sched_clock always 0 and no process time accounting with 3.11-rc1

From: baruch@tkos.co.il (Baruch Siach)
To: linux-arm-kernel@lists.infradead.org
Subject: sched_clock always 0 and no process time accounting with 3.11-rc1
Date: Wed, 17 Jul 2013 12:12:09 +0300	[thread overview]
Message-ID: <20130717091209.GL6950@tarshish> (raw)
In-Reply-To: <20130717090354.GK24642@n2100.arm.linux.org.uk>

Hi Russell,

On Wed, Jul 17, 2013 at 10:03:55AM +0100, Russell King - ARM Linux wrote:
> On Wed, Jul 17, 2013 at 09:19:25AM +0300, Baruch Siach wrote:
> > Apparently the expression '(1 << 32)' evaluates to 1 on xtensa cross gcc, and 
> > x86_84 native gcc. According to my limited understanding, the C compiler is 
> > allowed to do so. This caused sched_clock_32() to return constant 0. I wonder 
> > how it didn't bite the ARM people who are using this code from quite some time 
> > (added LAKL to CC).
> 
> It (a) used to be only 32-bit, and (b) <<32 on ARM evaluates to zero in
> a 32-bit context (it's not a rotate).
> 
> Patch looks fine.

Thanks for the confirmation, I'll take it as an ack if you don't mind.

> I wonder who takes it... after all, I used to look
> after that code (after all, I'm the author of it), but now I guess as
> it's been moved under kernel/time, it's someone elses responsibility.

Since the code move went through the tree of the timekeepers, John Stultz and 
Thomas Gleixner, I guess they should handle it.

baruch

-- 
     http://baruch.siach.name/blog/                  ~. .~   Tk Open Systems
=}------------------------------------------------ooO--U--Ooo------------{=
   - baruch at tkos.co.il - tel: +972.2.679.5364, http://www.tkos.co.il -