xfs
[Top] [All Lists]

Re: [PATCH]xfs: Fix xfs_swap_extents() after removal of xfs_flushinval_p

To: xfs@xxxxxxxxxxx
Subject: Re: [PATCH]xfs: Fix xfs_swap_extents() after removal of xfs_flushinval_pages()
From: "Michael L. Semon" <mlsemon35@xxxxxxxxx>
Date: Tue, 22 Jan 2013 01:23:02 -0500
Delivered-to: xfs@xxxxxxxxxxx
Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:x-received:date:message-id:subject:from:to :content-type; bh=6NZ/nME6SGkQZ7zwiXGlUQUTFhB+DQk8f7TOGYmauQY=; b=HIwBUG6SYIyVe40M0NyTQ0wULACfmj5CY62GuOnOjQW3njT3b4jeOEtIRv3BbaVCDP CvXf70NBFwX/KahwywYaglCicu6Ws3Cu1YD4NVI/WZKA/uQbq7TAHnNYzQJ5q17YCDHH dpZoeXs8aPCXzWcC7J4mYWV0NB/ScxvESCcKiZoiZPZ0mDKU67SH39pIhfErCTN30p4O g58uTEjdN3GQQ7hRqq4himXc6g78O96fU7mSv4Co9utr2Z2KZitB1qDuinXhRjlmyMAu EXFBA4OdVJ/Mh3ef6T3mWhd+bbB7opbkrmYyCE83OIaPr7iscC2XIgBJLW+VcrO18peY LHCw==
> From: Torsten Kaiser <just.for.lkml at googlemail.com>
>
> Commit fb59581404ab7ec5075299065c22cb211a9262a9 removed
> xfs_flushinval_pages() and changed its callers to use
> filemap_write_and_wait() and  truncate_pagecache_range() directly.
>
> But in xfs_swap_extents() this change accidental switched the argument
> for 'tip' to 'ip'. This patch switches it back to 'tip'

The patch works in casual testing, and I can defragment files using
Linux 3.8-rc4.  It solves the problem listed here:

http://oss.sgi.com/pipermail/xfs/2013-January/023379.html

However, I'm having trouble getting the filesystems to fragment again
so that a more formal test can be done...and that's one of the many
reasons that I use XFS ;-)

xfstests #042, #222, and #227 passed outright, with and without a lot
of debug options in the kernel.  Very nice!

Without debug options, xfstests #111 failed with a simple "Directory
defragmentation not supported" message but did not cause problems for
later tests.  With debug options, xfstests #111 causes a screenful of
oops and a stopped PC--it did this when defrag was broken, too--and I
don't know whether to blame the defrag code, the test code, or the
debug code.  I'm willing to write down that oops screen, but only on
request.

In other words, I'm happy with the patch.  Good work!

Michael

<Prev in Thread] Current Thread [Next in Thread>