linux-kernel - Re: [PATCH 4.14 17/89] futex: Prevent overflow by strengthen input validation

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:   Thu, 25 Jan 2018 16:21:51 +0100
From:   Jiri Slaby <jslaby@...e.cz>
To:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Cc:     Thomas Gleixner <tglx@...utronix.de>, linux-kernel@...r.kernel.org,
        stable@...r.kernel.org, Li Jinyue <lijinyue@...wei.com>,
        peterz@...radead.org, dvhart@...radead.org,
        "torvalds@...ux-foundation.org" <torvalds@...ux-foundation.org>
Subject: Re: [PATCH 4.14 17/89] futex: Prevent overflow by strengthen input
 validation

On 01/25/2018, 04:12 PM, Greg Kroah-Hartman wrote:
> On Thu, Jan 25, 2018 at 03:47:32PM +0100, Jiri Slaby wrote:
>> On 01/25/2018, 03:30 PM, Thomas Gleixner wrote:
>>> So what's the problem?
>>
>> The problem I see is that every stable kernel now requires updated
>> strace with their commit from yesterday to build correctly. In
>> particular, the new stable kernels cause rpm build failures of strace in
>> all our distros (based on those stable kernels). Sure, we can patch
>> strace in every distro every nth kernel update, but it's mere
>> impractical. Kernel should not break userspace, right?
> 
> Well, when userspace is doing something stupid... :)

No doubt... But does that mean we no longer maintain the "no userspace
breakage even if it is stupid" rule?

>> BTW why was the patch applied to stable? We actually do pass
>> -fno-strict-overflow.
> 
> The same reason it was applied upstream, it fixes a reported
> issue.
> 
> Does that mean that all UBSAN overflow error reports are not valid
> because of how we build the kernel?

IMO yes, because with the option, signed overflow is not undefined.

In the long term, it would be nice to get rid of *all* signed integer
overflows and kill the compiler option from Makefile. Therefore the
fixes are indeed very valid in upstream.

thanks,
-- 
js
suse labs