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>
2.2 KiB
| id | status | status_updated | status_note | component | severity | timestamp | last_seen | occurrences | files | tags | related | legacy | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| bug-562 | fixed | 2026-09-24 | lib | crash | 2026-06-14T23:55:00Z | 2026-09-19T09:00:00Z | 1 |
|
|
true |
bug-562: AddressSanitizer: heap-use-after-free in encode_RequestReply / strlen…
Description
AddressSanitizer: heap-use-after-free in encode_RequestReply / strlen (lib/controller.capnp.c:13168) on ctrl-2 startup, via xce_reply->xce_encode from OPCode_nop handler (cluster.c:307). Also: job submission succeeds but joblist shows no jobs.
Environment
Not recorded (migrated from buglog.json).
Reproduction
Not recorded (migrated from buglog.json).
Root cause
init_reply_by_req() only set seq+opcode, leaving the rest of the stack-declared xce_reply_t uninitialized. The Cap'n-C generated encode_RequestReply reads the common trailer field redirect_host UNCONDITIONALLY (gated only on != NULL), so a garbage stack pointer in reply.redirect_host was strlen'd. The garbage aliased a 4KB capnp segment freed moments earlier by the WAL-ack reply encode on the same thread -> UAF. Handlers that build a reply via init_reply_by_req and only set status/data_which/data (e.g. the NOP/state-report path) never cleared redirect_host. Empty joblist is downstream: ctrl-2 crash-looping on this UAF destabilizes replication/quorum, so a get request that fails over to the not-caught-up standby reads an empty jobmgr (standby serves get from its own committed state).
Fix
memset(reply,0,sizeof(*reply)) at the top of init_reply_by_req() (src/lib/request.c) before setting seq/opcode. All 30+ reply sites follow declare->init->populate, so zeroing in the shared initializer fixes the whole class (redirect_host/redirect_port/status/data garbage) at the source. Added regression test test_init_reply_by_req_zeroes_reply in src/lib/testsuite/test_request.c (poisons reply with 0xFF, asserts redirect_host==NULL etc).
Verification
Not recorded (migrated from buglog.json).
References
Not recorded (migrated from buglog.json).