Messages in this thread | | | Date | Thu, 19 Feb 2015 15:48:05 +0100 | From | David Hildenbrand <> | Subject | Re: [PATCH v2 0/5] Reenable might_sleep() checks for might_fault() |
| |
> On Mon, Jan 12, 2015 at 03:19:11PM +0100, David Hildenbrand wrote: > > Thomas, Peter, > > > > anything that speaks against putting the pagefault_disable counter into > > thread_info (my series) instead of task_struct (rt tree)? > > > > IOW, what would be the right place for it? > > I think we put it in task_struct because lazy; ARM seems one of the few > popular archs where current still goes through thread_info. > > And that I think is the only reason to maybe use thread_info, cost of > access. The down-side of using thread_info is of course that it reduces > stack size. > > In any case; I think that if you want to go do this; please consider the > route -rt took and completely separate the two, don't leave the > preempt_count_{inc,dec} remnant in pagefault_{en,dis}able() at all. > >
Thanks Peter,
I am currently preparing/testing a series that does the requested separation (getting rid of preempt_count_{inc,dec} ...) while putting the pagefault disable count into task_info.
Downside is that now that I have to touch all fault handlers, I have to go through all archs again.
Think I'll have something to show in a couple of days.
David
| |