linux-ext4 - Re: Exporting ext4-specific information through fsinfo attributes

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

Date:   Wed, 22 Apr 2020 09:12:42 -0700
From:   "Darrick J. Wong" <darrick.wong@...cle.com>
To:     Andreas Dilger <adilger@...ger.ca>
Cc:     David Howells <dhowells@...hat.com>,
        Eric Biggers <ebiggers@...nel.org>,
        "Theodore Ts'o" <tytso@....edu>,
        Ext4 Developers List <linux-ext4@...r.kernel.org>
Subject: Re: Exporting ext4-specific information through fsinfo attributes

On Tue, Apr 21, 2020 at 04:07:43PM -0600, Andreas Dilger wrote:
> On Apr 21, 2020, at 10:17 AM, David Howells <dhowells@...hat.com> wrote:
> > 
> > Darrick J. Wong <darrick.wong@...cle.com> wrote:
> > 
> >> The entire superblock as a binary blob? :)
> > 
> > How about the attached?  Please forgive the duplication of struct
> > ext4_super_block into the test program, but it's not in the UAPI.
> 
> I think (hope?) Darrick was joking?

<cough> 80% joking.  It /was/ April 1st, after all. :)

Ted has said on a few occasions that one of the big stumbling blocks to
making tune2fs safe wrt mounted ext4 is that there's no way to get or
set superblock fields in a way that's kernel-mediated, and maybe the
easiest way is to export the superblock instead of playing this game
with synchronous sb writes from the kernel and the strange
mask-and-write behavior that tune2fs does.

I'm not convinced that's the best way to do that, though at least for
/reading/ the superblock I guess it beats defining a whole new structure
and ioctl.

> At least IMHO, exporting the whole superblock as a binary blob is not
> a great user interface.  I guess it has the benefit of allowing access
> to various non-standard fields without accessing the device directly.
> Kind of like SCSI mode pages, but that can get ugly quickly...
> 
> I can definitely get behind adding generic properties like the ones
> you list below.

The other properties look fine (in principle) to me though. :)

--D

> > 
> > David
> > ---
> > fsinfo: Add support to ext4
> > 
> > Add support to ext4, including the following:
> > 
> > (1) FSINFO_ATTR_SUPPORTS: Information about supported STATX attributes and
> >     support for ioctls like FS_IOC_[GS]ETFLAGS and FS_IOC_FS[GS]ETXATTR.
> > 
> > (2) FSINFO_ATTR_FEATURES: Information about features supported by an ext4
> >     filesystem, such as whether version counting, birth time and name case
> >     folding are in operation.
> > 
> > (3) FSINFO_ATTR_VOLUME_NAME: The volume name from the superblock.
> > 
> > (4) FSINFO_ATTR_EXT4_SUPERBLOCK: The entirety of the on disk-format
> >     superblock record as an opaque blob.
> > 
> > Signed-off-by: David Howells <dhowells@...hat.com>
> > cc: "Theodore Ts'o" <tytso@....edu>
> > cc: Andreas Dilger <adilger.kernel@...ger.ca>
> > cc: "Darrick J. Wong" <darrick.wong@...cle.com>
> > cc: Eric Biggers <ebiggers@...nel.org>
> > cc: linux-ext4@...r.kernel.org
> 
> Cheers, Andreas
> 
> 
> 
> 
>