[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1180983647.4404.38.camel@chaos>
Date: Mon, 04 Jun 2007 21:00:47 +0200
From: Thomas Gleixner <tglx@...utronix.de>
To: Stephen Hemminger <shemminger@...ux-foundation.org>
Cc: Ulrich Drepper <drepper@...il.com>,
Maximilian Engelhardt <maxi@...monizer.de>,
Michael Buesch <mb@...sch.de>,
linux-kernel <linux-kernel@...r.kernel.org>,
linux-wireless <linux-wireless@...r.kernel.org>,
Arnaldo Carvalho de Melo <acme@...stprotocols.net>,
Jeff Garzik <jgarzik@...ox.com>,
Gary Zambrano <zambrano@...adcom.com>, netdev@...r.kernel.org,
Andrew Morton <akpm@...ux-foundation.org>,
Ingo Molnar <mingo@...e.hu>
Subject: Re: iperf: performance regression (was b44 driver problem?)
On Mon, 2007-06-04 at 10:51 -0700, Stephen Hemminger wrote:
> > I doubt that. This is in the iperf code itself.
> >
> > void thread_rest ( void ) {
> > #if defined( HAVE_THREAD )
> > #if defined( HAVE_POSIX_THREAD )
> > // TODO add checks for sched_yield or pthread_yield and call that
> > // if available
> > usleep( 0 );
> >
> > ----------^^^^
> >
> > It results in a nanosleep({0,0}, NULL)
> >
> > tglx
> >
>
> Yes, the following patch makes iperf work better than ever.
> But are other broken applications going to have same problem.
> Sounds like the old "who runs first" fork() problems.
Not really. The fork() "who runs first" problem is nowhere specified.
usleep(0) is well defined:
.... If the value of useconds is 0, then the call has no effect.
So the call into the kernel has been wrong for quite a time.
tglx
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists