Skip to content

Cleanup of plan5 follow-ups (branch claude/cleanup-small-2026-10-24-plan6)

Status: all verified on an Apple M5 Mac (make -C runtime test test-services, make -C gfx test: 0 failures).

  • hz_result.h: HZ_MOD_AUDIO (153; HZ_MOD_AUDIO_IN is now an alias) and HZ_R_AUDREC_BUFFER_COUNT_LIMIT / HZ_R_AUDREC_BUFFER_NOT_FOUND; audrec.c uses them. Values unchanged (still [unverified]). audin.c is not mine and keeps HZ_MOD_AUDIO_IN.
  • scan.py: the NVN regexes accept NVN_NOPROTO(. The scan reports NVN 523/523 implemented, 0 missing (no snapshot saved). The earlier "101 missing" was already gone because those entries moved to nvn_protos_ext.h and use plain NVN(; the regex change is future-proofing.
  • docs/runtime-services.md: aoc:u table row updated (DLC is mounted via fsp-srv cmd 30).
  • New runtime/tests/test_svc_fs_dataid.c (12 checks): cmd 30 opens a DLC romfs storage (GetSize, Read), missing romfs / unknown id / app id fail with NOT_FOUND_DATA.
  • gfx/tests/t_getters_linux.sh delegates to the Makefile target on macOS. t_getters_linux.c was already built and run by make -C gfx test (33 checks pass), so nothing to add there.
  • maxwell_translate.c: F2F F16 scratch buffer h 1100 -> 2200 bytes so gcc's -Wformat-truncation cannot trigger (clang never warned). hid.c: removed unused ret_u32_0, ret_u8_0.