make tests failure

[Darren Tucker]

Andreas Fehr wrote:
> - This is a test on the localhost, so I guess, there is no man in the 
>   middle attack (hosts file is setup correctly, localhost points to 
>   127.0.0.1)
> 
> - I don't want to use protocol 1 anyway, so how can I skip the test?

There's no easy way, but if you really want to you can do "make tests 
LTESTS="[list of tests]".  A better approach is to fix the problem (see 
below).

> - I copied the known host file from my already installed ssh 
>   (~/.ssh/known_hosts) to ./regress/ but that didn't help
> 
> - Just to be sure, I checked the 4.1 source of my current installation 
>   and it came up with the same error, but I can't remember seeing this 
>   error with the prev. installation. So I stopped the running sshd, 
>   removed the /etc/ssh directory and run the test again with the same 
>   result.
> 
> - I'm running uname -a: Linux linuxbox 2.6.12.5 #1 Tue Aug 16 08:57:47 
>   CEST 2005 ppc unknown unknown GNU/Linux

If you're using OpenSSL 0.9.7g then there's a bug in the PPC assember 
code which can cause this.  I don't know if any other versions are affected.

The attached patch fixes this (I didn't write it, I just happen to have 
it handy), or alternatively you can rebuild OpenSSL without assembler 
optimizations.

-- 
Darren Tucker (dtucker at zip.com.au)
GPG key 8FF4FA69 / D9A3 86E9 7EEE AF4B B2D4  37C9 C982 80C7 8FF4 FA69
     Good judgement comes with experience. Unfortunately, the experience
usually comes from bad judgement.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: openssl-0.9.7g-ppcasm.patch
Url: http://lists.mindrot.org/pipermail/openssh-unix-dev/attachments/20051012/00230784/attachment.ksh