[Gruzzberry]

What

profil was a throwing stub in libkernel_sys. FreeBSD 11.4's sys_profil fails with EINVAL when scale is above 1 << 16, stops profiling and returns 0 when scale is 0, and otherwise records the buffer and starts the profiling clock. It never reads the samples pointer at call time; the clock does that on each tick.

profil in libkernel_sys now returns EINVAL for a scale above 1 << 16 and 0 for a scale of 0 (profiling is never on, so there is nothing to stop). A non-zero scale still throws NotImplemented_nid_no_patch, because the profiling clock that fills the sample buffer is not modelled. The declaration in Api.hpp now carries the real four arguments.

Tested

Windows 11 (10.0.26200), WinLibs MinGW-w64 GCC 15.2.0 r7, Ninja, Release, -DBUILD_TESTING=ON.

• guest_kernel_sys: scale of (1 << 16) + 1 and 0xFFFFFFFF give -1 with EINVAL; scale 0 returns 0 with a null buffer and with the unreadable address 0x10 (the pointer is not dereferenced). On main the test does not build against the stub, which takes no arguments; with this change it passes 3 runs out of 3. The non-zero scale throw is not asserted.
• Full suite on this branch and on main: the same tests fail on both under load (pr_overlap_encoding, public_headers, agc_import_settle_initial_dcc, agc_host_import_cut_lifetime, agc_driver_depth_plane_partial_store, audio_out2_timing, guest_job_manager_failures, guest_pthread_start, guest_pthread_rwlock, guest_select, guest_mutex_contention, guest_named_shm_windows_short_temp); guest_ampr_threads fails intermittently on both (0 of 4 on main, 3 of 4 here). No build failures on either.
• python tools/check_conventions.py --base upstream/main: passes.
• Not tested: Linux, a console, a title run.

Overlaps

• Line overlaps from tools/pr_overlap.py: feat(libkernel): implement four pthread_attr forms #9156 (neighbouring stub in Export.Discovery.cpp), feat(libkernel_sys): implement sigqueue #9389 (Export.cpp L335) and feat(libkernel_sys): implement is_in_sandbox #10421 (GuestKernelSys.cpp L532). Proposed on each that it goes first and this PR rebases.
• The earlier head of this PR also carried a swapon implementation by mistake. That is removed; swapon is only in feat(libkernel_sys): implement swapon #11034.

Checklist

• Based on current main; no other open PR implements the same function
• 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); no silent stubs added
• No notes, investigation .md files or images added to the repository
• New third-party code is a submodule built from source: none added
• Generic behaviour, not specific to one title
• Depends on: feat(libkernel): implement four pthread_attr forms #9156; merge order: after feat(libkernel): implement four pthread_attr forms #9156, feat(libkernel_sys): implement sigqueue #9389 and feat(libkernel_sys): implement is_in_sandbox #10421 (neighbouring lines)
• Fixes: none
• Unblocks: none
• GPUs tested: n/a
• AI-assisted: yes

AI-assisted (Claude Code).

🤖 Generated with Claude Code

[MarcZX10]

Requesting changes for an unrelated implementation included in this profil PR. The diff also removes the swapon stub from Export.Discovery.cpp, adds a four-argument swapon declaration, implements path validation and host filesystem checks for swapon in Export.cpp, and adds GuestEnotblk. None of this is described in the PR's What/Test sections, has no tests here, and is a separate API topic. This conflicts with the repository's one-topic-per-PR rule.

Please remove the swapon changes from this branch; if that behavior is still wanted, submit it as a separate change after checking overlap and documenting/testing its contract. The profil portion itself follows FreeBSD 11.4's scale limit and scale-zero behavior (sys_profil): https://github.com/freebsd/freebsd-src/blob/release/11.4.0/sys/kern/subr_prof.c#L1988-L2025. The local and hosted product jobs are skipped, so I have not independently run the reported Windows test.

[github-actions]

Other open pull requests touch the same code. Merging one will break the other, or both implement the same thing. If this one builds on them, list them in Depends on.

Pull request	Conflicting files (lines on main)	Same exports	Same new files
PR 10421	core/libs/tests/GuestKernelSys.cpp L676

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/prx/libkernel_sys/Export.cpp
PR 9097	core/libs/prx/libkernel_sys/Export.Discovery.cpp core/libs/prx/libkernel_sys/Export.cpp core/libs/prx/libkernel_sys/include/Api.hpp core/libs/tests/GuestKernelSys.cpp
PR 9162	core/libs/prx/libkernel_sys/Export.Discovery.cpp core/libs/prx/libkernel_sys/include/Api.hpp
PR 9242	core/libs/prx/libkernel_sys/Export.Discovery.cpp core/libs/prx/libkernel_sys/include/Api.hpp
PR 9284	core/libs/prx/libkernel_sys/Export.Discovery.cpp
PR 9293	core/libs/prx/libkernel_sys/Export.Discovery.cpp
PR 9297	core/libs/prx/libkernel_sys/Export.Discovery.cpp
PR 9302	core/libs/prx/libkernel_sys/Export.Discovery.cpp
PR 9383	core/libs/prx/libkernel_sys/Export.Discovery.cpp
PR 9389	core/libs/prx/libkernel_sys/Export.Discovery.cpp core/libs/prx/libkernel_sys/Export.cpp
PR 9713	core/libs/prx/libkernel_sys/Export.Discovery.cpp core/libs/prx/libkernel_sys/Export.cpp core/libs/prx/libkernel_sys/include/Api.hpp core/libs/tests/GuestKernelSys.cpp
PR 10427	core/libs/prx/libkernel_sys/Export.cpp core/libs/tests/GuestKernelSys.cpp
PR 10657	core/libs/prx/libkernel_sys/Export.Discovery.cpp
PR 11267	core/libs/prx/libkernel_sys/Export.Discovery.cpp core/libs/prx/libkernel_sys/include/Api.hpp

[Gruzzberry]

Thanks, you're right: the swapon hunks were a sibling branch's commit that leaked into this one. The branch is now rebuilt on current main with only profil: the stub removal, the four-argument declaration, the function and the four test lines. GuestEnotblk and every swapon line are gone, and swapon stays in #11034. Focused guest_kernel_sys passes 3 of 3, check_conventions.py --base upstream/main passes, and the full suite matches the earlier run. I've also updated the overlap note.

[wachi35676]

Tested head 69ead78b5e merged onto main 2b5225605b, Windows 11 (WinLibs GCC 15.2.0, Release) and WSL2 Ubuntu 24.04 (GCC 13.3), BUILD_TESTING=ON. This covers the Linux run listed as not tested.

• guest_kernel_sys and guest_filesystem pass 3 runs out of 3 on both hosts. With the new body replaced by main's stub (same signature), guest_kernel_sys fails with what(): profil_nid_postfix not implemented.
• The head now touches only profil. No swapon or GuestEnotblk lines are left.
• It matches release/11.4.0 sys_profil: scale > (1 << 16) gives EINVAL, and scale == 0 only stops the clock and returns 0 without reading samples. A real profiling request throws here, which the conventions expect.

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

Dismissing a stale review: the requested changes were addressed after it (new head) and the merge queue blocks on it. Please request changes again if anything is still open.

[adriadam10]

Checked against FreeBSD 11.4 sys_profil: scale > 1 << 16 gives EINVAL, scale == 0 returns 0 without reading the buffer, and a non-zero scale still throws since the profiling clock is not modelled. The guest_kernel_sys rows pin those cases and fail on main; the head only touches profil now, so the earlier swapon request is met. Confirmed by @wachi35676 on Windows 11 and WSL2; merged.