Page MenuHomeFreeBSD

Measure latency of PMC interruptions
ClosedPublic

Authored by wma on Jul 23 2021, 8:09 AM.
Tags
None
Referenced Files
Unknown Object (File)
Mon, Nov 10, 12:32 AM
Unknown Object (File)
Fri, Nov 7, 5:30 AM
Unknown Object (File)
Thu, Nov 6, 5:44 AM
Unknown Object (File)
Wed, Nov 5, 11:10 PM
Unknown Object (File)
Wed, Nov 5, 10:42 PM
Unknown Object (File)
Sun, Nov 2, 5:25 AM
Unknown Object (File)
Sat, Nov 1, 5:39 AM
Unknown Object (File)
Fri, Oct 31, 5:27 AM
Subscribers

Diff Detail

Lint
Lint Skipped
Unit
Tests Skipped

Event Timeline

wma requested review of this revision.Jul 23 2021, 8:09 AM
wma created this revision.
wma added a reviewer: kib.
sys/kern/kern_intr.c
79

I suggest to have it_waiting defined always, the effect on memory usage is minuscule.

80

What is the locking protocol for it_waiting? Is it locked by it_thread thread lock?

1031

Can you explain the desired semantic for intr waiting_5 soft event?

Good point about locking. I modified the code to use atomics instead.

Also, removed waiting_5 and let the user to choose the threshold when the PMC event is triggered (default = 1).

In D31283#714889, @wma wrote:

Any further comments?

New soft events should be documented in the pmc.soft(3) man page.

sys/kern/kern_intr.c
1465

Why not pass frame here?

This revision was not accepted when it landed; it landed in state Needs Review.Sep 13 2021, 4:12 AM
This revision was automatically updated to reflect the committed changes.