged onto main 0c6a5bd (a3706ed2cf, clean), the platform the description lists as not tested. Windows 11 Home 10.0.26200, MinGW-w64 GCC 15.2.0 (ucrt-posix-seh r7), Release, BUILD_TESTING=ON.

Built libkernel_sys, guest_kernel_sys_barrier_tests and guest_kernel_sys_pthread_tests with no errors or warnings. ctest --repeat until-fail:2 on guest_kernel_sys_barrier and guest_kernel_sys_pthread: both pass on both repeats. Not run: a title.

[oneandonlydean]

@MarrasFederico the four new sce forms (scePthreadAttrGetscope, scePthreadAttrSetscope, scePthreadAtfork, scePthreadSetconcurrency) look right to me: each one calls the libkernel POSIX form and goes through #8987's SceResult, and guest_kernel_sys_pthread covers them.

One thing blocks it. This branch carries an older copy of #8987 (6131ec265d), but #8987 was rebased since then and is now 106e73669d. The new head also adds the 0x80000000 count check to GuestKernelSysBarrier.cpp and orders the registration after guest_kernel_sys_thread_defaults. The two branches now conflict in core/libs/CMakeLists.txt, Export.Discovery.cpp, GuestKernelSysBarrier.cpp and docs/dev/debt/libkernel_sys.md, whichever lands first. If this one landed first, it would also drop #8987's new check.

Could you rebase onto #8987's current head (106e73669d), so this PR only adds your commit on top of it? Nothing else needs to change.

[MarrasFederico]

@oneandonlydean done. The branch now carries #8987's current head 106e73669d and only my commit on top. I can't force-push from here, so I didn't rebase; I rebuilt the commit with git cherry-pick onto 106e73669d, merged current main, and gave this branch exactly that tree with a merge commit (git diff between the two is empty).

• GuestKernelSysBarrier.cpp is feat(libkernel_sys): implement the scePthreadBarrier, cpuclockid, concurrency and msync forms #8987's file byte for byte, 0x80000000 count check included. The registration stays where feat(libkernel_sys): implement the scePthreadBarrier, cpuclockid, concurrency and msync forms #8987 put it, after guest_kernel_sys_thread_defaults, and guest_kernel_sys_pthread follows it.
• Two conflicts came up. In CMakeLists.txt the Windows PATH line now lists all four tests. In Export.Discovery.cpp scePthreadSetconcurrency now sits on feat(libkernel_sys): implement the scePthreadBarrier, cpuclockid, concurrency and msync forms #8987's base, without the stale scePthreadSetDefaultstacksize/scePthreadSetName stubs, which main moved to Export.cpp.
• All six guest_kernel_sys* tests pass, and tools/check_conventions.py --base upstream/main is clean.

[MarcZX10]

Reviewed exact head e8f8e59. The dependent #8987 changes are retained, and the checklist declares that dependency. The four new SCE forms delegate to the existing POSIX providers with SCE errno mapping; tests cover scope state/error preservation, concurrency, and atfork registration order. Local diff whitespace, conventions, registration, and current-main checks pass. The PR reports a Linux full build/focused tests, and an independent Windows build plus repeated barrier/pthread runs; I did not execute those binaries locally. Hosted product jobs are skipped by policy, and console behavior remains documented as unmeasured.