Re: sched_clock() uses are broken

[Russell King]

On Tue, May 02, 2006 at 03:05:22PM -0400, Nicolas Pitre wrote:
> On Tue, 2 May 2006, Russell King wrote:
> > On Tue, May 02, 2006 at 01:18:25PM -0400, Nicolas Pitre wrote:
> > > On Tue, 2 May 2006, Andi Kleen wrote:
> > > 
> > > > On Tuesday 02 May 2006 18:50, Russell King wrote:
> > > > 
> > > > > You're right assuming you have a 64-bit TSC, but ARM has at best a
> > > > > 32-bit cycle counter which rolls over about every 179 seconds - with
> > > > > gives a range of values from sched_clock from 0 to 178956970625 or
> > > > > 0x29AAAAAA81.
> > > > > 
> > > > > That's rather more of a problem than having it happen every 208 days.
> > > > 
> > > > Ok but you know it's always 32bit right? You can fix it up then
> > > > with your proposal of a sched_diff()
> > > > 
> > > > The problem would be fixing it up with a unknown number of bits.
> > > 
> > > Just shift it left so you know you always have the most significant bits 
> > > valid.  The sched_diff() would take care of scaling it back to nanosecs.
> > 
> > sched_clock is currently defined to return nanoseconds so this isn't
> > a possibility.
> 
> If we're discussing the addition of a sched_clock_diff(), why whouldn't 
> shed_clock() return anything it wants in that context?  It could be 
> redefined to have a return value meaningful only to shed_clock_diff()?

If we're talking about doing that, we need to replace sched_clock()
to ensure that we all users are aware that it has changed.

I did think about that for my original fix proposal, but stepped back
because that's a bigger change - and is something for post-2.6.17.
The smallest fix (suitable for -rc kernels) is as I detailed.

-- 
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 Serial core
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/