2023-09-19 17:33:17

by Daniel Gomez

Subject: [PATCH v2 2/6] shmem: return freed pages in shmem_free_swap

Both shmem_free_swap callers require to get the number of pages in the
folio after calling shmem_free_swap. Make shmem_free_swap return the
expected value directly and return 0 number of pages being freed
to avoid error handling in the external accounting.

Suggested-by: Matthew Wilcox <[email protected]>
Signed-off-by: Daniel Gomez <[email protected]>
---
mm/shmem.c | 14 ++++++++------
1 file changed, 8 insertions(+), 6 deletions(-)

diff --git a/mm/shmem.c b/mm/shmem.c
index de0d0fa0349e..5c9e80207cbf 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -846,16 +846,18 @@ static void shmem_delete_from_page_cache(struct folio *folio, void *radswap)
/*
* Remove swap entry from page cache, free the swap and its page cache.
*/
-static int shmem_free_swap(struct address_space *mapping,
+static long shmem_free_swap(struct address_space *mapping,
pgoff_t index, void *radswap)
{
void *old;

old = xa_cmpxchg_irq(&mapping->i_pages, index, radswap, NULL, 0);
if (old != radswap)
- return -ENOENT;
+ return 0;
+
free_swap_and_cache(radix_to_swp_entry(radswap));
- return 0;
+
+ return folio_nr_pages((struct folio *)radswap);
}

/*
@@ -1008,7 +1010,7 @@ static void shmem_undo_range(struct inode *inode, loff_t lstart, loff_t lend,
if (xa_is_value(folio)) {
if (unfalloc)
continue;
- nr_swaps_freed += !shmem_free_swap(mapping,
+ nr_swaps_freed += shmem_free_swap(mapping,
indices[i], folio);
continue;
}
@@ -1077,12 +1079,12 @@ static void shmem_undo_range(struct inode *inode, loff_t lstart, loff_t lend,
if (xa_is_value(folio)) {
if (unfalloc)
continue;
- if (shmem_free_swap(mapping, indices[i], folio)) {
+ nr_swaps_freed += shmem_free_swap(mapping, indices[i], folio);
+ if (!nr_swaps_freed) {
/* Swap was replaced by page: retry */
index = indices[i];
break;
}
- nr_swaps_freed++;
continue;
}

--
2.39.2

2023-09-19 19:59:31

by Matthew Wilcox

Subject: Re: [PATCH v2 2/6] shmem: return freed pages in shmem_free_swap

On Tue, Sep 19, 2023 at 01:55:47PM +0000, Daniel Gomez wrote:
> +++ b/mm/shmem.c
> @@ -846,16 +846,18 @@ static void shmem_delete_from_page_cache(struct folio *folio, void *radswap)
> /*
> * Remove swap entry from page cache, free the swap and its page cache.
> */
> -static int shmem_free_swap(struct address_space *mapping,
> +static long shmem_free_swap(struct address_space *mapping,
> pgoff_t index, void *radswap)
> {
> void *old;
>
> old = xa_cmpxchg_irq(&mapping->i_pages, index, radswap, NULL, 0);
> if (old != radswap)
> - return -ENOENT;
> + return 0;
> +
> free_swap_and_cache(radix_to_swp_entry(radswap));
> - return 0;
> +
> + return folio_nr_pages((struct folio *)radswap);
> }

Oh my goodness. I have led you astray; my apologies.

shmem_free_swap() is called when the 'folio' is NOT actually a folio.
It's an 'exceptional' / 'value' entry. We can't do this.

Do we encode the size of the swap entry in the swp_entry_t or do
we have to get that information from the XArray (which no longer
knows it after we've stored a NULL there)?