Re: 4.5.0+ panic when setup loop device

[Jens Axboe]

On 03/17/2016 01:20 PM, Thomas Gleixner wrote:
> On Thu, 17 Mar 2016, Jens Axboe wrote:
> > On 03/17/2016 09:42 AM, Jens Axboe wrote:
> > > On 03/17/2016 05:01 AM, Thomas Gleixner wrote:
> > > > On Thu, 17 Mar 2016, Peter Zijlstra wrote:
> > > > > On Thu, Mar 17, 2016 at 12:39:46PM +0100, Thomas Gleixner wrote:
> > > > > > But we have to clarify and document whether holes in
> > > > > > cpu_possible_mask are not
> > > > > > allowed at all or if code like the above is simply broken.
> > > > >
> > > > > So the general rule is that cpumasks can have holes, and exempting one
> > > > > just muddles the water.
> > > > >
> > > > > Therefore I'd call the code just plain broken.
> > > >
> > > > Agreed.
> > > >
> > > > That macro is not really helping the readability of the code at all. So a
> > > > simple for_each_possible_cpu() loop would have avoided that wreckage.
> > >
> > > Does the attached work? The rest of blk-mq should deal with holes just
>
> Bah. Attachements ...

You'll live. Let's face it, all mailers suck in one way or another.

> > > fine, we found some of those issues on sparc. Not sure why this one
> > > slipped through the cracks.
> >
> > This might be better, we need to start at -1 to not miss the first one...
> > Still untested.
>
> > +static inline struct blk_mq_ctx *next_ctx(struct request_queue *q, int *i)
> > +{
> > +	do {
> > +		(*i)++;
> > +		if (*i < q->nr_queues) {
> > +			if (cpu_possible(*i))
> > +				return per_cpu_ptr(q->queue_ctx, *i);
> > +			continue;
> > +		}
> > +		break;
> > +	} while (1);
> > +
> > +	return NULL;
> > +}
> > +
> > +#define queue_for_each_ctx(q, ctx, i)					\
> > +	for ((i) = -1; (ctx = next_ctx((q), &(i))) != NULL;)
> > +
>
> What's wrong with
>
>          for_each_possible_cpu(cpu) {
>         		ctx = per_cpu_ptr(q->queue_ctx, cpu);
>
> 		....
> 	}
>
> instead of hiding it behind an incomprehensible macro mess?

We might not have mapped all of them.

--
Jens Axboe