wdgzyjx/myxce
1
0
Fork 0
forked from lei.chen/myxce
Code Pull requests Activity

chore: update OpenWolf session state (Group 1 validate-impl batch + bug-1447/1449)

Browse source
This commit is contained in:
Rongsong Shen 2026-07-18 13:28:10 +00:00
parent 426728bb98
commit 8f614b6421
6 changed files with 2822 additions and 350 deletions
Download patch file Download diff file Expand all files Collapse all files

103
.wolf/memory.md
View file

@ -5034,3 +5034,106 @@
| 07:35 | Scripted fix across 83 controller/testsuite/*.c group functions | UnitySetTestFile(__FILE__) added; 19 files' nested UNITY_BEGIN/UNITY_END removed | ~2100 |
| 07:36 | Verified: controller test suite 972 Tests 0 Failures 0 Ignored OK (single consolidated summary, matches ground-truth RUN_TEST count) | testlog.txt | ~50 |
| 07:37 | Updated .wolf/cerebrum.md | added Unity nested-BEGIN/END masking fix entry | ~450 |
## Session: 2026-07-18 17:02
| Time | Action | File(s) | Outcome | ~Tokens |
|------|--------|---------|---------|--------|
| 17:17 | Edited .kiro/specs/raft-sim-conformance/spec.json | expanded (+11 lines) | ~1609 |
| 17:19 | Session end: 1 writes across 1 files (spec.json) | 8 reads | ~22642 tok |
## Session: 2026-07-18 17:27
| Time | Action | File(s) | Outcome | ~Tokens |
|------|--------|---------|---------|--------|
| 17:28 | Edited .kiro/specs/libxce-failover/spec.json | expanded (+10 lines) | ~787 |
| 00:35 | Wrote GO decision to spec.json (validate_impl block) | .kiro/specs/libxce-failover/spec.json | ready_for_implementation false->true, phase tasks-generated->implementation-complete | ~450 |
| 17:29 | Session end: 1 writes across 1 files (spec.json) | 0 reads | ~787 tok |
| 17:36 | Edited .kiro/specs/ha-entity-snapshots/spec.json | modified commit() | ~1450 |
| 00:35 | /kiro-validate-impl ha-entity-snapshots: verified task 7.1's 07-02 blocker genuinely resolved upstream (raft-ha-core-safety 5.10a), built controller (192/192), full suite 962 PASS/0 FAIL, wrote GO decision to spec.json | .kiro/specs/ha-entity-snapshots/spec.json | ready_for_implementation stays true, phase tasks-generated->implementation-complete, cleaned up 331 session /tmp/xce_* dirs | ~14000 |
| 17:38 | Session end: 2 writes across 1 files (spec.json) | 9 reads | ~91160 tok |
| 00:56 | /kiro-validate-impl ha-commit-durability: build clean (0 :FAIL:/791 PASS), re-verified durable_lsn/CAS invariant intact despite heavy churn since 2026-07-02 (bug-1433/1434/1445), fresh live re-soak on ext4 (4-round rolling+cold-restart-all, 8/8 PASS 0 loss), diagnosed deterministic-mode false RED as BusyBox `date +%N` incompatibility (logged bug-1446), wrote GO to spec.json | .kiro/specs/ha-commit-durability/spec.json, .wolf/buglog.json | phase tasks-generated->implementation-complete, bug-1446 appended | ~55000 |
| 17:54 | Session end: 2 writes across 1 files (spec.json) | 15 reads | ~121365 tok |
## Session: 2026-07-18 18:01
| Time | Action | File(s) | Outcome | ~Tokens |
|------|--------|---------|---------|--------|
## Session: 2026-07-18 18:06
| Time | Action | File(s) | Outcome | ~Tokens |
|------|--------|---------|---------|--------|
## Session: 2026-07-18 18:13
| Time | Action | File(s) | Outcome | ~Tokens |
|------|--------|---------|---------|--------|
| 18:19 | Edited .kiro/specs/raft-drive-async/spec.json | added error handling | ~1860 |
| 01:20 | /kiro-validate-impl raft-drive-async: 10-round sequential HA soak crashed round 9/10 (ASAN SEGV, ssl3_read_bytes via netio.c handle_ssl_handshake, ctrl-3 died mid reconnect-offload burst); logged bug-1447 (NOT-FIXED, blocking, LOCAL to Task 2.1/2.2 reconnect offload); wrote NO-GO to spec.json validate_impl | .kiro/specs/raft-drive-async/spec.json, .wolf/buglog.json | NO-GO | ~1200 |
