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

feat(xce-test-cli-ux): build interruptible-call heap-ownership mechanism

Browse source 
xcetest_run_interruptible() runs a blocking call on a detached worker
thread while the caller polls xcetest_interrupted() every 50ms. A single
atomic CAS (RUNNING -> DONE_UNCLAIMED | ABANDONED) decides which side
frees the heap-allocated call block, so an abandoned call's worker
thread never double-frees or leaks once it eventually finishes. Proven
under active ASAN with a real detached thread and a real SIGINT-forced
abandonment. 1282/1282 controller suite passing (+2).

Task 1.4 of .kiro/specs/xce-test-cli-ux/tasks.md.
Requirements: 1.1, 1.2, 1.3, 1.4

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019jaMf9QEwpugH2W9KAuc7f
This commit is contained in:
Rongsong Shen 2026-09-09 08:29:14 +00:00
parent b4f5a460fc
commit 9a06047da2
No known key found for this signature in database
GPG key ID: 5BCDAAE0123E9FE2
6 changed files with 467 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

3
.kiro/specs/xce-test-cli-ux/tasks.md
View file

@ -23,7 +23,7 @@
- Observable completion: a small standalone test that sends an interrupt signal to the process during a blocking read confirms the read returns immediately with an interrupted result rather than continuing to block
- _Requirements: 9.9_
- [ ] 1.4 Build the interruptible-call adapter's heap-block-ownership mechanism
- [x] 1.4 Build the interruptible-call adapter's heap-block-ownership mechanism
- Implement a wrapper that runs a given blocking call on a detached worker thread while the caller polls the interruption accessor from 1.3 on a short interval
- Implement the atomic ownership handoff so that whichever side (the polling caller or the worker) is left holding the call's heap-allocated state is the one that frees it, with no path that frees it from both sides
- Observable completion: the unit test suite shows a normal (uninterrupted) call returning its real result and freeing cleanly, and a forced-timeout call returning immediately with no double-free and no leak (verified under the project's leak/ASAN tooling)
@ -178,4 +178,5 @@
## Implementation Notes
- `src/tools/xce-build.sh controller asan xce.build --incremental` writes its build dir at `src/xce.build/` (relative to the script's own location), not a repo-root `xce.build/` — meson logs are at `src/xce.build/meson-logs/testlog.txt`. (task 1.1)
- New Unity test files need two wiring points, not just meson.build's source list: also register an `extern void test_<module>(void);` + `TEST(test_<module>);` in `src/controller/testsuite/controller_test.c`, mirroring the existing `test_rcfile` registration, or the suite silently never runs. (task 1.1)
- `interruptcall.c`'s forced-abandonment test proves `-EINTR` via return-value disambiguation against a slow `fn`, not a direct wall-clock elapsed-time assertion — sound today (poll interval is 50ms, matching design.md) but a future regression widening the poll interval wouldn't be caught unless the test also asserts an elapsed-time bound. Non-blocking; noted for task 4.1's end-to-end Ctrl-C validation to consider. (task 1.4)
- After task 1.3's SIGINT/EINTR fix alone (before task 3.2 wires the REPL loop to check `xcetest_interrupted()`), a real Ctrl-C at `xce_test`'s top-level interactive prompt now causes the process to EXIT (readline's EINTR-driven NULL return is currently treated the same as EOF) rather than gracefully redisplaying the prompt. This is expected/deferred, not a regression — no current test exercises that path — but task 3.2's implementer must fix it as part of wiring the REPL loop, not just the RPC call sites. (task 1.3)