Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Mon, 09 May 2011 21:19:30 +0200
From: magnum <rawsmooth@...dband.net>
To: john-dev@...ts.openwall.com
Subject: Re: Help! mskrb5 64-32-bit builds crashing hard while
 native builds does not?

On 2011-05-08 22:37, magnum wrote:
> On 2011-05-08 17:01, magnum wrote:
>> I have a weird problem. My mskrb5 format works fine on 64-bit and
>> 32-bit, but not when building a 64-32 target and running it on 64-bit.
>>
>> It smashes the stack. When I trace it, everything is fine until I call
>> RC4_set_key() the first time, after that stack is crap (I can't step
>> inside RC4_set_key, maybe I should look for a debug version of the lib)
>
> It gets weirder. I commented the RC4 stuff out, problem disappeared but
> obviously the self test fail (no kidding). Then I replaced the OpenSSL
> calls with some unknown (slow) code I found on the 'net. The problem is
> gone and all is fine (except performance, and unknown licensing).
>
> What conclusions should I draw? I just happened to hide a problem in my
> code that is still there? It's simple code. I have checked it a hundred
> times. This drives me mad!

Tried some more with the original OpenSSL code today. The 32-bit binary 
I create from 64-bit host will fail (not crash though) on a 32-bit 
machine too. That is, it fails on mskrb5 only.

But a binary compiled on a 32-bit host works just fine on any of the two 
hosts. All tests pass. One would think it should be more or less the 
same binary, apart from slightly different versions of gcc and stuff.

I think I'm gonna assume (for the time being) this is not a problem with 
my code but something else. Is that a bad idea?

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.