[reyani-dev]

Resubmission of #2216, which was closed because no title was shown stopping on rand_r. With tzset (#3124) merged, PS5MC (PPSA99420) now stops at rand_r on main 06da117, per @babyhuey's catalog run with sce_module/libc.prx unwrapped (comment):

app.elf: symbol lookup error: app.elf: undefined symbol: dW3xsu3EgFI

dW3xsu3EgFI is rand_r. With this change the title moves on to rewind (3QIPIh-GDjw, #1871), per the same run. rand_r is still absent on current main, and no open PR implements it. The branch is rebased onto current main. Windows 11 (WinLibs GCC 15.2.0), Release: guest_math passes 3/3, and check_conventions.py and pr_overlap.py --branch pass.

---

What

Exports rand_r. In FreeBSD, rand and rand_r share one generator step (do_rand in lib/libc/stdlib/rand.c): rand advances the global state, and rand_r advances the caller's unsigned int. This moves the step that rand_nid_postfix already implements into an internal helper (NextRand_nid_no_patch, not exported) used by both, so rand_r matches this libc's rand exactly. rand's behaviour is unchanged. A null state pointer throws std::invalid_argument.

Title: PPSA99420 PS5MC imports it from libSceLibcInternal (resolved through libc.prx), per the #1578 audit. I haven't run the title.

Tested

Windows 11, WinLibs GCC 15.2.0 r7, Release, -DBUILD_TESTING=ON:

• New checks in guest_math_tests, next to tanf:
  • known answers from the Park-Miller step: state 1 → 33613 and state 0 → 16806, each written back;
  • for seeds 0, 1, 12345, 0x7ffffffe and 0xffffffff, eight rand_r(&seed) calls equal srand(seed) followed by eight rand() calls;
  • null state throws.
• Mutation check: not writing the state back, or advancing the global state instead, each make guest_math fail.
• cmake --build build --target libs: the patched libc.prx exports dW3xsu3EgFI (rand_r) and still cpCOXWMgha0 (rand), and the helper isn't exported.
• tools/check_conventions.py passes, and tools/pr_overlap.py --branch reports no overlap. It also merges cleanly with feat(libc): implement lrand48 and srand48 #2038 (lrand48/srand48, added just below srand), checked with git merge-tree.

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

🤖 Generated with Claude Code

[github-actions]

Progress report (06da117 → 86db61c)

➖ System libraries: 85.96% (+0.0%, +1 implemented functions)

✅ 1 implemented

Library	Name
libc	rand_r_nid_postfix

[reyani-dev]

Fork CI for this head merged onto current main (run 38099847203):

• Windows: 759/759 passed.
• Linux: 752/755 passed. The 3 failures are agc_unit_shadow, agc_driver_flat_apertures (timeout) and agc_driver_image_packed16_load, which fail identically in main's own push build (run 38098335801). This PR only touches libc, so they're unrelated.
• Relinker jobs: green.

[babyhuey]

Checked with #3190, #3191 and #3192 stacked on main d31b57a (all merge cleanly), catalog builds with sce_module/libc.prx unwrapped. Linux (Ubuntu 24.04, GCC 13) and Windows 11 (WinLibs GCC 15.2), Release: guest_math and guest_strings pass 5/5 (--repeat until-fail:5) on both, with no warnings in Math.cpp or StringExtensions.cpp.

PS5MC (PPSA99420) gets past rand_r on both and stops at setstate:

• Linux: app.elf: symbol lookup error: app.elf: undefined symbol: woQta4WRpk0
• Windows: FAIL: unresolved ELF import woQta4WRpk0 (setstate) from libSceLibcInternal.prx