From: tytso@mit.edu
To: Peter Hallberg <hallberg68@hotmail.com>
Cc: linux-kernel@vger.kernel.org
Subject: Re: What happened to mount option "noextents"
Date: Sun, 30 May 2010 14:13:00 -0400 [thread overview]
Message-ID: <20100530181300.GP26177@thunk.org> (raw)
In-Reply-To: <COL116-W2944FC5A891172B1FECD52DAED0@phx.gbl>
On Sun, May 30, 2010 at 03:52:10PM +0200, Peter Hallberg wrote:
>
> > Basically, since we couldn't think of a good use case where the mount
> > options made sense, and ext4 has way too many mount options already, we
> > decided to remove them.
>
> The ext2/3 Windows driver doesn't support extents yet so one common
> use would be to do a mount -o remount,noextents and make a dir and
> copy the files you want to read from Windows to it. I think mount -o
> remount is something that is well known, if you use tune2fs you
> would need to do research before you know if the change starts
> working right a way or after a reboot or if it is valid only until
> reboot or is static, with mount -o remount there is no risk that you
> forget to set the flag back to the right value.
Yeah, but.... it's a horrible interface to support. If one or more
directories happen to be extent-mapped, then Windows won't be able to
read it. And if there are other files that are created while the
filesystem is remounted noextents, they will be created w/o extents,
which means that fallocate won't work. The opportunities for
hard-to-understand failures are very large, and especially for
clueless Ubuntu users, it's just not something where I want to see
whining on Launchpad because they read somewhere that they could do
it, and it doesn't quite work right.
This is more something where if someone wants to patch the kernel do
something hacky like this, it's almost better because then it's not
something where we have to worry about confused users wondering why
something broke when they tried to use this interface. Something that
works 80% of the time can be far worse than something that doesn't
work at all.
The real right answer is to fix the Windows driver to support extents.
Or create a separate filesystem just for moving files from Linux to
Windows --- or just write the files directly onto the Windows
partition.
> Also when I tryed tune2fs as you said, I got the following error message:
>
> peter@sanna:~$ sudo tune2fs -O ^extents /dev/mapper/sannavg-home
> tune2fs 1.41.4 (27-Jan-2009)
> Clearing filesystem feature 'extent' not supported.
> peter@sanna:~$
Yes, right now you can only enable extents; it's not something you
can disable.
- Ted
next prev parent reply other threads:[~2010-05-30 18:13 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-05-30 12:07 What happened to mount option "noextents" Peter Hallberg
2010-05-30 12:32 ` tytso
2010-05-30 13:52 ` Peter Hallberg
2010-05-30 18:13 ` tytso [this message]
2010-05-30 13:53 ` Peter Hallberg
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20100530181300.GP26177@thunk.org \
--to=tytso@mit.edu \
--cc=hallberg68@hotmail.com \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox