Writeback, partial page writes and data corruption (was Re: [PATCH v3] ext4: fix data integrity sync in ordered mode)

[Dave Chinner <david@fromorbit.com>]

[cc linux-fsdevel as a heads-up]

On Thu, May 08, 2014 at 08:16:24AM +0900, Namjae Jeon wrote:
> When we perform a data integrity sync we tag all the dirty pages with
> PAGECACHE_TAG_TOWRITE at start of ext4_da_writepages.
> Later we check for this tag in write_cache_pages_da and creates a
> struct mpage_da_data containing contiguously indexed pages tagged with this
> tag and sync these pages with a call to mpage_da_map_and_submit.
> This process is done in while loop until all the PAGECACHE_TAG_TOWRITE pages
> are synced. We also do journal start and stop in each iteration.
> journal_stop could initiate journal commit which would call ext4_writepage
> which in turn will call ext4_bio_write_page even for delayed OR unwritten
> buffers. When ext4_bio_write_page is called for such buffers, even though it
> does not sync them but it clears the PAGECACHE_TAG_TOWRITE of the corresponding
> page and hence these pages are also not synced by the currently running data
> integrity sync. We will end up with dirty pages although sync is completed.
> 
> This could cause a potential data loss when the sync call is followed by a
> truncate_pagecache call, which is exactly the case in collapse_range.
> (It will cause generic/127 failure in xfstests)

Yes, this is a patch that went into 3.16, but I only just found out
about it because Brian just found a very similar data corruption bug
in XFS. i.e. a partial page write was starting writeback and hence
clearing PAGECACHE_TAG_TOWRITE before the page was fully cleaned and
hence WB_SYNC_ALL wasn't writing the entire page.

http://oss.sgi.com/pipermail/xfs/2014-September/038150.html
http://oss.sgi.com/pipermail/xfs/2014-September/038167.html

IOWs, if a filesystem does write-ahead in ->writepages() or
relies on the write_cache_pages() layer to reissue dirty pages in
partial page write situations for data integrity purposes, then it
needs to be converted to use set_page_writeback_keepwrite() until
the page is fully clean, at which point it can then use
set_page_writeback().

For everyone: if one filesystem is using the generic code
incorrectly, then it is likely the same or similar bugs exist in
other filesystems. As a courtesy to your fellow filesystem
developers, if you find a data corruption bug caused by interactions
with the generic code can the fixes please be CC'd to linux-fsdevel
so everyone knows about the issue? This is especially important if
new interfaces in the generic code have been added to avoid the
problem.

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>