[Fishybeing]

What

The guest's wchar_t is 16 bits (UTF-16, see WindowsFormatting.hpp; FormattingWide.cpp and _Mbtowcx already use 16-bit units). The string functions in core/libs/prx/libc/src/Strings.cpp took wchar_t* and forwarded to the host's std::wcs*/std::wmem*, which only matches the guest on Windows. On Linux the host's wchar_t is 32 bits, so these read and wrote guest strings in 4-byte units: wcslen of a guest string returned a wrong length, comparisons were wrong, and the copies could overrun the guest buffer.

This implements them on char16_t:

wmemchr, wmemcmp, wmemcpy, wmemmove, wmemset, wcslen, wcscmp, wcsncmp, wcscpy, wcsncpy, wcschr, wcsrchr, wcsstr, wcspbrk, wcsspn

Comparisons are on unsigned code units and return -1, 0 or 1, as the Windows build returned before, so Windows behaviour does not change.

The wide conversions (wcstod, wcstof, wcstold, wcstol, wcstoll, wcstoul, wcstoull, wcscoll, wcsxfrm, wcsrtombs) and swprintf have the same problem on Linux but need a UTF-16 conversion, so they are left for a follow-up and listed in TechnicalDebt.

guest_wide_strings covers every changed function, including code units above 0x7FFF, embedded terminators, wcsncpy padding and truncation, overlapping wmemmove and writes past the requested count.

Tested

• Windows 11, MinGW-w64 GCC 15.2.0 (WinLibs 15.2.0posix-14.0.0-ucrt-r7), Release, -DBUILD_TESTING=ON, on d5adabae
• Strings.cpp and GuestWideStrings.cpp build without warnings, also with -Wall -Wextra.
• ctest -E 'agc|shader|recompiler|Vulkan|vulkan|video_out': 197/197 passed, including guest_wide_strings and guest_wide_formatting. GPU tests not run.
• Not tested on Linux: I have no Linux machine. The new test is what would have failed there before this change.

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 in TechnicalDebt
• 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
• AI-assisted: yes

[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
#1092	docs/dev/TechnicalDebt.md L211
#1095	docs/dev/TechnicalDebt.md L211
#1128		wcschr_nid_postfix wcscmp_nid_postfix wcscpy_nid_postfix wcslen_nid_postfix wcsncmp_nid_postfix wcsrchr_nid_postfix wcsspn_nid_postfix wmemchr_nid_postfix wmemcmp_nid_postfix wmemset_nid_postfix
#1184	docs/dev/TechnicalDebt.md L211

[fisherjoey]

This looks like the same change as #1126: both add wcslen_nid_postfix, wcscpy_nid_postfix and wmemcpy_nid_postfix to libc/src/Strings.cpp for 16-bit guest wide strings, with a GuestWideStrings.cpp test under the same guest_wide_strings ctest name. Linking so they can be compared.

[adriadam10]

Checked the 15 rewrites against the C semantics (unsigned unit comparison, wcsncpy padding/truncation, overlapping wmemmove, empty needle in wcsstr, terminator in wcschr/wcsrchr) and ran guest_wide_strings on Linux: it passes, and fails with Strings.cpp reverted, so it covers the 32-bit host wchar_t case. Full ctest green (424/424).

Only issue was a TechnicalDebt.md adjacency conflict with main; I rebased and kept all lines. Merging once CI is green on the rebased head.

#1126 is the follow-up and will need a rebase after this lands.