diff mbox

donor file data inconsistent after EXT4_IOC_MOVE_EXT

Message ID 4ADBFC4D.8050405@gmail.com
State New, archived
Headers show

Commit Message

Peng Tao Oct. 19, 2009, 5:42 a.m. UTC 
Hi, Akira,

Peng Tao wrote:
> Hi, Akira,
> Akira Fujita wrote:
>> Hi Peng,
>>
>> This is a known issue, and I sent a patch to linux-ext4 2 weeks ago.
>> Unfortunately it is not included in the ext4 patch queue yet.
>>
>> http://marc.info/?l=linux-ext4&m=125447192709338&w=2
>>
>> Would you retry your test case with above my patch?
> It didn't work. I still got the old donor file data.
invalidate_mapping_pages() in your patch won't invalidate the locked page returned by
grab_cache_page(). The following patch addresses the bug by calling
invalidate_inode_pages2_range() instead.

commit e88c6fa454c38039e9f342196129f8ef782f5edb
Author: Peng Tao <bergwolf@gmail.com>
Date:   Mon Oct 19 10:57:51 2009 +0800

    ext4: Invalidate donor file pagecache in EXT4_IOC_MOVE_EXT
    
    After calling EXT4_IOC_MOVE_EXT, the donor file will have data blocks previously
    owned by the orig file. But the pagecache for donor file still refers to old
    blocks in memory. The patch invalidates all pagecaches for donor file corresponding
    to replaced blocks. So later reading donor file won't hit the wrong pagecache.
    
    Signed-off-by: Peng Tao <bergwolf@gmail.com>




--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Comments

Akira Fujita Oct. 19, 2009, 8:28 a.m. UTC | #1 
Hi Peng,
Peng Tao wrote:
> Hi, Akira,
> 
> Peng Tao wrote:
>> Hi, Akira,
>> Akira Fujita wrote:
>>> Hi Peng,
>>>
>>> This is a known issue, and I sent a patch to linux-ext4 2 weeks ago.
>>> Unfortunately it is not included in the ext4 patch queue yet.
>>>
>>> http://marc.info/?l=linux-ext4&m=125447192709338&w=2
>>>
>>> Would you retry your test case with above my patch?
>> It didn't work. I still got the old donor file data.
> invalidate_mapping_pages() in your patch won't invalidate the locked page returned by
> grab_cache_page(). The following patch addresses the bug by calling
> invalidate_inode_pages2_range() instead.

Though I tested my patch, I seem to have made a mistake.
Anyway, I confirmed your patch works fine, thank you.

Regards,
Akira Fujita
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/fs/ext4/move_extent.c b/fs/ext4/move_extent.c
index 25b6b14..dcdfec8 100644
--- a/fs/ext4/move_extent.c
+++ b/fs/ext4/move_extent.c
@@ -826,6 +826,7 @@  move_extent_per_page(struct file *o_filp, struct inode *donor_inode,
 	const struct address_space_operations *a_ops = mapping->a_ops;
 	handle_t *handle;
 	ext4_lblk_t orig_blk_offset;
+	pgoff_t donor_page_offset = orig_page_offset;
 	long long offs = orig_page_offset << PAGE_CACHE_SHIFT;
 	unsigned long blocksize = orig_inode->i_sb->s_blocksize;
 	unsigned int w_flags = 0;
@@ -920,6 +921,8 @@  move_extent_per_page(struct file *o_filp, struct inode *donor_inode,
 	ext4_ext_invalidate_cache(orig_inode);
 	ext4_ext_invalidate_cache(donor_inode);
 
+	invalidate_inode_pages2_range(donor_inode->i_mapping, donor_page_offset,
+					donor_page_offset);
 	if (!page_has_buffers(page))
 		create_empty_buffers(page, 1 << orig_inode->i_blkbits, 0);