Date: Sun, 9 Feb 2003 00:40:40 -0800 (PST) From: Julian Elischer <julian@elischer.org> To: Tim Robbins <tjr@FreeBSD.org> Cc: Bruce Evans <bde@zeta.org.au>, Julian Elischer <julian@FreeBSD.org>, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/sys proc.h src/sys/kern kern_clock.c Message-ID: <Pine.BSF.4.21.0302090036030.31703-100000@InterJet.elischer.org> In-Reply-To: <20030209192134.A10064@dilbert.robbins.dropbear.id.au>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 9 Feb 2003, Tim Robbins wrote:
> On Sat, Feb 08, 2003 at 10:06:46PM -0800, Julian Elischer wrote:
>
> [...]
> > So it turns out the code is correct.
>
> Thanks for the explanation. It seems reasonable now. My other concern
> is protecting addupc_{intr,task}() from profil() and also protecting
> addupc_task() from itself.
There is LOTS of room for improvement.
We are just trying to get thing working in a threaded world.
>
> There's an splstatclock() call in profil() which was presumably there to
> stop profil() from interfering with p_stats->p_prof while addupc_* were
> running. Maybe we could have a spinlock in struct uprof to synchronise
> access to it...
that is one possibilty.
>
> I'm not entirely comfortable about this piece of code in the presence
> of multiple threads, either. It looks like it could lose ticks.
yes, on an MP machine it can.
David and I discussed this but we decided that the problem is not high
enough for us to work on yet.. (if ever :-)
> addr = prof->pr_base + i;
> if (copyin(addr, &v, sizeof(v)) == 0) {
> v += ticks;
> if (copyout(&v, addr, sizeof(v)) == 0)
>
>
> Tim
>
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0302090036030.31703-100000>