linux-ext4 - Re: [PATCH] e2fsck: fix portability problems caused by unaligned accesses

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:   Tue, 4 May 2021 09:49:15 -0400
From:   "Theodore Ts'o" <tytso@....edu>
To:     harshad shirwadkar <harshadshirwadkar@...il.com>
Cc:     Andreas Dilger <adilger@...ger.ca>,
        Ext4 Developers List <linux-ext4@...r.kernel.org>,
        Harshad Shirwadkar <harshads@...gle.com>
Subject: Re: [PATCH] e2fsck: fix portability problems caused by unaligned
 accesses

On Tue, May 04, 2021 at 02:40:08AM -0700, harshad shirwadkar wrote:
> Hi Ted,
> 
> Thanks for the patch. While I now see that these accesses are safe,
> ubsan still complains about it the dereferences not being aligned.
> With your changes, the way we read journal_block_tag_t is now safe.
> But IIUC, ubsan still complains mainly because we still pass the
> pointer as "&tag->t_flags" and at which point ubsan thinks that we are
> accessing member t_flags in an aligned way. Is there a way to silence
> these errors?

Yeah, I had noticed that.  I was thinking perhaps of doing something
like casting the pointer to void * or char *, and then adding offsetof
to work around the UBSAN warning.  Or maybe asking the compiler folks
if they can make the UBSAN warning smarter, since what we're doing
should be perfectly safe. 

> 
> I was wondering if it makes sense to do something like this for known
> unaligned structures:
> 
> journal_block_tag_t local, *unaligned;
> ...
> memcpy(&local, unaligned, sizeof(&local));

I guess that would work too.  The extra memory copy is unfortunate,
although I suspect the performance hit isn't measurable, and journal
replay isn't really a hot path in either the kernel or e2fsprogs.
(Note that want to keep recovery.c in sync between the kernel and
e2fsprogs, so whatever we do needs to be something we're happy with in
both places.)

						- Ted