Rongsong Shen
d26ff8870d
The buglog was one 27K-line JSON file whose fields covered only symptom,
root cause and fix, as free text: 2% of entries named the code version a
bug was found on, 12% the build profile, 36% a way to reproduce, and 58% of
the file was OpenWolf hook auto-detections ("Missing guard clause",
"Function not marked async" on .md files).
- .claude/skills/buglog/bug-record.md: the record standard. Eight
sections (Summary as the title, Description, Environment, Reproduction;
Root cause, Fix, Verification, References once fixed), each with the
content it must contain -- e.g. Environment needs the OS, the libc the
failing binary loads (from its ELF interpreter: glibc 2.42 in the nix
shell vs 2.40 on the host), toolchain, build profile and commit.
- .wolf/bugs/<id>.md: 500 records migrated from buglog.json; every field
verified to round-trip. They carry legacy: true and an explicit "not
recorded" where the old log had no such field; component derived from
file paths, severity from keywords (else unknown), fix_commits from
hashes in status_note/fix that git resolves (91 of 463 fixed).
- .wolf/buglog-archive.json: the 710 hook auto-detections, verbatim.
- .wolf/buglog.json is now an index generated from the records, keeping
the fields OpenWolf's hooks and `openwolf bug search` read.
- scripts/buglog.py rewritten: new (template + Environment collected from
the machine and --binary), check, status (refuses a status the record
doesn't support; fixed needs Root cause, Fix, Verification and
--commit hashes git resolves), note, recur, set, show, list, triage,
env, index. 29 unittest cases in scripts/tests/test_buglog.py.
- scripts/git-hooks/pre-commit runs `check --staged`: incomplete staged
records, a legacy record hand-changed to fixed without the fix-time
sections, a new record claiming legacy: true, or a stale index refuse
the commit.
- scripts/gitea-sync.py reads the records: issue body = the record's
sections plus linked fix commits (origin remote, GITEA_COMMIT_REPO to
override), and a body hash so an issue's text is refreshed when the
record changes.
- bug-1866 rewritten to the full standard as the worked example.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2.8 KiB
| id | status | status_updated | status_note | component | severity | timestamp | last_seen | occurrences | files | tags | related | legacy | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| bug-1794 | fixed | 2026-09-24 | controller | crash | 2026-09-09T19:03:51Z | 2026-09-19T09:00:00Z | 1 |
|
|
|
true |
bug-1794: xce_test cmd_jobrun crashes (ASAN allocation-size-too-big abort in alloc_packet, requested size…
Description
xce_test cmd_jobrun crashes (ASAN allocation-size-too-big abort in alloc_packet, requested size ~0xffffffffc3c3c4b0) when run live against a real controller, after xce-test-cli-ux task 3.2 wrapped its xcecli_jobdispatch() call through the interruptible-call adapter (heap-allocated ctx block instead of a stack-local xce_job_alloc_t).
Environment
Not recorded (migrated from buglog.json).
Reproduction
Not recorded (migrated from buglog.json).
Root cause
PRE-EXISTING, latent: fill_job_alloc() (test.c) only ever sets xce_job_alloc_t.nhosts/.allocs -- it never initializes the struct's other two fields, nres_shared and allocs_shared (xce/types.h). Before task 3.2, xce_job_alloc_t jalloc was a plain stack-local in cmd_jobrun, so those two never-set fields held whatever incidental value that stack frame happened to have -- harmless in practice, apparently usually small/zero. Task 3.2 moved this struct into a heap-allocated jobrun_call_t block (required by the interruptible-call adapter's contract: fn must only touch heap-owned memory, never the caller's stack, since an abandoned call's worker thread can outlive the caller's own stack frame) allocated via plain malloc(), which leaves ALL bytes as genuine heap garbage -- confirmed live (real single-node deploy, jobrun -j -m worker-1:1): the garbage nres_shared/allocs_shared corrupted xce_jobdispatch_req's wire-size computation in lib/request.c, producing a multi-exabyte allocation request that ASAN correctly aborted on.
Fix
FIXED (2026-09-09, xce-test-cli-ux task 3.2). Every per-call-site heap block (all ~26 _call_t typedefs test.c's interrupt-wrapping introduces) now allocates via calloc(1, sizeof(*call)) instead of malloc(sizeof(*call)), zero-initializing every field uniformly rather than auditing each wrapped request struct for a similarly-never-fully-set field one at a time -- the same defensive default an accidentally-zero stack frame was already silently relied on for. Verified by re-running the exact live repro (jobrun against a real single-node deploy) after the fix: completes normally with no ASAN error.
Verification
Not recorded (migrated from buglog.json).
References
- Related: bug-1791