Re: [PATCH -v3] use per cpu data for single cpu ipi calls

[Jens Axboe <jens.axboe@oracle.com>]

On Fri, Jan 30 2009, Peter Zijlstra wrote:
> On Fri, 2009-01-30 at 13:38 +0100, Jens Axboe wrote:
> 
> > >  		if (!wait) {
> > > +			/*
> > > +			 * We are calling a function on a single CPU
> > > +			 * and we are not going to wait for it to finish.
> > > +			 * We first try to allocate the data, but if we
> > > +			 * fail, we fall back to use a per cpu data to pass
> > > +			 * the information to that CPU. Since all callers
> > > +			 * of this code will use the same data, we must
> > > +			 * synchronize the callers to prevent a new caller
> > > +			 * from corrupting the data before the callee
> > > +			 * can access it.
> > > +			 *
> > > +			 * The CSD_FLAG_LOCK is used to let us know when
> > > +			 * the IPI handler is done with the data.
> > > +			 * The first caller will set it, and the callee
> > > +			 * will clear it. The next caller must wait for
> > > +			 * it to clear before we set it again. This
> > > +			 * will make sure the callee is done with the
> > > +			 * data before a new caller will use it.
> > > +			 * We use spinlocks to manage the callers.
> > > +			 */
> > 
> > That last sentence appears stale now, since there's no locking involved
> > for the per-cpu csd.
> 
> Gah, I edited that, and generated the patch using
> 
> # git diff --stat -p 4f4b6c1a94a8735bbdc030a2911cf395495645b6.. kernel/smp.c | xclip
> 
> on Ingo's tip/master. However it seems such a diff does not include
> non-commited changes, even though a regular git diff does:
> 
> # git diff kernel/smp.c
> diff --git a/kernel/smp.c b/kernel/smp.c
> index 9eead6c..bbedbb7 100644
> --- a/kernel/smp.c
> +++ b/kernel/smp.c
> @@ -251,7 +251,6 @@ int smp_call_function_single(int cpu, void (*func) (void *info), void *info,
>                          * it to clear before we set it again. This
>                          * will make sure the callee is done with the
>                          * data before a new caller will use it.
> -                        * We use spinlocks to manage the callers.
>                          */
>                         data = kmalloc(sizeof(*data), GFP_ATOMIC);
>                         if (data)
> 
> What is the recommended way to do such a diff?

Drop the .. and it should work. Don't ask me why :-)

-- 
Jens Axboe