linux-kernel - Re: [tip:x86/urgent] x86/cpufeature: Add AVX512_4VNNIW and AVX512

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]

Message-ID: <1476697859.27884.21.camel@intel.com>
Date:   Mon, 17 Oct 2016 09:55:07 +0000
From:   "Luc, Piotr" <Piotr.Luc@...el.com>
To:     "hpa@...or.com" <hpa@...or.com>, "bp@...en8.de" <bp@...en8.de>
CC:     "mingo@...nel.org" <mingo@...nel.org>,
        "peterz@...radead.org" <peterz@...radead.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "torvalds@...ux-foundation.org" <torvalds@...ux-foundation.org>,
        "tglx@...utronix.de" <tglx@...utronix.de>,
        "linux-tip-commits@...r.kernel.org" 
        <linux-tip-commits@...r.kernel.org>,
        "jpoimboe@...hat.com" <jpoimboe@...hat.com>,
        "brgerst@...il.com" <brgerst@...il.com>,
        "luto@...nel.org" <luto@...nel.org>,
        "Hansen, Dave" <dave.hansen@...el.com>,
        "dvlasenk@...hat.com" <dvlasenk@...hat.com>
Subject: Re: [tip:x86/urgent] x86/cpufeature: Add AVX512_4VNNIW and
 AVX512_4FMAPS features

On Mon, 2016-10-17 at 00:42 +0200, Borislav Petkov wrote:
> On Sun, Oct 16, 2016 at 11:42:26AM -0700, hpa@...or.com wrote:
> > 
> > It's needlessly adding complexity for no reason, at least for the
> 
> What complexity? The init_scattered_cpuid_features() version is a
> trivial patch in comparison to the current version.
> 
> > 
> > leaves that are going to add bits over time.
> 
> Sure, except they don't get added or we don't need them or whatever,
> and
> we end up with only a small number of bits actually being used.
> 
> I don't mind moving them to x86_capability later, when a high
> percentage
> of the respective leaf is actually being used but not for a couple of
> bits. That's just waste.
> 
> > 
> > The x86_capability array is not an expensive resource.
> 
> 0.1% here, 0.1% there, the creeping bloat thing.
> 
> And again, the init_scattered_cpuid_features() hunk is much smaller.
> 
I agree, the scattered solution reduces data segment footprint in case
many cores.

Regards,
Piotr