[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20160509082331.GB11897@quack2.suse.cz>
Date: Mon, 9 May 2016 10:23:31 +0200
From: Jan Kara <jack@...e.cz>
To: Christoph Hellwig <hch@...radead.org>
Cc: Toshi Kani <toshi.kani@....com>, dan.j.williams@...el.com,
jack@...e.cz, david@...morbit.com, viro@...iv.linux.org.uk,
axboe@...com, boaz@...xistor.com, tytso@....edu,
adilger.kernel@...ger.ca, ross.zwisler@...ux.intel.com,
micah.parrish@....com, linux-nvdimm@...1.01.org,
linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v3 4/5] ext2: Add alignment check for DAX mount
On Sun 08-05-16 01:59:37, Christoph Hellwig wrote:
> Not really for the patch, but given that we have the right people
> on CC:
>
> Do we really want to keep DAX support in ext2 in the long run? ext2
> is missing a lot of the useful features for a modern FS, shouldn't
> we direct people to use ext4 (in non-journal mode if needed) if they
> want to use DAX? ext4 will even support the unmodified ext2 fs, so it
> shouldn't be a big hurdle, and it would avoid a lot of churn in ext2.
I've heard concerns that embedded people use ext2 driver (due to smaller
code size than ext4 - 1.7 MB object for ext2 vs over 7 MB object for ext4
in my build) including old XIP support. DAX has replaced the old XIP code
so removing DAX from ext2 would be a regression for them - either they'd
have to go with significantly larger module or without XIP. So for now I'd
leave DAX in ext2.
Honza
--
Jan Kara <jack@...e.com>
SUSE Labs, CR
Powered by blists - more mailing lists