public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: "Theodore Ts'o" <tytso@mit.edu>
To: Andries Brouwer <aebr@win.tue.nl>
Cc: Pankaj Agarwal <pankaj@pnpexports.com>,
	Andreas Schwab <schwab@suse.de>,
	linux-kernel@vger.kernel.org
Subject: Re: how to identify filesystem type
Date: Fri, 20 Aug 2004 23:25:21 -0400	[thread overview]
Message-ID: <20040821032520.GA15772@thunk.org> (raw)
In-Reply-To: <20040820215858.GA5771@pclin040.win.tue.nl>

On Fri, Aug 20, 2004 at 11:58:58PM +0200, Andries Brouwer wrote:
> 
> I suppose this code started as part of mount(8). For example,
> 
> # mount --guess-fstype /dev/hdb2
> reiserfs

Actually, it was fsck's filesystem type detection code, and it's since
been completely rewritten.  

> However, I cannot stress often enough that these are unreliable guesses.
> It is really undesirable when standard infrastructure starts depending
> on 99.7% guesses.

It's certainly possible for mistakes to be made, but it is really
quite reliable at this point --- escpecially since various
filesystem's mkfs programs and various lvm partition initialization
progams are pretty good about erasing other filesystems' signatures as
part of the mkfs/partition init process.  

Everything is really a "guess" at some level; there is no guarantee
that /etc/fstab is 100% accurate, or that the partition table's type
fields are accurate.  I will say that the ID code in the blkid library
is pretty paranoid about sanity checks, although of course it could be
better.

> Consequently, "blkid" is a really bad name. It gives no indication
> of the guessed nature of its results.
> 
> (I see that my current version is also broken:
> # blkid -v
> blkid 1.0.0 (12-Feb-2003)
> # blkid
> ...
> /dev/sda4: LABEL="ZIP-100" UUID="34D8-1C07" TYPE="msdos" 
> /dev/sda1: UUID="1ac5969c-8fdf-4f69-934a-c6103d93c05d" TYPE="ext2" 
> /dev/sdb4: LABEL="ZIP-100" UUID="34D8-1C07" TYPE="msdos" 
> /dev/sdb1: LABEL="CF_CARD032M" UUID="2001-1207" TYPE="msdos" 
> ...
> Here no /dev/sda1 and no /dev/sdb4 exist.)

Blkid deliberately doesn't revalidate devices without any command-line
arguments, because certain devices might timeout or block for a
long-time.  If you use "blkid /dev/sdb4", or use the library
interfaces, it will revalidate any entries found in the cache file
before returning them.

						- Ted

  reply	other threads:[~2004-08-21  3:25 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-08-19  9:07 how to identify filesystem type Pankaj Agarwal
2004-08-19  9:46 ` Andreas Schwab
2004-08-19 14:03   ` Pankaj Agarwal
2004-08-20 15:36     ` Frank Steiner
2004-08-20 20:46     ` Andreas Dilger
2004-08-20 21:58       ` Andries Brouwer
2004-08-21  3:25         ` Theodore Ts'o [this message]
2004-08-21  8:51           ` Andries Brouwer

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=20040821032520.GA15772@thunk.org \
    --to=tytso@mit.edu \
    --cc=aebr@win.tue.nl \
    --cc=linux-kernel@vger.kernel.org \
    --cc=pankaj@pnpexports.com \
    --cc=schwab@suse.de \
    /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