2014-04-07 14:41:37

by Namjae Jeon

[permalink] [raw]
Subject: [PATCH] ext4: fix shared/002 and 004 fail issue in xfstests

From: Namjae Jeon <[email protected]>

When mounting ext4 with data=journal option, shared/002 and 004 are
failed on xfstests. Calling ext4_force_commit before
filemap_write_and_wait_range is needed to flush all data into journal
on this case.

Signed-off-by: Namjae Jeon <[email protected]>
Signed-off-by: Ashish Sangwan <[email protected]>
---
fs/ext4/extents.c | 7 +++++++
1 file changed, 7 insertions(+)

diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index 243a02e..dc9333f 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -5375,6 +5375,13 @@ int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len)
punch_start = offset >> EXT4_BLOCK_SIZE_BITS(sb);
punch_stop = (offset + len) >> EXT4_BLOCK_SIZE_BITS(sb);

+ /* Call ext4_force_commit to flush all data in case of data=journal. */
+ if (ext4_should_journal_data(inode)) {
+ ret = ext4_force_commit(inode->i_sb);
+ if (ret)
+ return ret;
+ }
+
/* Write out all dirty pages */
ret = filemap_write_and_wait_range(inode->i_mapping, offset, -1);
if (ret)
--
1.7.9.5



2014-04-07 16:25:24

by Theodore Ts'o

[permalink] [raw]
Subject: Re: [PATCH] ext4: fix shared/002 and 004 fail issue in xfstests

On Mon, Apr 07, 2014 at 11:41:15PM +0900, Namjae Jeon wrote:
> From: Namjae Jeon <[email protected]>
>
> When mounting ext4 with data=journal option, shared/002 and 004 are
> failed on xfstests. Calling ext4_force_commit before
> filemap_write_and_wait_range is needed to flush all data into journal
> on this case.

Hi Namjae,

Thanks for working for this!

Could you include in your commit description of why adding a call to
ext4_force_commit() is necessary to avoid the problem, and what wa
going wrong in the data=journal case?

Thanks,

- Ted

2014-04-07 23:12:36

by Namjae Jeon

[permalink] [raw]
Subject: Re: [PATCH] ext4: fix shared/002 and 004 fail issue in xfstests

2014-04-08 1:25 GMT+09:00, Theodore Ts'o <[email protected]>:
> On Mon, Apr 07, 2014 at 11:41:15PM +0900, Namjae Jeon wrote:
>> From: Namjae Jeon <[email protected]>
>>
>> When mounting ext4 with data=journal option, shared/002 and 004 are
>> failed on xfstests. Calling ext4_force_commit before
>> filemap_write_and_wait_range is needed to flush all data into journal
>> on this case.
>
> Hi Namjae,
>
> Thanks for working for this!
>
> Could you include in your commit description of why adding a call to
> ext4_force_commit() is necessary to avoid the problem, and what wa
> going wrong in the data=journal case?
Hi Ted.
Yes Sure. I will post the patch with detail description again.
Thanks!
>
> Thanks,
>
> - Ted
>