Josef Bacik
2012-Oct-05 17:49 UTC
[PATCH] Btrfs: cleanup pages properly when ENOMEM in compression
We were freeing non-existent pages which was causing a panic for a user who
was suffering from ENOMEM. This patch fixes the problem. Thanks,
Reported-by: Jérôme Poulin <jeromepoulin@gmail.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
fs/btrfs/compression.c | 12 +++++++++---
1 files changed, 9 insertions(+), 3 deletions(-)
diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c
index 43d1c5a..36eb5f2 100644
--- a/fs/btrfs/compression.c
+++ b/fs/btrfs/compression.c
@@ -577,6 +577,7 @@ int btrfs_submit_compressed_read(struct inode *inode, struct
bio *bio,
u64 em_start;
struct extent_map *em;
int ret = -ENOMEM;
+ int faili = 0;
u32 *sums;
tree = &BTRFS_I(inode)->io_tree;
@@ -626,8 +627,11 @@ int btrfs_submit_compressed_read(struct inode *inode,
struct bio *bio,
for (pg_index = 0; pg_index < nr_pages; pg_index++) {
cb->compressed_pages[pg_index] = alloc_page(GFP_NOFS |
__GFP_HIGHMEM);
- if (!cb->compressed_pages[pg_index])
+ if (!cb->compressed_pages[pg_index]) {
+ faili = pg_index - 1;
+ ret = -ENOMEM;
goto fail2;
+ }
}
cb->nr_pages = nr_pages;
@@ -713,8 +717,10 @@ int btrfs_submit_compressed_read(struct inode *inode,
struct bio *bio,
return 0;
fail2:
- for (pg_index = 0; pg_index < nr_pages; pg_index++)
- free_page((unsigned long)cb->compressed_pages[pg_index]);
+ while (faili >= 0) {
+ __free_page(cb->compressed_pages[pg_index]);
+ faili--;
+ }
kfree(cb->compressed_pages);
fail1:
--
1.7.7.6
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs"
in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
David Sterba
2012-Oct-05 20:06 UTC
Re: [PATCH] Btrfs: cleanup pages properly when ENOMEM in compression
On Fri, Oct 05, 2012 at 01:49:19PM -0400, Josef Bacik wrote:> --- a/fs/btrfs/compression.c > +++ b/fs/btrfs/compression.c > @@ -577,6 +577,7 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio, > u64 em_start; > struct extent_map *em; > int ret = -ENOMEM; > + int faili = 0; > u32 *sums; > > tree = &BTRFS_I(inode)->io_tree; > @@ -626,8 +627,11 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio, > for (pg_index = 0; pg_index < nr_pages; pg_index++) { > cb->compressed_pages[pg_index] = alloc_page(GFP_NOFS | > __GFP_HIGHMEM); > - if (!cb->compressed_pages[pg_index]) > + if (!cb->compressed_pages[pg_index]) { > + faili = pg_index - 1; > + ret = -ENOMEM; > goto fail2; > + } > } > cb->nr_pages = nr_pages; > > @@ -713,8 +717,10 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio, > return 0; > > fail2: > - for (pg_index = 0; pg_index < nr_pages; pg_index++) > - free_page((unsigned long)cb->compressed_pages[pg_index]); > + while (faili >= 0) { > + __free_page(cb->compressed_pages[pg_index]);^^^^^^^^ s/pg_index/faili/ Also, faili (how do you pronounce that?) is initialized to 0, the while loop will always fire (and burn a page), initialize it to -1. Other than that, I like the fix, thanks.> + faili--; > + } > > kfree(cb->compressed_pages); > fail1:david -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
David Sterba
2012-Oct-05 20:32 UTC
Re: [PATCH] Btrfs: cleanup pages properly when ENOMEM in compression
On Fri, Oct 05, 2012 at 10:06:17PM +0200, David Sterba wrote:> Also, faili (how do you pronounce that?) is initialized to 0, the while > loop will always fire (and burn a page), initialize it to -1.My oversight sorry, it takes a different codepath and this will not happen. david -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html