Make p1003_1b.aio_listio_max a tunable
ClosedPublic
Actions

Authored by asomers on Jul 14 2017, 4:31 PM.

Details

Reviewers

jhb
kib

Commits

rS326322: MFC r322258, r324941, r324956, r325018
rS322258: Make p1003_1b.aio_listio_max a tunable

Summary

Make AIO_LISTIO_MAX a tunable

p1003_1b.aio_listio_max is now a tunable. Its value is reflected in the
sysctl of the same name, and the sysconf(3) variable _SC_AIO_LISTIO_MAX.
Its value will be bounded from below by the compile-time constant
AIO_LISTIO_MAX and from above by the compile-time constant
MAX_AIO_QUEUE_PER_PROC and the tunable vfs.aio.max_aio_queue.

Test Plan

sys/aio/lio_kqueue_test with different values for IOCBS_PER_LIO and
p1003_1b.aio_listio_max

Diff Detail

Repository

rS FreeBSD src repository - subversion

Lint

Lint Not Applicable

Unit

Tests Not Applicable

Event Timeline

asomers created this revision.Jul 14 2017, 4:31 PM

Herald added a subscriber: imp. · View Herald TranscriptJul 14 2017, 4:31 PM

Harbormaster completed remote builds in B10500: Diff 30790.Jul 14 2017, 4:31 PM

jhb added inline comments.Jul 17 2017, 5:17 PM

sys/kern/posix4_mib.c
80 ↗	(On Diff #30790)	Even though this is not enabled, probably best to still add the macro here as well. Eventually we should probably just remove the disabled code.
sys/kern/vfs_aio.c
393 ↗	(On Diff #30790)	Blank line before comment (bde@ would probably say the comment is banal and can just be removed)
418 ↗	(On Diff #30790)	Add missing spaces around '*' while here.
1955 ↗	(On Diff #30790)	I'd rather make aio_listio_max a static global variable in vfs_aio.c and not call p31b_getcfg() in all these places.
2161 ↗	(On Diff #30790)	Extra blank line?

Store aio_listio_max as a static variable in vfs_aio.c instead of in
posix4_mib.c's facility array.

Harbormaster completed remote builds in B10536: Diff 30885.Jul 17 2017, 7:14 PM

jhb added inline comments.Jul 17 2017, 9:50 PM

sys/kern/vfs_aio.c
172 ↗	(On Diff #30885)	Hmm, I think I do like this approach and would be a fan of moving all the AIO-related MIBs into this file. However, I think it would be useful to check in with kib@ and possibly bde@ before moving away from p1b_*().