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>
3.3 KiB
| id | status | status_updated | status_note | component | severity | timestamp | last_seen | occurrences | files | tags | related | legacy | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| bug-1574 | open | 2026-09-24 | xce_test resupdate string/type crash: set_resval() lacks svalue case, vals[] uninitialized (src/harness/test.c) | controller | crash | 2026-08-17T18:55:00Z | 2026-09-19T09:00:00Z | 1 |
|
|
|
true |
bug-1574: xce_test segfaults (SIGSEGV, exit 139) immediately on 'resupdate -n model -v ' (and by the…
Description
xce_test segfaults (SIGSEGV, exit 139) immediately on 'resupdate -n model -v ' (and by the same code path, -n type) run in the worker context against a live controller -- no output printed before the crash
Environment
Not recorded (migrated from buglog.json).
Reproduction
Not recorded (migrated from buglog.json).
Root cause
Pre-existing, predates the resource-value-verification feature entirely (confirmed via git show on the pre-task-2.1 baseline commit 98b8c34b^: the original set_resval()'s switch(valtype) already had no case for Value_u_svalue, same shape as today). set_resval() only assigns vals[idx].name inside a case block of switch(valtype); Value_u_svalue (type/model) has no case, so for these two names vals[idx].name is never assigned and keeps whatever uninitialized garbage was already on the local stack array xce_resval_t vals[MAX_N_RES] (never zero-initialized) in cmd_resupdate(). That garbage pointer is then passed unconditionally into xcecli_update_resvals()'s wire encoder inside &valptrs[0], which (most likely) dereferences it as a C string, causing the segfault. Task 2.1's own comment at test.c (near set_resval, 'the switch falls through with vals[idx].name/.val->which left unset ... intentionally not fixed by this task, see CONCERNS') already flagged this exact gap and deliberately preserved it byte-for-byte per Requirement 1.4, rather than introducing it. Discovered live during resource-value-verification task 4.2's validation (apiserver trace-log addition, live-testing the string-value code path with resupdate -n model -v CustomModelX) -- unrelated to task 4.2's own change (src/frontend/apiserver/backend/handler.go), which only touches Go/apiserver code.
Fix
NOT FIXED -- out of boundary for task 4.2 (scope: src/frontend/apiserver/backend/handler.go only). Belongs to test.c/set_resval()/resval_typeresolve.c territory (tasks 2.1/2.2 of this same spec, or a dedicated follow-up). Worked around during task 4.2 live validation by using a numeric resource (r15m) for the primary trace-log confirmation and avoiding type/model live resupdate calls thereafter. Likely fix: zero-initialize xce_resval_t vals[MAX_N_RES] in cmd_resupdate() (e.g. = {0}) so an unassigned .name is NULL instead of garbage, PLUS give set_resval() a real Value_u_svalue case that sets vals[idx].name and vals[idx].val->u.svalue (currently the string value passed to resupdate for type/model is silently never stored in the union either -- a second, related but distinct defect covered by the same missing switch case).
Verification
Not recorded (migrated from buglog.json).
References
- Related: bug-1570, bug-1573