## Session: 2026-07-18 18:21
| Time | Action | File(s) | Outcome | ~Tokens |
|------|--------|---------|---------|--------|
| 18:35 | Edited .kiro/specs/ha-wal-replication/spec.json | added error handling | ~2214 |
| 02:10 | /kiro-validate-impl ha-wal-replication: ASAN build clean, controller_unit_test 972/972 PASS, cross-task wiring (repl_add_standby/repl_remove_standby/WalAckReply/repl_on_ack) confirmed by code read, 10-round sequential HA soak PASS (33 jobs, zero loss, all 3 nodes led); explicitly grepped for bug-1447's SEGV signature (raft-drive-async's sibling NO-GO from same session) — did NOT reproduce; wrote GO to spec.json validate_impl, phase advanced tasks-generated → implementation-complete | .kiro/specs/ha-wal-replication/spec.json | GO | ~2000 |
| 18:38 | Session end: 1 writes across 1 files (spec.json) | 8 reads | ~13024 tok |
| 18:40 | Session end: 1 writes across 1 files (spec.json) | 12 reads | ~24736 tok |
| 18:44 | Session end: 1 writes across 1 files (spec.json) | 16 reads | ~155010 tok |
| 18:57 | Session end: 1 writes across 1 files (spec.json) | 16 reads | ~155010 tok |
| 20:01 | Session end: 1 writes across 1 files (spec.json) | 16 reads | ~155010 tok |
| 20:12 | Session end: 1 writes across 1 files (spec.json) | 16 reads | ~155010 tok |
| 20:13 | Session end: 1 writes across 1 files (spec.json) | 16 reads | ~155010 tok |
| 21:07 | Session end: 1 writes across 1 files (spec.json) | 16 reads | ~155010 tok |
| 23:14 | Session end: 1 writes across 1 files (spec.json) | 16 reads | ~155010 tok |
| 01:26 | Session end: 1 writes across 1 files (spec.json) | 16 reads | ~155010 tok |
| 01:28 | Session end: 1 writes across 1 files (spec.json) | 16 reads | ~155010 tok |
| 02:12 | Session end: 1 writes across 1 files (spec.json) | 16 reads | ~155010 tok |
## Session: 2026-07-18 02:19
| Time | Action | File(s) | Outcome | ~Tokens |
|------|--------|---------|---------|--------|
| 02:28 | Edited .kiro/specs/ha-heartbeat-committed-propagation/spec.json | modified 0() | ~2754 |
| 09:28 | /kiro-validate-impl ha-heartbeat-committed-propagation: GO. 972 unit tests pass, 10/10 sequential HA soak clean (33 jobs, zero loss, all 3 nodes led), zero bug-1447 signature | .kiro/specs/ha-heartbeat-committed-propagation/spec.json | GO | ~180 |
| 02:30 | Session end: 1 writes across 1 files (spec.json) | 2 reads | ~7744 tok |
| 02:43 | Created .kiro/specs/ha-prevote/spec.json | — | ~3803 |
| 09:56 | /kiro-validate-impl ha-prevote: GO. 293 raft-lib + 972 controller unit tests pass (52 PreVote-specific), 10/10 sequential HA soak clean (33 jobs, zero loss, all 3 nodes led, live PreVote WON/DENIED/mid-round-cancel activity observed), zero bug-1447 signature at round 9 | .kiro/specs/ha-prevote/spec.json | GO | ~210 |
| 02:45 | Session end: 2 writes across 1 files (spec.json) | 10 reads | ~145243 tok |
## Session: 2026-07-18 02:57
| Time | Action | File(s) | Outcome | ~Tokens |
|------|--------|---------|---------|--------|
| 09:56 | kiro-validate-impl ha-elect-ctx-lifecycle: GO (13/13 tasks, 972/0 suite, leak-detect 5/5 detect_leaks=1, HA soak retry 10/10 clean after 1 harness-timing FAIL) | .kiro/specs/ha-elect-ctx-lifecycle/spec.json, .wolf/buglog.json (bug-1449) | GO, phase=implementation-complete | ~45k |
## Session: 2026-07-18 03:09
| Time | Action | File(s) | Outcome | ~Tokens |
|------|--------|---------|---------|--------|
| 04:58 | Edited .claude/skills/xce-test-ha/run-ha-soak.sh | modified node_config_tag() | ~727 |
| 04:59 | Edited .claude/skills/xce-test-ha/run-ha-soak.sh | expanded (+18 lines) | ~411 |
