linux-kernel - Re: Re: [PATCH 1/1] Preventive patch in the proc file-system to handle NULL check.

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: <20180817033606.GB6515@ZenIV.linux.org.uk>
Date:   Fri, 17 Aug 2018 04:36:06 +0100
From:   Al Viro <viro@...IV.linux.org.uk>
To:     Srikanth Korangala Hari <srikanth.h@...sung.com>
Cc:     Luis Chamberlain <mcgrof@...nel.org>,
        "keescook@...omium.org" <keescook@...omium.org>,
        "adobriyan@...il.com" <adobriyan@...il.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-fsdevel@...r.kernel.org" <linux-fsdevel@...r.kernel.org>
Subject: Re: Re: [PATCH 1/1] Preventive patch in the proc file-system to
 handle NULL check.

On Fri, Aug 17, 2018 at 08:51:42AM +0530, Srikanth Korangala Hari wrote:
> > Thanks for the patch! Do you have a reproducer or is this theoretical?
> > This will affect if it should go to stable or not.
> 
> Dear Luis, this is theoretical as I observed in most of the call's to api - "proc_mkdir" the NULL check is being done. Hence I thought of adding one here.

Realistically, if you get allocation failures that early in the boot,
oops is the least of your problems - it won't get through mounting
the root or lauching init (or unpacking initramfs, etc.)

Sure, might as well check it there - nothing wrong with that, but do
keep in mind that
	* it's very certain to end up in panic() very shortly afterwards,
no matter what
	* the odds of exhausting memory (and that would be extremely
low-memory setup) precisely at that point (i.e. even getting to
proc_root_init()) are not high.

Might be interesting to try lower and lower mem=... values passed to the
kernel in attempt to step into this one; I wouldn't put large odds on
being able to hit precisely that place, but it would be educational anyway.