[plexerPK]

What

Copy the complete guest KernelTimespec through the current Time.cpp ReadGuestTime helper before validating fields or starting a sleep. A stable inaccessible page now returns EFAULT through nanosleep and _nanosleep, or SCE_KERNEL_ERROR_EFAULT through sceKernelNanosleep, instead of faulting the host. Preserve errno across the copy; the POSIX wrappers set their error as before.

Extend GuestNanosleep with protected-page cases for all three exports: wholly inaccessible requests, a readable first field with an inaccessible second field, readable same-address controls, complete requests immediately before the protected page, restored permissions, and unchanged remaining-time output on failure. Existing duration/invalid-value/alias checks remain.

The request snapshot preserves the existing validation and sleep behavior. This revision rebases onto main 801d5b9bc11049e6274daca09d70690bd5b8d0a9 and reuses its checked ReadGuestTime helper; the SocketRuntime include and GuestSocketReadGuest call are removed.

Overlap: #8916 changes cancellation policy, while this change snapshots the input before validation. Prior source/component composition was checked at its head a80a247; that historical check is not a runtime validation of this revision. The manually inspected clock-setter changes in #11042 and #11476 and clock-source change in #9677 affect different functions. No dependency is required.

Tested

Current head 7af32724b934054e17db210ffa2d63f23511be31, on main 801d5b9bc11049e6274daca09d70690bd5b8d0a9:

• macOS arm64 synthetic component: exact SleepForRequest, three export bodies and current ReadGuestTime implementation, plus the complete modified GuestNanosleep fixture. Main aborts with SIGBUS; candidate passes.
• Candidate component passes AddressSanitizer and UndefinedBehaviorSanitizer with no diagnostics.
• Both complete changed translation units compile to x86_64 objects on macOS. Compile-only; no linked/native x86_64 execution claim.
• Conventions against origin/main and whitespace checks pass.
• The component uses no-op GuestArena read preparation and host timing/clock adapters, with errno preservation. It does not validate the full linked library, cancellation or write-watch integration.

Pending: the focused linked guest_nanosleep runtime test on this new current-main integration. No supported Linux/Windows runtime host is available locally. The previous approval was superseded by the review requesting this integration and retest.

Historical external Linux evidence: jotitadev tested previous head c95820f composed with main cb45ae2 on i7-8700/Ubuntu26.04/GCC15.2 Release. Linked guest_nanosleep passed 3/3 and focused sleep/time 16/16; main Time.cpp with the new test crashed. Those results do not validate the changed helper integration in this head. The reported 1361/1361 suite belonged to main only; combined SHA/raw logs were not supplied.

Not tested on this head: linked native Linux/Windows, real guest/title execution, or hosted runtime CI. No proprietary modules used.

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: Fixes libkernel: cross-page nanosleep request crashes the host instead of returning EFAULT #11568
• Unblocks: none
• GPUs tested: n/a
• AI-assisted: yes

[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/prx/libkernel/Time/Time.cpp
PR 7804	core/libs/prx/libkernel/Time/Time.cpp
PR 12130	core/libs/prx/libkernel/Time/Time.cpp

[MarcZX10]

Reviewed the current draft head 0d6a3fe10617cfff2bc3164f905521785eaf73e5. The checked copy precedes all KernelTimespec field reads, restores host errno, and returns the existing guest EFAULT mapping; the boundary cases cover inaccessible and split-page requests across all three exports. I found no source-level blocker in this diff.

I independently configured a fresh Linux Release build and built both libkernel and guest_nanosleep_tests successfully. The focused guest_nanosleep CTest aborts during HostCpu static initialization, before reaching assertions, because this Xeon E5-1620 v2 lacks the PS5-required LZCNT/BMI/AVX2/FMA features. This is not a passing runtime test; the macOS component-harness results in the description remain the available runtime evidence. The branch is 261 commits behind current main but merges cleanly. Please update/revalidate the integration on a supported host before marking the draft ready.

[MarcZX10]

Re-reviewed exact rebased head c95820f4e4511fc0aa8ce507f73d434f4d675fa6 against main 74acfad65b3f0fd4eab653f3a3989d298fba1157. The authored patch is unchanged from the previously reviewed head (same stable patch ID); rebase introduces no additional code changes. The checked full-structure copy still precedes all timespec field reads, preserves host errno, and returns the existing POSIX/SCE EFAULT mapping. Boundary controls and all three exported aliases remain covered.

Validation evidence: the prior exact-code Linux Release build of libkernel and guest_nanosleep_tests succeeded, but this host's CTest run aborted before assertions during HostCpu initialization because required CPU features are unavailable. The author reports the rebased macOS component test, ASan/UBSan, x86_64 translation-unit compiles, conventions and whitespace checks pass; the component's adapters and limits are described in the PR. Hosted conventions, eligibility, labels and report pass; native product jobs are skipped by path policy. No full linked runtime or title run is claimed. No source blocker remains in this focused change.

[jotitadev]

First full native run of guest_nanosleep on Linux, which the Tested section asks for (the earlier Linux attempt aborted in HostCpu init on a CPU without AVX2/BMI).

Exact head c95820f4e4511fc0aa8ce507f73d434f4d675fa6, merged on main cb45ae2dd. Linux x86_64 (bare metal, not WSL), i7-8700, Ubuntu 26.04, kernel 7.0, GCC 15.2, glibc 2.43, CMake 4.2.3, Ninja, Release, BUILD_TESTING=ON, full build.

• guest_nanosleep passes, 3 of 3 runs, with the linked executable and the real libkernel.
• The existing sleep/time tests pass, 16 of 16: guest_nanosleep, guest_nanosleep_long_request, guest_sleep_wake, the 8 guest_sleep_cancel_*, the 2 guest_timed_join_*, guest_freebsd_clocks, guest_time_conversion, guest_time_extras.
• Full non-GPU suite on the same main without the PR: 1361/1361.
• With only core/libs/prx/libkernel/Time/Time.cpp reverted to main (new test kept), guest_nanosleep crashes with SIGSEGV. So the unreadable-request case also crashes on Linux, and the test catches it there.

Written with AI assistance (Claude Code). I ran the builds and tests above on my machine.

[babyhuey]

Windows run of this head c95820f4e45, merged onto main 27ad7244d22 (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 2122 tests: 1422 pass, 0 fail, 700 skip. No failures. 27ad7244d22 itself skips the same tests on this machine; one is machine-local (pad_sdl, because a controller is plugged in).

AI-assisted: yes (Claude Code)

[MarcZX10]

Follow-up after comparing the reviewed head with current upstream main 801d5b9bc1: this PR still targets base 74acfad65b, and core/libs/prx/libkernel/Time/Time.cpp has since changed. Current main now contains a checked ReadGuestTime helper in that same file, used by other time APIs, while this PR adds SocketRuntime.hpp and a separate GuestSocketReadGuest call in SleepForRequest. Please rebase and integrate the full-request snapshot with the current helper/path, then rerun the focused linked guest_nanosleep test and conventions on the merged tree. The independent Linux 3/3 + 16/16 result at the submitted head is valuable, but does not validate this current-main integration. The earlier approval is superseded until that recheck.