Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Sun, 20 Jul 2014 01:18:09 +0200
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: CUDA Test Results on ARM w/CUDA

On 2014-07-18 04:08, Lukas Odzioba wrote:
> 2014-07-18 3:22 GMT+02:00 Lukas Odzioba <lukas.odzioba@...il.com>:
>> For some reason this format fails with sm_20 and grater when password is
> empty.
>
> This is caused by memcpy function itself.
> The following patch seems to fix the problem, but sice it has performance
> impact I am not convinced to using it in production code without SM
> detection:
>
> --- cryptmd5.cu 2014-07-18 01:50:16.412332266 +0000
> +++ cryptmd5-2.cu 2014-07-18 01:50:48.020313773 +0000
> @@ -35,7 +35,8 @@ __device__ void ctx_update(md5_ctx * ctx
>    uint8_t *dest = &ctx->buffer[*ctx_buflen];
>    uint8_t *src = (uint8_t *) string;
>    *ctx_buflen += len;
> - memcpy(dest, src, len);
> + if(len)
> + memcpy(dest, src, len);
>   }
>
>

Did you see a significant drop? I see no impact at all here. I'll commit 
this for now.

magnum

Powered by blists - more mailing lists

Your e-mail address:

Powered by Openwall GNU/*/Linux - Powered by OpenVZ