_umtx_op: document UMTX_OP_SEM2_WAIT copyout behavior
ClosedPublic
Actions

Authored by kevans on Nov 10 2020, 4:48 AM.

Details

Reviewers

kib
vangyzen
0mp

Group Reviewers

manpages

Commits

rS368325: MFC r367742-r367743: _umtx_op: documentation and compat32 fix
rS367742: _umtx_op: document UMTX_OP_SEM2_WAIT copyout behavior

Summary

I tripped over a librt test that exercises this under qemu-bsd-user, and was quite confused for a solid minute.

This clever technique to get a time remaining back was added to support sem_clockwait_np.

Diff Detail

Repository

rS FreeBSD src repository - subversion

Lint

Lint Not Applicable

Unit

Tests Not Applicable

Event Timeline

kevans created this revision.Nov 10 2020, 4:48 AM

Herald added a reviewer: manpages. · View Herald TranscriptNov 10 2020, 4:48 AM

Herald added a subscriber: imp. · View Herald Transcript

kevans requested review of this revision.Nov 10 2020, 4:48 AM

Harbormaster completed remote builds in B34720: Diff 79373.Nov 10 2020, 4:48 AM

If the content makes sense, it's good to go.

PS MFC maybe?

This revision is now accepted and ready to land.Nov 10 2020, 9:01 AM

I do not like how this description is structured.

The block of UMTX_OP_SEM2_WAIT starts with explanation of arguments, and currently it only mentions obj. I suggest to add that uaddr1 can take size of memory pointed to by uaddr2, and uaddr2 points to struct umtx_time32 followed by struct timespec.

Then if UMTX_ABSTIME is not set, <follow your description>.

Take 2; explain uaddr/uaddr2 up front and slightly wordsmith.

This revision now requires review to proceed.Nov 12 2020, 3:27 AM

Harbormaster completed remote builds in B34754: Diff 79454.Nov 12 2020, 3:27 AM

kib added inline comments.Nov 12 2020, 4:26 AM

lib/libc/sys/_umtx_op.2
1105 ↗	(On Diff #79454)	s/structure/memory/
1139 ↗	(On Diff #79454)	This makes impression that UMTX_ABSTIME can be set during operation. `... was not set, and the operation was interrupted ...`