Re: [PATCH v2 12/14] mm/shmem: fix truncate loop after boundary swapin failure

[Baolin Wang]

On 10/28/26 4:21 PM, David Hildenbrand (Arm) wrote:
> On 10/28/26 02:30, Baolin Wang wrote:
> > On 10/27/26 9:03 PM, David Hildenbrand (Arm) wrote:
> > > If swapping in a large entry at the new EOF fails, shmem_undo_range()
> > > leaves the entry in place. Its final scan skips the entry, then restarts
> > > from the same start index and finds it again. Truncation can loop forever
> > > while holding i_rwsem.
> >
> > Right. But ...
> >
> > > Advance start past the skipped entry so that a restarted scan can finish,
> > > as we already do for folios that cannot be split.
> >
> > IIRC, we originally avoided advancing start because a concurrent swapin fault
> > could split the boundary large swap entry via shmem_split_large_entry(), leaving
> > small swap entries below the new start.
> >
> > Sigh.
> >
> > There doesn't seem to be a simple fix. We may need to swap in the large entry as
> > shmem_get_partial_folio() does, and then split the large folio. I'll work on
> > this as a follow-up.
>
> Thanks!
>
> So you are fine with the patch as is to fix the endless loop?

Yes, let's get this fixed first.

Reviewed-by: Baolin Wang <baolin.wang@xxxxxxxxxxxxxxxxx>