[xXPBSXx]

What

Based on #6544 (notaxeltv, head 83e675746e; the first commit keeps its authorship). The change answers the asks of the review of 17 October one by one, on top of the lookup order that #8626 introduced:

1. Same directory first: a module next to the title, found by its exact name or ignoring case, wins over a file of the same name in prx, sce_module or sce_modules; the nested folders are tried only when the own directory has no match.
2. The nested lookup ignores case, like the lookup in the own directory.
   The order of RelinkedModulePath is now: exact name, own directory ignoring case, prx / sce_module / sce_modules ignoring case, then the stem fallback from fix(libc): find a relinked module by stem when the exact name misses (carries #6226) #8626. The first commit's nested lookup is unchanged apart from the new surrounding code. Two tests next to the original nested test: a same-directory module with a differently cased name wins over an exact-case nested file, and a nested module is found when its name differs in case.

Fixes: none

Tested

One base for all arms: main e59364ea98, head e2dda5eba0 (these two commits; git merge-tree against main is clean). Same checkout for every arm, only General.cpp swapped and rebuilt, the tests of the head kept. ctest -R '^guest_dynamic_loader$', 5 runs per arm:

• Linux (WSL2 Ubuntu 24.04: a virtualised kernel, not bare-metal; GCC 13.3, Release): head 5 of 5 pass; with General.cpp of the first commit alone (nested folders first) 0 of 5 pass, the same-directory test fails (GuestDynamicLoader.cpp:248); with General.cpp of main 0 of 5 pass, the nested test fails (GuestDynamicLoader.cpp:231).
• Windows 10 Pro 19045 (WinLibs GCC 15.2.0, Release): head 5 of 5 pass; with General.cpp of the first commit alone 5 of 5 pass, because the file system ignores case, so a Windows run cannot tell the two orders apart; with General.cpp of main 0 of 5 pass.
• tools/check_conventions.py --base upstream/main passes.
  Not tested: macOS, a title run, the variant with a case-sensitive nested lookup (the review of 23 October reports the case test fails there).

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

Co-authored by the original author of #6544.
AI-assisted (Claude Code).

🤖 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/libc/src/General.cpp
PR 7201	core/libs/prx/libc/src/General.cpp
PR 9371	core/libs/prx/libc/src/General.cpp

[harisibrar1]

Checked against main 6c52b64 and the relinker sources on main (NativeModules.cpp, GuestModuleBuilder.cpp):

• The new lookup matches the order agreed in #6544 (same directory, then the nested case-insensitive lookup, then the stem), and only covers directories the relinker itself writes modules to (sce_module/sce_modules, prx). It merges cleanly with main.
• I checked whether prx and sce_module could both hold a different module of the same name. GuestModuleBuilder.cpp:161-166 already refuses that ("Ambiguous guest dependency name"), so the fixed prx → sce_module → sce_modules order can't choose between two relinked copies. The exception is names that differ only in case on a case-sensitive host, where the relinker's map is case-sensitive but this lookup ignores case and takes prx first. Probably not worth handling.
• No overlap with #9415: that PR changes only DynamicLoader.cpp and matches by basename against any loaded module, and none of the review findings there apply to this lookup.

Reviewed with AI assistance (Claude Code); the checks above were run locally on my machine.

[MarcZX10]

Reviewed exact head e2dda5eba0037be211cb7b9489e461e1c12c4dbd. Lookup now checks the exact relinked path, then case-insensitive same-directory matches, then nested prx/sce_module/sce_modules, preserving the #8626 stem fallback. The tests cover nested resolution, case-insensitive nested lookup, and same-directory precedence over a conflicting nested file. The PR reports five-run focused red/green checks on Linux and Windows; hosted conventions, intake, eligibility, labels, relinker matrix, register-context, write-watch, and ASan checks pass. Product Linux/Windows builds were skipped, and I did not reproduce the local tests.