[milescoviello]

What

sceKernelRead and sceKernelWrite threw std::invalid_argument for a null buffer, even with a byte count of 0. The POSIX read, _read, write and _write in Stdio.cpp already handle a null buffer: a non-zero count fails with EFAULT, and a zero count is treated as an empty buffer, so the descriptor is still checked.

The sceKernel entry points now do the same: 0x8002000e for a null buffer with a non-zero count, otherwise a zero-length transfer on the descriptor (0, or 0x80020009 for a descriptor that isn't open). The guest socket path is unchanged.

This is the same approximation the POSIX wrappers use. FreeBSD looks the descriptor up first and fails with EFAULT only when it copies through the buffer. So on a closed descriptor it reports EBADF, and a read at end of file returns 0.

I don't know of a title that hits this. I found it while going through the argument throws in libkernel.

Tested

Ubuntu 24.04, GCC 13.3, Release, on main 6a32550.

• guest_posix_scalar_io: new checks for sceKernelRead and sceKernelWrite with a null buffer: a count of 1 gives 0x8002000e, a count of 0 gives 0, and a count of 0 on descriptor -1 gives 0x80020009. With main's Open.cpp the test aborts on the first of these (the throw). With this change it passes 3 of 3.
• ctest -R '^guest_|libc': 403 of 403 pass, with this and my other small libkernel argument fixes from today applied together.
• tools/check_conventions.py passes. tools/pr_overlap.py finds no conflicting lines. It couldn't compare against feat(aot): consolidate native ARM64 compiler, runtime and consumers #5021, test: organize library tests and label CTest suites #6044 and fix(libkernel): preserve directory identity across Windows renames #6228, which conflict with main. None of their diffs touches sceKernelRead or sceKernelWrite.
• Windows 11 Pro, Ryzen 7 5700X, WinLibs GCC 15.2.0 r7, Release, with the ten PRs of this series merged in order: this PR's test passes 3 of 3, and with only this PR's source reverted it fails with main's behaviour. (series run by @screamm)

Not checked: macOS, a console.

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

🤖 Generated with Claude Code

[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/File/src/Open.cpp
PR 6044	core/libs/tests/GuestPosixScalarIo.cpp

[screamm]

Windows cross-validation of the whole series (stacked on main 5f61879) is in #7023 (comment): this PR's test passes 3/3, and with only this PR's source reverted it fails with main's throw. Suite, conventions and two title runs show no regression.

Written with Claude Code assistance (validation by a subagent, reviewed in this session).