| 12:12 | Verified bug-1449 fix: 3x back-to-back `run-ha-soak.sh --sequential 6` on ext4 (/home/claude/tmp) | run-ha-soak.sh, .wolf/buglog.json, .wolf/anatomy.md, .wolf/cerebrum.md | PASS 3/3 (6 rounds/24 jobs/zero loss/all 3 nodes led each run); config-convergence gate confirmed before every round 1; cold-start timing: election 10.3-11.7s, config-push 40-46ms after election (ample margin over ~4s min dwell); cleaned up 3 deploy dirs+logs | ~9000 |
| 05:13 | Session end: 2 writes across 1 files (run-ha-soak.sh) | 2 reads | ~142893 tok |
| 05:23 | Edited src/lib/netio.c | modified XCE_TESTING() | ~800 |
| 05:23 | Edited src/lib/netio.c | modified io_worker_dispatch_batch() | ~117 |
| 05:23 | Edited src/include/netio.h | modified XCE_TESTING() | ~165 |
| 05:24 | Created src/lib/testsuite/test_netio_batch_dispatch_uaf.c | — | ~1879 |
| 05:24 | Edited src/lib/testsuite/test_netio_batch_dispatch_uaf.c | 2→1 lines | ~9 |
| 05:24 | Edited src/lib/meson.build | 2→3 lines | ~51 |
| 05:24 | Edited src/lib/testsuite/test.c | 2→3 lines | ~39 |
| 05:24 | Edited src/lib/testsuite/test.c | 2→3 lines | ~32 |
| 05:25 | Edited src/lib/netio.c | modified io_worker_dispatch_batch_for_test() | ~214 |
| 05:26 | Edited src/include/netio.h | modified in() | ~150 |
| 05:26 | Edited src/lib/netio.c | modified io_worker_dispatch_batch() | ~95 |
| 05:30 | Session end: 13 writes across 6 files (run-ha-soak.sh, netio.c, netio.h, test_netio_batch_dispatch_uaf.c, meson.build) | 8 reads | ~167605 tok |
| 06:11 | Session end: 13 writes across 6 files (run-ha-soak.sh, netio.c, netio.h, test_netio_batch_dispatch_uaf.c, meson.build) | 8 reads | ~167605 tok |
| 06:13 | Session end: 13 writes across 6 files (run-ha-soak.sh, netio.c, netio.h, test_netio_batch_dispatch_uaf.c, meson.build) | 8 reads | ~167605 tok |
| 06:19 | Session end: 13 writes across 6 files (run-ha-soak.sh, netio.c, netio.h, test_netio_batch_dispatch_uaf.c, meson.build) | 8 reads | ~167605 tok |
| 12:26 | bug-1447 root-caused: NOT two io_worker threads (only one exists per iomgr) — same-batch UAF: on_standby_connected's endpoint_unref on a stale duplicate peer ep frees it while a LATER event in the SAME epoll_wait() batch still points to it | src/lib/netio.c, src/controller/ha.c (read only) | root cause confirmed via code reading, not yet fixed | ~40k |
| 12:26 | RED test established: io_worker_dispatch_batch (pure refactor, no fix yet) + io_worker_dispatch_batch_for_test seam + test_netio_batch_dispatch_uaf.c; ran standalone -> ASAN heap-use-after-free at netio.c:587, freed at netio.c:843 via injected evictor handler | src/lib/netio.c, src/include/netio.h, src/lib/testsuite/test_netio_batch_dispatch_uaf.c, src/lib/meson.build, src/lib/testsuite/test.c | RED confirmed for the right reason | ~15k |
| 12:28 | Fix applied: endpoint_ref/endpoint_unref bracket around the whole io_worker batch dispatch (ref all before dispatch, unref all after) | src/lib/netio.c | GREEN: test passes, full lib suite 294/0/0, controller suite 972/0/0, zero ASAN hits | ~20k |
| 13:00-13:20 | Live verification: 2x independent fresh sequential 10-round /xce-test-ha soaks on real ext4 (~/tmp), both fully clean, zero crash-signature matches | ~/tmp/xce-bug1447-verify-1 (33 jobs), ~/tmp/xce-bug1447-verify-2 (36 jobs, incl. forced ctrl-3 election) | PASS 2/2, all 3 nodes led each run | ~30k |
| 13:24 | Updated .wolf/buglog.json bug-1447 (root_cause + fix rewritten, tags FIXED); updated .kiro/specs/raft-drive-async/spec.json validate_impl NO-GO->GO with full evidence; cleaned up both soak dirs + stray logs | .wolf/buglog.json, .kiro/specs/raft-drive-async/spec.json | bug-1447 FIXED, raft-drive-async GO | ~12k |
| 06:24 | Session end: 13 writes across 6 files (run-ha-soak.sh, netio.c, netio.h, test_netio_batch_dispatch_uaf.c, meson.build) | 11 reads | ~190470 tok |