linux-ext4 - Re: Fast symlinks stored slow

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20170713202720.GV31999@redhat.com>
Date:   Thu, 13 Jul 2017 21:27:20 +0100
From:   "Richard W.M. Jones" <rjones@...hat.com>
To:     Theodore Ts'o <tytso@....edu>
Cc:     linux-ext4@...r.kernel.org, Tahsin Erdogan <tahsin@...gle.com>
Subject: Re: Fast symlinks stored slow

On Thu, Jul 13, 2017 at 02:50:37PM -0400, Theodore Ts'o wrote:
> On Thu, Jul 13, 2017 at 06:13:35PM +0100, Richard W.M. Jones wrote:
> > In this case we're using libext2fs to build an appliance filesystem,
> > used to boot a small Linux system which is then run under qemu by
> > libguestfs.  This appliance is completely rebuilt automatically under
> > many circumstances, for example a host package upgrade (eg. upgrading
> > the kernel), so it's not a long-lived filesystem that would cause a
> > problem.  Rebuilding only takes a few seconds.
> > 
> > The process is described in more detail here:
> > http://libguestfs.org/supermin.1.html#SUPERMIN-APPLIANCES
> > 
> > From our point of view the only issue are some prebuilt appliances
> > which we have provided to other distributions that cannot / don't want
> > to use supermin (http://download.libguestfs.org/binaries/appliance/)
> > and at some point I'm going to have to rebuild these using the fixed
> > supermin.
> 
> OK, so the risk is if there are other people who are using supermin to
> create appliances.  (One potential use case we might need to
> investigate are services such as SuSE Studio, since it can create
> turnkey VM appliances for its users.)  If these applianes are
> distributed end users (as opposed to being automatically rebuilt as in
> your use case), that's when we would potentially be at risk.

I can't speak about SuSE Studio, but supermin appliances aren't
distributed to end users, but get built on the fly on end user
machines.

I think you may be confusing supermin and libguestfs.  Supermin is a
component we use to make libguestfs work, but it's not how libguestfs
makes new filesystems.

For example if you write:

  $ guestfish -N disk.img=fs:ext4 -m /dev/sda1 touch /foo : ln-s /foo /bar
  $ ll disk.img 
  -rw-rw-r--. 1 rjones rjones 104857600 Jul 13 21:26 disk.img

then the result is a new ext4 filesystem in a disk image, containing a
symlink.  But it was created using the *kernel* + the symlink(2)
system call (not using libext2fs), and in all cases it was in the past
and will be in the future created correctly.

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-top is 'top' for virtual machines.  Tiny program with many
powerful monitoring features, net stats, disk stats, logging, etc.
http://people.redhat.com/~rjones/virt-top