linux-kernel - Re: [PATCH v2] x86/mm/kaslr: Use _ASM

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, 26 Apr 2017 14:29:13 -0700
From:   Greg Hackmann <ghackmann@...gle.com>
To:     hpa@...or.com, Kees Cook <keescook@...omium.org>
Cc:     Matthias Kaehlcke <mka@...omium.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>,
        "x86@...nel.org" <x86@...nel.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Grant Grundler <grundler@...omium.org>,
        Michael Davidson <md@...gle.com>
Subject: Re: [PATCH v2] x86/mm/kaslr: Use _ASM_MUL macro for multiplication

On 04/26/2017 02:24 PM, hpa@...or.com wrote:
>>> This really feels like a "fix your compiler" issue.
>>
>> We already use the other forms, what's so bad about adding mul too?
>> And if this lets us build under clang, all the better.
>>
>> -Kees
>
> It's not bad per se, but if this doesn't eventually gets fixed in clang we'll have no end of this crap.
>

AIUI the "problem" is that clang is spilling mix_const into memory 
rather than assigning it to a register.  This is perfectly legal since 
mix_const has a constraint of "rm".  But mul needs a suffix when the 
input is a memory location, since it can't infer the multiplication 
width from the input operand anymore.

You get the same error message with gcc if you force it to use a memory 
location, by narrowing the constraint from "rm" to "m".