Re: [patch V2 08/10] timer: Implement the hierarchical pull model

[Peter Zijlstra]

On Wed, Apr 19, 2017 at 11:44:45AM +0200, Peter Zijlstra wrote:
> On Wed, Apr 19, 2017 at 11:09:14AM +0200, Peter Zijlstra wrote:
> 
> > Would it be very onerous to rewrite that into regular loops? That avoids
> > us having to think (and worry) about blowing our stack.
> 
> void walk_groups(bool (*up)(void *), void (*down)(void *), void *data)
> {
> 	struct tmigr_cpu *tmc = this_cpu_ptr(&tmigr_cpu);
> 	struct group *stack[tmigr_hierarchy_levels];
> 	struct group *group = tmc->group;
> 	int i = 0;
> 
> 	raw_spin_lock(&tmc->lock);
> 
> 	do {
> 		stack[i++] = group;
> 
> 		if (up(data))
> 			break;
> 
> 	} while ((group = group->parent));
> 
> 	do {
> 		group = stack[--i];
> 
> 		down(data);
> 
> 	} while (group != tmc->group);
> 
> 	raw_spin_unlock(&tmc->lock);
> }
> 
> Something like so, iterates the hierarchy for the current CPU and calls
> @up and @down at each level in the proper order. And has obvious stack
> usage.

I should, of course, have added @group as an argument to @up,@down.
Otherwise its a bit hard for them to know what to do.