2023-04-18 06:30:42

by David Stevens

Subject: [PATCH] mm/shmem: Fix race in shmem_undo_range w/THP

From: David Stevens <[email protected]>

Split folios during the second loop of shmem_undo_range. It's not
sufficient to only split folios when dealing with partial pages, since
it's possible for a THP to be faulted in after that point. Calling
truncate_inode_folio in that situation can result in throwing away data
outside of the range being targeted.

Fixes: b9a8a4195c7d ("truncate,shmem: Handle truncates that split large folios")
Cc: [email protected]
Signed-off-by: David Stevens <[email protected]>
---
mm/shmem.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/shmem.c b/mm/shmem.c
index 9218c955f482..317cbeb0fb6b 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -1033,7 +1033,7 @@ static void shmem_undo_range(struct inode *inode, loff_t lstart, loff_t lend,
}
VM_BUG_ON_FOLIO(folio_test_writeback(folio),
folio);
- truncate_inode_folio(mapping, folio);
+ truncate_inode_partial_folio(folio, lstart, lend);
}
folio_unlock(folio);
}
--
2.40.0.634.g4ca3ef3211-goog

2023-04-18 08:31:39

by David Stevens

Subject: Re: [PATCH] mm/shmem: Fix race in shmem_undo_range w/THP

On Tue, Apr 18, 2023 at 3:22 PM David Stevens <[email protected]> wrote:
>
> From: David Stevens <[email protected]>
>
> Split folios during the second loop of shmem_undo_range. It's not
> sufficient to only split folios when dealing with partial pages, since
> it's possible for a THP to be faulted in after that point. Calling
> truncate_inode_folio in that situation can result in throwing away data
> outside of the range being targeted.
>
> Fixes: b9a8a4195c7d ("truncate,shmem: Handle truncates that split large folios")
> Cc: [email protected]
> Signed-off-by: David Stevens <[email protected]>
> ---
> mm/shmem.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/mm/shmem.c b/mm/shmem.c
> index 9218c955f482..317cbeb0fb6b 100644
> --- a/mm/shmem.c
> +++ b/mm/shmem.c
> @@ -1033,7 +1033,7 @@ static void shmem_undo_range(struct inode *inode, loff_t lstart, loff_t lend,
> }
> VM_BUG_ON_FOLIO(folio_test_writeback(folio),
> folio);
> - truncate_inode_folio(mapping, folio);
> + truncate_inode_partial_folio(folio, lstart, lend);

It was pointed out to me that truncate_inode_partial_folio only
sometimes frees the target pages. So this patch does fix the data
loss, but it ends up making partial hole punches on a THP not actually
free memory. I'll send out a v2 that properly calls
truncate_inode_folio after splitting a THP.

-David

> }
> folio_unlock(folio);
> }
> --
> 2.40.0.634.g4ca3ef3211-goog
>