Re: [PATCH] Btrfs: reserve sufficient space for ioctl clone
From: Miao Xie <hidden>
Date: 2011-08-16 02:09:31
On tue, 9 Aug 2011 10:46:37 -0700, Sage Weil wrote:
quoted hunk ↗ jump to hunk
Fix a crash/BUG_ON in the clone ioctl due to insufficient reservation. We need to reserve space for: - adjusting the old extent (possibly splitting it) - adding the new extent - updating the inode Signed-off-by: Sage Weil <redacted> --- fs/btrfs/ioctl.c | 7 ++++++- 1 files changed, 6 insertions(+), 1 deletions(-)diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index a3c4751..f038d4a 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c@@ -2320,7 +2320,12 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd, else new_key.offset = destoff; - trans = btrfs_start_transaction(root, 1); + /* + * 1 - adjusting old extent (we may have to split it)
I don't think it is enough. If we have lots of file extents and their extent items are stored in many contiguous leaves, the drop operation may cause those leaves to be COWed. So I think we must calculate the number of leaves which will be COWed at first. Thanks Miao
+ * 1 - add new extent
+ * 1 - inode update
+ */
+ trans = btrfs_start_transaction(root, 3);
if (IS_ERR(trans)) {
ret = PTR_ERR(trans);
goto out;