Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 29 Jun 2022 21:14:03 +0000
From:      bugzilla-noreply@freebsd.org
To:        bugs@FreeBSD.org
Subject:   [Bug 264867] kevent EVFILT_TIMER, timer expiration with twice the period
Message-ID:  <bug-264867-227-74oohqE2AS@https.bugs.freebsd.org/bugzilla/>
In-Reply-To: <bug-264867-227@https.bugs.freebsd.org/bugzilla/>
References:  <bug-264867-227@https.bugs.freebsd.org/bugzilla/>

next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D264867

--- Comment #6 from Mark Johnston <markj@FreeBSD.org> ---
To expand on my question a bit, I think there is a race somewhere in the
eventtimer code which occurs when a callout is scheduled on a different CPU=
.=20
In particular, when an EVFILT_TIMER event is created, it is "pinned" to the
current CPU, and all future callouts will fire on the CPU.  I am not sure w=
hy
kevent does this, it doesn't make much sense to me.

If I change filt_timertouch() to set kc->cpuid =3D PCPU_GET(cpuid) each tim=
e the
event is re-added, then the bug is much harder to trigger.  Alternately, if=
 the
test is changed so that each thread pins itself to a different CPU, then the
problem is not reproducible for me.

(In reply to felice.tufo from comment #4)
Does your test module specify the target CPU when arming a callout?

--=20
You are receiving this mail because:
You are the assignee for the bug.=



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-264867-227-74oohqE2AS>