Hi there!
Bug Report Filed: https://bugzilla.kernel.org/show_bug.cgi?id=70091
Linux Version [3.13]
Configuration: Default configuration for x86
In function (swap_inode_boot_loader) in file (linux-3.13/fs/ext4/ioctl.c):
The structures: (inode->i_mutex) and (inode_bl->i_mutex) gets
successfully locked at line (133) by (lock_two_nondirectories(inode,
inode_bl), but both are not unlocked when the function returns after
line (147), however they got unlocked only at line (208) by
(unlock_two_nondirectories(inode, inode_bl)).
A possible solution is to call (unlock_two_nondirectories(inode,
inode_bl)) before line (147).
Thanks,
Ahmed
Hi Ahmed,
On Wed, Feb 05, 2014 at 01:39:06PM -0600, Ahmed Tamrawi wrote:
> Hi there!
>
> Bug Report Filed: https://bugzilla.kernel.org/show_bug.cgi?id=70091
> Linux Version [3.13]
> Configuration: Default configuration for x86
>
>
> In function (swap_inode_boot_loader) in file (linux-3.13/fs/ext4/ioctl.c):
>
> The structures: (inode->i_mutex) and (inode_bl->i_mutex) gets
> successfully locked at line (133) by (lock_two_nondirectories(inode,
> inode_bl), but both are not unlocked when the function returns after
> line (147), however they got unlocked only at line (208) by
> (unlock_two_nondirectories(inode, inode_bl)).
>
> A possible solution is to call (unlock_two_nondirectories(inode,
> inode_bl)) before line (147).
Thanks for reporting this issue. I believe the following patch that can
fix the issue.
Regards,
- Zheng
Subject: [PATCH] ext4: release ->i_mutex and resume unlocked dio in swap_inode_boot_loader()
From: Zheng Liu <[email protected]>
In swap_inode_boot_loader() we forgot to release ->i_mutex and resume
unlocked dio for inode and inode_bl. This commit fixes this issue.
Reported-by: Ahmed Tamrawi <[email protected]>
Cc: Andreas Dilger <[email protected]>
Cc: "Theodore Ts'o" <[email protected]>
Cc: Dr. Tilmann Bubeck <[email protected]>
Signed-off-by: Zheng Liu <[email protected]>
---
fs/ext4/ioctl.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/fs/ext4/ioctl.c b/fs/ext4/ioctl.c
index 6bea806..a2a837f 100644
--- a/fs/ext4/ioctl.c
+++ b/fs/ext4/ioctl.c
@@ -140,7 +140,7 @@ static long swap_inode_boot_loader(struct super_block *sb,
handle = ext4_journal_start(inode_bl, EXT4_HT_MOVE_EXTENTS, 2);
if (IS_ERR(handle)) {
err = -EINVAL;
- goto swap_boot_out;
+ goto journal_err_out;
}
/* Protect extent tree against block allocations via delalloc */
@@ -198,6 +198,7 @@ static long swap_inode_boot_loader(struct super_block *sb,
ext4_double_up_write_data_sem(inode, inode_bl);
+journal_err_out:
ext4_inode_resume_unlocked_dio(inode);
ext4_inode_resume_unlocked_dio(inode_bl);
--
1.7.9.7
On Sun, Feb 09, 2014 at 08:27:45PM +0800, Zheng Liu wrote:
>
> In swap_inode_boot_loader() we forgot to release ->i_mutex and resume
> unlocked dio for inode and inode_bl. This commit fixes this issue.
>
> Reported-by: Ahmed Tamrawi <[email protected]>
> Cc: Andreas Dilger <[email protected]>
> Cc: "Theodore Ts'o" <[email protected]>
> Cc: Dr. Tilmann Bubeck <[email protected]>
> Signed-off-by: Zheng Liu <[email protected]>
Thanks, applied.
- Ted