btrfs: don't hold CPU for too long when defragging a file

adam900710 · kdave · commit ea0eba69a2a8 · 2022-02-09T18:50:56.000+01:00
There is a user report about "btrfs filesystem defrag" causing 120s timeout problem. For btrfs_defrag_file() it will iterate all file extents if called from defrag ioctl, thus it can take a long time. There is no reason not to release the CPU during such a long operation. Add cond_resched() after defragged one cluster. CC: stable@vger.kernel.org # 5.16 Link: https://lore.kernel.org/linux-btrfs/10e51417-2203-f0a4-2021-86c8511cc367@gmx.com Signed-off-by: Qu Wenruo <wqu@suse.com> Reviewed-by: David Sterba <dsterba@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
@@ -1629,6 +1629,7 @@ int btrfs_defrag_file(struct inode *inode, struct file_ra_state *ra,
 			ret = 0;
 			break;
 		}
+		cond_resched();
 	}
 
 	if (ra_allocated)

Original file line number	Diff line number	Diff line change
`@@ -1629,6 +1629,7 @@ int btrfs_defrag_file(struct inode inode, struct file_ra_state ra,`
`1629`	`1629`	`ret = 0;`
`1630`	`1630`	`break;`
`1631`	`1631`	`}`
	`1632`	`+ cond_resched();`
`1632`	`1633`	`}`
`1633`	`1634`
`1634`	`1635`	`if (ra_allocated)`