Repair two function symbol identities without changing extents #720

Merged
coah80 merged 1 commits from agent/bittle/symbol-name-repairs into main 2026-10-02 04:25:16 +00:00
coah80 commented 2026-10-02 04:11:10 +00:00 (Migrated from github.com)

Metadata only — no new code or matching gain

Two reconstructed labels prevent exact-name tooling from finding the existing source functions:

  1. Remove the stray trailing backtick from pdm_disk_convert_block_into_sector. Its address remains 0x815C79D8, size 0xB0. The existing C function is already 44/44 instructions exact. After correction, all 17/17 functions in fa/pdm_disk pass exact-name ctxdiff with zero differences.
  2. Correct the omitted context parameter in __nupGetBootVersion__FP14ESTitleVersion to __nupGetBootVersion__FP14NUPContextInfoP14ESTitleVersion. Address remains 0x815F0A40, size 0x28C, scope local. The original caller moves context into r3 and title into r4 before the call; the callee saves r4 as its title pointer and reads the context's owned-title count/array from r3+0x28/r3+0x2C. This is a two-parameter ABI, matching the existing C++ definition and emitted symbol.

The NUP function is not a new match: it still has 163/163 instructions with 17 register-allocation differences, and now receives its honest exact-name objdiff score of 99.35583% instead of being unmapped. NUP's fuzzy unit score rises 93.442955 -> 99.46117% solely from attribution; exact code remains 7,496/10,764, exact functions 19/23, and data 1,720/1,720.

Verification on 184bee01

  • Only two names in config/43U/symbols.txt change; no address, size, split, scope, source, or padding changes
  • Rebuilt extraction, all_source, progress, report and DOL successfully
  • Compared allocated-section bytes and normalized relocations across all 1,027 source objects, including duplicate-named ELF sections: zero changes
  • All exact-code, exact-data, function-match and link measures unchanged; only the described NUP fuzzy attribution changes
  • pdm_disk ctxdiff 17/17 zero differences; NUP pool 28/28 identical
  • Workflow/pool/jump-table tests 7/8/27 pass; diff and main/leaf worktree checks pass
  • DOL SHA1 unchanged: 26116613f624061ba99c8d1a299aaa6efa85670d

Tested/published tree 428c78ab4c8605abe94c3426d2d4ae944219fbbe. Kept separate from source-matching PRs, as requested. Whole-project completion remains incomplete.

Note

Fix two symbol identities in config/43U/symbols.txt without changing extents

Removes an erroneous trailing backtick from the disk block-to-sector symbol entry. Adds the missing context-information parameter to the boot-version symbol's recorded signature. Symbol extents are unchanged.

Macroscope summarized 8fbd3a7.

## Metadata only — no new code or matching gain Two reconstructed labels prevent exact-name tooling from finding the existing source functions: 1. Remove the stray trailing backtick from `pdm_disk_convert_block_into_sector`. Its address remains **0x815C79D8**, size **0xB0**. The existing C function is already **44/44 instructions exact**. After correction, all **17/17 functions** in `fa/pdm_disk` pass exact-name ctxdiff with zero differences. 2. Correct the omitted context parameter in `__nupGetBootVersion__FP14ESTitleVersion` to `__nupGetBootVersion__FP14NUPContextInfoP14ESTitleVersion`. Address remains **0x815F0A40**, size **0x28C**, scope local. The original caller moves context into **r3** and title into **r4** before the call; the callee saves r4 as its title pointer and reads the context's owned-title count/array from **r3+0x28/r3+0x2C**. This is a two-parameter ABI, matching the existing C++ definition and emitted symbol. The NUP function is **not a new match**: it still has **163/163 instructions with 17 register-allocation differences**, and now receives its honest exact-name objdiff score of **99.35583%** instead of being unmapped. NUP's fuzzy unit score rises **93.442955 -> 99.46117% solely from attribution**; exact code remains **7,496/10,764**, exact functions **19/23**, and data **1,720/1,720**. ## Verification on 184bee01 - Only two names in `config/43U/symbols.txt` change; **no address, size, split, scope, source, or padding changes** - Rebuilt extraction, all_source, progress, report and DOL successfully - Compared allocated-section bytes and normalized relocations across **all 1,027 source objects**, including duplicate-named ELF sections: **zero changes** - All exact-code, exact-data, function-match and link measures unchanged; only the described NUP fuzzy attribution changes - pdm_disk ctxdiff **17/17 zero differences**; NUP pool **28/28 identical** - Workflow/pool/jump-table tests **7/8/27 pass**; diff and main/leaf worktree checks pass - DOL SHA1 unchanged: **26116613f624061ba99c8d1a299aaa6efa85670d** Tested/published tree `428c78ab4c8605abe94c3426d2d4ae944219fbbe`. Kept separate from source-matching PRs, as requested. Whole-project completion remains incomplete. <!-- Macroscope's pull request summary starts here --> <!-- Macroscope will only edit the content between these invisible markers, and the markers themselves will not be visible in the GitHub rendered markdown. --> <!-- If you delete either of the start / end markers from your PR's description, Macroscope will append its summary at the bottom of the description. --> > [!NOTE] > ### Fix two symbol identities in `config/43U/symbols.txt` without changing extents > Removes an erroneous trailing backtick from the disk block-to-sector symbol entry. Adds the missing context-information parameter to the boot-version symbol's recorded signature. Symbol extents are unchanged. > > <!-- Macroscope's review summary starts here --> > > <sup><a href="https://app.macroscope.com">Macroscope</a> summarized 8fbd3a7.</sup> > <!-- Macroscope's review summary ends here --> > <!-- Macroscope's pull request summary ends here -->
coah80 commented 2026-10-02 04:25:14 +00:00 (Migrated from github.com)

Reviewed and merged: rebased on main, full build ok, DOL sha1 26116613f624061ba99c8d1a299aaa6efa85670d, 0 regressions, 0 forbidden patterns.

Reviewed and merged: rebased on main, full build ok, DOL sha1 26116613f624061ba99c8d1a299aaa6efa85670d, 0 regressions, 0 forbidden patterns.
Sign in to join this conversation.