Bug 220398
| Summary: | lio_listio(2) never sends signal if nent==0 | ||||||
|---|---|---|---|---|---|---|---|
| Product: | Base System | Reporter: | Alan Somers <asomers> | ||||
| Component: | kern | Assignee: | Alan Somers <asomers> | ||||
| Status: | Closed FIXED | ||||||
| Severity: | Affects Some People | CC: | emaste, jhb | ||||
| Priority: | --- | ||||||
| Version: | CURRENT | ||||||
| Hardware: | Any | ||||||
| OS: | Any | ||||||
| Attachments: |
|
||||||
A commit references this bug: Author: asomers Date: Mon Jul 17 18:33:31 UTC 2017 New revision: 321082 URL: https://svnweb.freebsd.org/changeset/base/321082 Log: Add regression tests for bugs 220459 and 220398 Bug 220398 - lio_listio(2) never sends asynchronous notification if nent==0 Bug 220459 - lio_listio(2) doesn't support SIGEV_THREAD PR: 220459 PR: 220398 Reviewed by: cem, jhb MFC after: 3 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D11470 Changes: head/tests/sys/aio/Makefile head/tests/sys/aio/lio_test.c A commit references this bug: Author: ngie Date: Wed Aug 2 09:49:41 UTC 2017 New revision: 321918 URL: https://svnweb.freebsd.org/changeset/base/321918 Log: Fix cosmetic issue with error message Add missing space in error message related to PR noted. MFC after: 2 weeks PR: 220398 Changes: head/tests/sys/aio/lio_test.c A commit references this bug: Author: asomers Date: Fri Aug 25 14:37:23 UTC 2017 New revision: 322891 URL: https://svnweb.freebsd.org/changeset/base/322891 Log: MFC r321082: Add regression tests for bugs 220459 and 220398 Bug 220398 - lio_listio(2) never sends asynchronous notification if nent==0 Bug 220459 - lio_listio(2) doesn't support SIGEV_THREAD PR: 220459 PR: 220398 Reviewed by: cem, jhb Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D11470 Changes: _U stable/11/ stable/11/tests/sys/aio/Makefile stable/11/tests/sys/aio/lio_test.c A commit references this bug: Author: kib Date: Tue Dec 1 22:53:34 UTC 2020 New revision: 368265 URL: https://svnweb.freebsd.org/changeset/base/368265 Log: lio_listio(2): send signal even if number of jobs is zero. Right now, if lio registered zero jobs, syscall frees lio job structure, cleaning up queued ksi. As result, the realtime signal is dequeued and never delivered. Fix it by allowing sendsig() to copy ksi when job count is zero. PR: 220398 Reported and reviewed by: asomers Sponsored by: The FreeBSD Foundation MFC after: 1 week Differential revision: https://reviews.freebsd.org/D27421 Changes: head/sys/kern/vfs_aio.c Changing the bug title, because the lack of signal notification and lack of kqueue notification turned out to be separate bugs. Kib's commit only fixed the former. I'll open a separate bug for kqueue notification. A commit references this bug: Author: asomers Date: Wed Dec 2 00:53:03 UTC 2020 New revision: 368272 URL: https://svnweb.freebsd.org/changeset/base/368272 Log: AIO tests: update expected failure messages after r368265 PR: 220398, 251515 MFC after: 1 week MFC-With: r368265 Changes: head/tests/sys/aio/lio_test.c A commit in branch stable/12 references this bug: URL: https://cgit.FreeBSD.org/src/commit/?id=0efb22eebbbbd5f164cab39dca72a78cce82afe3 commit 0efb22eebbbbd5f164cab39dca72a78cce82afe3 Author: Alan Somers <asomers@FreeBSD.org> AuthorDate: 2020-01-04 18:59:46 +0000 Commit: Alan Somers <asomers@FreeBSD.org> CommitDate: 2020-12-30 03:02:47 +0000 MFC r356352, r368272 lio_listio_empty_nowait_thread sometimes does *not* hang. The other tests consistently do hang though. Sponsored by: DellEMC (cherry picked from commit 0cb4586a0be3b1b5909b73b24798032ddf2f4002) AIO tests: update expected failure messages after r368265 PR: 220398, 251515 (cherry picked from commit 23693bd8f3e36b4721d7b59d52154d3e3f49e3cd) tests/sys/aio/lio_test.c | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-) |
Created attachment 183968 [details] Reproduction program It may seem stupid, but you can call lio_listio(2) with an empty aiocb list. I would naively expect lio_listio to do nothing and return in that case. Indeed, that's what it does when mode == LIO_WAIT. However, if mode == LIO_NOWAIT, then asynchronous notifications never get delivered. With SIGEV_KEVENT, no kevent is ever delivered, and with SIGEV_SIGNAL, no signal is ever delivered. I haven't checked other SIGEV_ values. The attached program demonstrates the problem: it calls lio_listio with LIO_NOWAIT and notification via SIGTERM. But no signal ever gets delivered, so the program exits normally. jhb feels that the Open Group's specification is ambiguous on this point, but that sending asynchronous notification immediately would be sensible.