On Thu, Mar 23, 2017 at 3:26 AM, Jeff Layton <jlayton@xxxxxxxxxx> wrote:
> I think you're correct that NOFS would have prevented the recursion
> shown in the stack trace below.
>
> However, if you (for instance) had a userland program that was
> accessing the krbd device directly with buffered I/O, then I think you
> could still end up deadlocked here.
>
> NOIO is more restrictive than NOFS and should prevent that situation in
> addition to the one in the patch description.
What Jeff said and also, less importantly, there is no corresponding
memalloc_nofs_{save,restore}() API -- it's still being debated on the
mailing lists.
Thanks,
Ilya
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
