lei.chen/myxce
1
0
Fork 1
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

fix(job-array-single-wal-record): close bug-1443's bulk-insert partial-visibility gap

Browse source 
restore_job_array()'s bulk-insert pass could fail partway through
(key_add() on realloc growth exhaustion), leaving an already-inserted
prefix permanently visible in c->jobmgr but excluded from
out_new_jobs -- so those elements existed and were individually
addressable but never reached HOOK_DETAIL_JOB_NEW, never scheduled or
pushed, not on this attempt or any later successful retry. A literal
violation of Requirements 2.3 and 9.4, tracked as bug-1443.

Adds kvmgr_reserve()/xzarray_reserve(), new purely-additive primitives
in shared kvmgr.c/zarray.c (following the existing kvmgr_capture_keys()
template -- bare extern function, no RTTI vtable slot, caller-holds-
write-lock contract), that pre-grow c->jobmgr's backing store before
the bulk-insert pass begins, closing the dominant N-scaling failure
trigger structurally.

Found a genuine off-by-one in the existing, unmodified
xzarray_ensure_items(): its strict less-than sufficiency check means
reserving exactly array_size and then performing array_size real
stores still triggers one more growth cycle on the last store.
Reserving array_size + 1 is the exact minimal correct compensation,
independently re-derived and confirmed by review with concrete
arithmetic.

For the now-much-narrower residual case (kvtab_add()'s own small
per-entry allocation, or a content collision -- both non-N-scaling,
honestly documented as still theoretically possible): restore_job_array()
now populates out_new_jobs with whatever prefix was genuinely inserted
before a failure, and apply_committed_entry() fires HOOK_DETAIL_JOB_NEW
for that non-empty partial prefix even when the overall return is
negative, while still propagating the negative rc so the LSN correctly
halts and the Live Apply Resource Fence still engages for the
remainder. A genuinely-inserted element is never left un-pushed again.

Task 7.1 of job-array-single-wal-record (Requirements 2.3, 9.4, 9.6).
This commit is contained in:
Rongsong Shen 2026-07-16 15:08:02 +00:00
parent 575f928bbc
commit dcf9f39c22
13 changed files with 910 additions and 36 deletions
Download patch file Download diff file Expand all files Collapse all files

36
src/controller/persist.c
View file

@ -2820,16 +2820,44 @@ int apply_committed_entry(persist_t* p, uint32_t wal_type, uint64_t lsn,
* and only, what the hook fire below carries — Req 3.4); -EEXIST =
* every index in the record already existed (idempotent
* whole-record no-op, Req 3.3) — no hook; a negative errno = a
* genuine construction/resource-exhaustion failure (Req 9.4) — no
* hook, propagated so persist_apply_pump_run() (task 4.1) can halt
* and fence on it rather than silently skipping it. */
* genuine construction/resource-exhaustion failure (Req 9.4) —
* propagated so persist_apply_pump_run() (task 4.1) can halt and
* fence on it rather than silently skipping it. **Task 7.1
* (bug-1443) update**: a negative return no longer unconditionally
* means "no hook" — restore_job_array()'s bulk-insert-phase residual
* gap can leave a genuinely, permanently inserted prefix in
* out_new_jobs even on failure (job.c's doc comment, "Residual
* atomicity gap"); see below. */
xlist_t out_new_jobs;
memset(&out_new_jobs, 0, sizeof(out_new_jobs));
int rc = restore_job_array(c, payload, len, lsn, &out_new_jobs);
if (rc == -EEXIST)
return 0; /* idempotent re-apply (replay / resync) — no hook */
if (rc < 0)
if (rc < 0) {
/* job-array-single-wal-record task 7.1 (bug-1443 partial-prefix
* fix): out_new_jobs is untouched (count == 0, from the memset
* above) for a reservation or scratch-build-phase failure —
* nothing was ever inserted, so no hook fires, unchanged from
* before this task. But for the narrower bulk-insert-phase
* residual gap, out_new_jobs now carries a non-empty, genuinely
* and permanently inserted prefix (job.c) that must still get
* its one HOOK_DETAIL_JOB_NEW fire — a job that is really in
* c->jobmgr must never be silently un-notified. The overall
* apply is still, correctly, reported as failed via the
* negative rc returned below, so persist_apply_pump_run() halts
* the LSN and the Live Apply Resource Fence engages for the
* un-constructed remainder (Req 2.3, 9.4, 9.6). */
if (out_new_jobs.count > 0) {
t_apply_seam_notifying = true;
xce_run_hooks(&c->hooksys, XCE_CONTROLLER_HOOK_JOB,
HOOK_AFTER_EVENT, HOOK_DETAIL_JOB_NEW, c,
&out_new_jobs);
t_apply_seam_notifying = false;
}
xlist_fini(&out_new_jobs);
return rc;
}
/* Exactly ONE fire carrying every newly created element in one
* list — never once per element (Req 3.4). out_new_jobs is already