Mihawk a0a4f9df51 Watch the kernel log, launch the title, and record why it crashes ...

The console's PPSA99005 was launched with its kernel log captured. It does not
start, and the cause is measured rather than guessed: its eboot.bin was a
link-stage ELF (51,870,448 bytes), not the converted application image the loader
requires, so it faulted before main() ran - SIGSEGV, page fault at address 0x1,
thread "eboot.bin", /app0/sce_module/libc.prx named in the crash block.

Evidence: evidence/ppsa-99005-startup-crash/ holds the distilled record and the
raw capture (klog/ppsa99005-122733.log, 214 lines); tools/evidence.py compare
replays both records cleanly.

Two things are recorded as findings rather than smoothed over:
- the installed image was not converted, which is what tools/install-title.sh
  exists to prevent: it installs only the converted image, verifies the stored
  bytes byte-for-byte, and keeps the previous one as eboot.bin.previous;
- the /data/homebrew/PPSA99005/ directory disappeared during that replace, with
  every step reporting success. The neighbouring folders were untouched, so an
  install here must write a whole folder and verify it rather than patch a file
  inside one.

tools/evidence.py also gains a correction: a raw capture that has aged out of the
ignored klog/ tree is reported, not treated as a failure, because the distilled
record is the evidence and a cleanup must not turn a gate red.

2026-09-18 12:30:29 -04:00

..

m1-baseline-loads

Prove the Option 1 baseline on the console

2026-09-18 12:08:41 -04:00

ppsa-99005-startup-crash

Watch the kernel log, launch the title, and record why it crashes

2026-09-18 12:30:29 -04:00