On Sat, Jun 11, 2011 at 07:51:36PM +0200, Johannes Weiner wrote:

This is a problem with the patch, but I think Kame's analysis and
approach to fix it are still correct.

I agree with Kame's analysis too. This explains why removing the
mmap_sem read mode introduced the problem, it was quite some
unexpected subtleness not apparent to the naked eye, as memcg didn't
explicitly relay on mmap_sem but it did implicitly during exit because
of the __khugepaged_exit waiting if we were collapsing an hugepage...

So the fix is safe because the task struct is freed with
delayed_put_task_struct and that won't run until we rcu_read_unlock
after mem_cgroup_from_task.

Reviewed-by: Andrea Arcangeli <redacted>

Thanks,
Andrea

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>