Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 7 May 2014 05:39:37 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: autoconf

Hi magnum,

On Sun, May 04, 2014 at 11:57:06AM +0200, magnum wrote:
> Just for a heads-up: We (mainly Jim as of yet) are trying to add 
> autoconf to bleeding-jumbo.
> 
> Some discussions:
> https://github.com/magnumripper/JohnTheRipper/issues/575
> 
> Topic branch:
> https://github.com/magnumripper/JohnTheRipper/commits/autoconf

Thanks!  Yes, we might have to bite that bullet in jumbo.

A quick comment: I think HAVE_SSE*, HAVE_AVX*, and HAVE_XOP are
undesirable.  I think it is better to continue to check the compiler's
__SSE2__, etc. macros instead, and use -march=native unless requested
otherwise (and if so use -march=whatever-is-requested-via-configure, but
still don't alter any HAVE_* macros).  Why would one want to build code
that possibly uses e.g. SSE2 anyway (because the compiler can, in the
code it generates!), but not enable it for intrinsics and for our
explicit assembly?

The one place where I think we may want to alter arch flavor settings
with the compiler is when building john.c vs. the rest of JtR, to allow
for arch flavor fallbacks.  john.spec on Owl does this.  But I guess
this is not currently under scope for your autoconf'ication, so your
HAVE_SSE*, etc. probably don't make sense even in this context.

Am I missing something?

Alexander

Powered by blists - more mailing lists

Your e-mail address:

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