linux-kernel - Re: [PATCH 1/2] arm64:vdso: Rewrite gettimeofday into C.

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: <31273d31-57a8-2bfd-283f-025d3a88c96f@caviumnetworks.com>
Date:   Mon, 24 Apr 2017 11:49:10 -0700
From:   Andrew Pinski <apinski@...iumnetworks.com>
To:     Catalin Marinas <catalin.marinas@....com>,
        Andrew Pinski <apinski@...ium.com>
Cc:     linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/2] arm64:vdso: Rewrite gettimeofday into C.

On 4/24/2017 8:21 AM, Catalin Marinas wrote:
> On Sun, Apr 23, 2017 at 04:47:00PM -0700, Andrew Pinski wrote:
>> This allows the compiler to optimize the divide by 1000.
>> And remove the other divide.
>>
>> On ThunderX, gettimeofday improves by 32%.  On ThunderX 2,
>> gettimeofday improves by 18%.
> Is this with or without the second patch (removing the ISB)?

Hi Caralin,
   This is without the second patch that removes the ISB.  Maybe I 
should not have sent them as the same patch set to make it clear that 
way.  As I tried to make a mention of, the improvement is the compiler 
not outputting the udiv instruction for the division by 1000.    I 
should also mention I tested this patch on a softiron using GCC 4.8.5 
(SUSE's default compiler) and the performance was the same; GCC 4.8.5 
emits udiv still in this case.  I did not try a newer compiler to see 
the performance there.

Thanks,
Andrew Pinski

>