From: Christoph Hellwig Date: Thu, 23 Oct 2025 15:17:03 +0000 (+0200) Subject: xfs: document another racy GC case in xfs_zoned_map_extent X-Git-Tag: v6.18-rc4~7^2 X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=0db22d7ee462c42c1284e98d47840932792c1adb;p=thirdparty%2Fkernel%2Flinux.git xfs: document another racy GC case in xfs_zoned_map_extent Besides blocks being invalidated, there is another case when the original mapping could have changed between querying the rmap for GC and calling xfs_zoned_map_extent. Document it there as it took us quite some time to figure out what is going on while developing the multiple-GC protection fix. Signed-off-by: Christoph Hellwig Reviewed-by: Hans Holmberg Reviewed-by: Damien Le Moal Reviewed-by: Carlos Maiolino Reviewed-by: Darrick J. Wong Signed-off-by: Carlos Maiolino --- diff --git a/fs/xfs/xfs_zone_alloc.c b/fs/xfs/xfs_zone_alloc.c index 23cdab4515bb..040402240807 100644 --- a/fs/xfs/xfs_zone_alloc.c +++ b/fs/xfs/xfs_zone_alloc.c @@ -246,6 +246,14 @@ xfs_zoned_map_extent( * If a data write raced with this GC write, keep the existing data in * the data fork, mark our newly written GC extent as reclaimable, then * move on to the next extent. + * + * Note that this can also happen when racing with operations that do + * not actually invalidate the data, but just move it to a different + * inode (XFS_IOC_EXCHANGE_RANGE), or to a different offset inside the + * inode (FALLOC_FL_COLLAPSE_RANGE / FALLOC_FL_INSERT_RANGE). If the + * data was just moved around, GC fails to free the zone, but the zone + * becomes a GC candidate again as soon as all previous GC I/O has + * finished and these blocks will be moved out eventually. */ if (old_startblock != NULLFSBLOCK && old_startblock != data.br_startblock)