Re: [PATCH] idr: Use this_cpu_ptr() for percpu_ida

[Andrew Morton]

On Wed, 21 Aug 2013 14:32:55 +0000 Christoph Lameter <cl@xxxxxxxxx> wrote:

> On Tue, 20 Aug 2013, Nicholas A. Bellinger wrote:
> 
> > On Thu, 2013-08-08 at 14:32 +0000, Christoph Lameter wrote:
> > > On Wed, 7 Aug 2013, Kent Overstreet wrote:
> > >
> > > > One thing that was bugging me - I was never able to figure out for sure
> > > > if smp_processor_id() returns a number in the range [0, nr_cpu_ids), at
> > > > least I couldn't find where it was documented - could you tell me if
> > > > that's true?
> > >
> > > I always assumed that it was in the range 0 ... nr_cpu_ids - 1 and that is
> > > the assumption under which the kernel code was written. Things would break
> > > horribly if smp_process_id would return nr_cpu_ids or higher.
> > >
> >
> > Hi guys,
> >
> > Just a heads up that I've put Kent's standalone percpu-ida patch (with
> > Christoph's recommend changes) into target-pending/for-next here:
> >
> > https://git.kernel.org/cgit/linux/kernel/git/nab/target-pending.git/commit/?h=for-next&id=47bd524a5b3eb6429b058b8b562b45329ab2c9e7
> >
> > I've got a number of target patches that depend on this code for v3.12,
> > and a delay on this particular piece would be painful to endure..
> >
> > Sooo, please yell loudly if there is an objection to percpu-ida merge as
> > a completely standalone item, that does not effect any existing ida
> > code.
> 
> Well the performance is still going to be limited due to the spinlock in
> the percpu handling. You do not need the spinlock. Once preempt is off you
> should have exclusive access to the per cpu data.

The lock is needed so that one cpu can steal tags from another cpu's cache.
See (the needlessly inlined!) steal_tags().
--
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/