| Message ID | 20190424084655.31615-1-nborisov@suse.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | btrfs: Fix off-by-one error btrfs_trim_free_extents | expand |
On Wed, Apr 24, 2019 at 11:46:55AM +0300, Nikolay Borisov wrote: > btrfs_trim_free_extents always caps the range it's going to trim based > on the size of the device. This happens if find_first_clear_extent_bit > detects that untrimmed range is past the last allocated range. Since it > doesn't have knowledge of the size of the device it just returns (u64)-1 > for end. Then btrfs_trim_free_extents caps this to device->total_bytes. > However, btrfs_trim_free_extent calculates 'len' based off of the > actual usable end byte - in this case total_bytes - 1. > > Signed-off-by: Nikolay Borisov <nborisov@suse.com> > > Fixes: 4d877687cbbc ("btrfs: Switch btrfs_trim_free_extents to find_first_clear_extent_bit") > --- > > David you might want to squash that in the 'Fixes' commit. With this I see > generic/500 passing and also full xfstest didn't uncover any new regressions. Folded, thanks.
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index d9e2e35700fd..5a4b81259413 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -11175,7 +11175,7 @@ static int btrfs_trim_free_extents(struct btrfs_device *device, * end of the device it will set end to -1, in this case it's up * to the caller to trim the value to the size of the device. */ - end = min(end, device->total_bytes); + end = min(end, device->total_bytes - 1); len = end - start + 1; /* We didn't find any extents */
btrfs_trim_free_extents always caps the range it's going to trim based on the size of the device. This happens if find_first_clear_extent_bit detects that untrimmed range is past the last allocated range. Since it doesn't have knowledge of the size of the device it just returns (u64)-1 for end. Then btrfs_trim_free_extents caps this to device->total_bytes. However, btrfs_trim_free_extent calculates 'len' based off of the actual usable end byte - in this case total_bytes - 1. Signed-off-by: Nikolay Borisov <nborisov@suse.com> Fixes: 4d877687cbbc ("btrfs: Switch btrfs_trim_free_extents to find_first_clear_extent_bit") --- David you might want to squash that in the 'Fixes' commit. With this I see generic/500 passing and also full xfstest didn't uncover any new regressions. fs/btrfs/extent-tree.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)