Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Tue, 12 Feb 2013 20:13:03 +0100
From: magnum <>
Subject: Re: sha2 in unstable vs bleeding

OK, no problem I'll revert and re-do it the other way :)


On 12 Feb, 2013, at 20:10 , "jfoug" <> wrote:

> Magnum,
> I was totally backwards.  The newest version was in bleeding, not unstable.
> Sorry, my bad.  I was reading the diff data wrong, I had the wrong version
> on the left.
> So what I said was completely backwards. The proper version was in bleeding,
> and should be moved to unstable.    
> I do not think this will change the alias warning, however.
> Jim.
> From: magnum On 12 Feb, 2013, at 18:30
>> The sha2.h in unstable has been reduced in complexity, and should be 
>> the one used.  Also, the sha2.c in unstable is the most current.  The 
>> version is bleeding was one of the earlier versions, before I lined up 
>> the macros, between the 32 bit and 64 bit versions.
>> So in other words, copy unstable's sha2.h and sha2.c to bleeding.
>> Jim.
>> From: magnum  Sent: Tuesday, February 12, 2013 2:03
>>> Jim,
>>> sha2.c and sha2.h are slightly different in unstable vs bleeding. Why?
>> Which version is better?
>>> The unstable version seem to be a couple days newer and the header 
>>> has
>> #else clauses that we probably want and that bleeding lacks.
> Okay, fixed now.
> BTW I got strict-aliasing warnings from gcc 4.7.2 on ppc for the BE versions
> of OUTBE32() and OUTBE64() despite you are already using a proper union. I
> can't see why. I'll try using m.wlen as input to the macro and cast it the
> other way round, but it really should not be needed.
> magnum

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.