linux-kernel - Re: [PATCH v2 10/30] Add x86-specific parity functions

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, 7 Apr 2016 08:31:09 +0200
From:	Dmitry Vyukov <dvyukov@...gle.com>
To:	Andi Kleen <andi@...stfloor.org>
Cc:	zengzhaoxiu@....com, Thomas Gleixner <tglx@...utronix.de>,
	Ingo Molnar <mingo@...hat.com>,
	"H. Peter Anvin" <hpa@...or.com>,
	Denys Vlasenko <dvlasenk@...hat.com>, bp@...e.de,
	Andrew Morton <akpm@...ux-foundation.org>,
	Kees Cook <keescook@...omium.org>,
	LKML <linux-kernel@...r.kernel.org>,
	Zhaoxiu Zeng <zhaoxiu.zeng@...il.com>
Subject: Re: [PATCH v2 10/30] Add x86-specific parity functions

On Wed, Apr 6, 2016 at 9:45 PM, Andi Kleen <andi@...stfloor.org> wrote:
> zengzhaoxiu@....com writes:
>
>> From: Zhaoxiu Zeng <zhaoxiu.zeng@...il.com>
>>
>> Use alternatives, lifted from arch_hweight
>
> Is there actually anything performance critical in the kernel that uses
> parity?
>
> FWIW the arch hweight custom calling convention is a problem for LTO
> because it needs different special flags, so I usually have to disable
> it. Likely other reasonable usages, such as automatic source code
> analysis, and other tool chain based usages have similar problems.
>
> As far as I can tell both for hweight and likely for parity it is
> badly overengineering and normal calling conventions would work as well,
> and cause much less problems.
>
> So if parity is really worth adding here (which I find doubtful,
> but you may have numbers), please add it without these magic
> calling hacks.


Hweight custom calling convention caused crashes with KCOV coverage.
We had to disable instrumentation of the file.