On Thu, Feb 22, 2018 at 6:09 PM, Daniel Colascione <dancol@xxxxxxxxxx> wrote:
Thanks for taking a look.On Feb 22, 2018 6:01 PM, "Minchan Kim" <minchan@xxxxxxxxxx> wrote:> diff --git a/include/linux/mm.h b/include/linux/mm.hReally heavy functioin iterates all of processes and threads.
> index ad06d42adb1a..f8129afebbdd 100644
> --- a/include/linux/mm.h
> +++ b/include/linux/mm.h
> @@ -1507,14 +1507,28 @@ extern int mprotect_fixup(struct vm_area_struct *vma,
> */
> int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
> struct page **pages);
> +
> +#ifdef SPLIT_RSS_COUNTING
> +/* Flush all task-buffered MM counters to the mm */
> +void sync_mm_rss_all_users(struct mm_struct *mm);
Just all processes and the threads of each process attached to the mm. Maybe that's not much better.
Another option would be to maintain a list (with the list_head in the mm) of all the tasks with unflushed counters. This way, we at least wouldn't have to scan the world.
