[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20230816124848.GA641857@lorien.usersys.redhat.com>
Date: Wed, 16 Aug 2023 08:48:48 -0400
From: Phil Auld <pauld@...hat.com>
To: Sebastian Andrzej Siewior <bigeasy@...utronix.de>
Cc: Peter Zijlstra <peterz@...radead.org>, tglx@...utronix.de,
linux-kernel@...r.kernel.org, bsegall@...gle.com,
boqun.feng@...il.com, swood@...hat.com, bristot@...hat.com,
dietmar.eggemann@....com, mingo@...hat.com, jstultz@...gle.com,
juri.lelli@...hat.com, mgorman@...e.de, rostedt@...dmis.org,
vschneid@...hat.com, vincent.guittot@...aro.org,
longman@...hat.com, will@...nel.org
Subject: Re: [PATCH 3/6] sched: Extract __schedule_loop()
On Wed, Aug 16, 2023 at 02:20:07PM +0200 Sebastian Andrzej Siewior wrote:
> On 2023-08-16 07:39:45 [-0400], Phil Auld wrote:
> > I do. Admittedly I'm not an expert in how the wchan unwinding works but
> > we have a slightly older version of this patch in our kernel (schedule_loop
> > not __schedule_loop). When I added __sched it fixed it. Maybe there
> > is something else but that seemed pretty obvious.
> >
> >
> > /* Attach to any functions which should be ignored in wchan output. */
> > #define __sched __section(".sched.text")
> >
> > I can't explain why you are not seeing it.
>
> as peterz pointed out, it is marked __always_inline so the compiler
> shouldn't make a separate function out of it.
> Could you check with _this_ series? The schedule_loop variant is in RT
> and does not have this inline thingy. So it would be good if the issue
> you report actually exists in the series that has been posted.
>
Hhm, yes. I was looking at the issue in our tree when these patches
came by. Sorry... I seem to have glossed over the __always_inline.
That would certainly work as well and, of course, does explain why you
aren't seeing it.
read more. talk less :)
Cheers,
Phil
> > Cheers,
> > Phil
>
> Sebastian
>
--
Powered by blists - more mailing lists