[gjonn]

What

The POSIX pread export threw on a null buffer, a negative offset and any failed host read, which terminates the process. It now returns -1 and sets the guest errno exactly like pwrite and sceKernelPread on main: EFAULT for a null buffer with a non-zero length, EINVAL for a negative offset, and the mapped host error otherwise (EBADF for an invalid or write-only descriptor).

PPSA01317 stops on its first boot on Windows when a read from its save folder fails, at:

terminate called after throwing an instance of 'std::runtime_error'
  what():  pread_nid_postfix: pread failed, fd=..., errno=...

Tested

Windows 11, MinGW, guest_file_vector adds pread/pwrite checks for a null buffer, a negative offset, descriptor 9999 and a write-only descriptor.

On main:

terminate called after throwing an instance of 'std::invalid_argument'
  what():  pread_nid_postfix: invalid argument
1/1 Test #869: guest_file_vector ................Exit code 0xc0000409***Exception

With the argument checks skipped, the invalid descriptor check on main stops at:

terminate called after throwing an instance of 'std::runtime_error'
  what():  pread_nid_postfix: pread failed, fd=9999, errno=9

With this change:

1/1 Test #869: guest_file_vector ................   Passed    0.05 sec
100% tests passed, 0 tests failed out of 1

guest_aio and guest_shm_open also pass.

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: PPSA01317 Gran Turismo 7 (PPSA01317) #4697
• AI-assisted: yes

[nicolasfalesy]

Linux test (Arch, GCC 16.2.1, glibc, Release), this PR merged onto main 7e181fc together with #4725, #4719, #4585 and #4582 (no conflicts): guest_file_vector passes. With this PR's source change reverted (test kept), it fails on Linux, so the check guards the change on Linux too.

[adriadam10]

Rebased onto current main (the partial EBADF handling in pread_nid_postfix overlapped this rewrite; kept this PR's version). guest_file_vector fails without the Stdio.cpp change and passes with it; scoped ctest (libs, libc, libkernel, 217 tests) is green on the rebase; merging once the full build is green.