[OpenAFS-devel] Re: AIX 6.1 support - initial patchset

[Niklas Edmundsson]

On Mon, 25 Aug 2008, Niklas Edmundsson wrote:

Hmm. The first mail probably got stuck in a moderator queue due to its 
size, seems like best practice is mailing patches to openafs-bugs and 
refer to the RT url here ;)

>> I can now read from AFS, it was mostly build system issues with the spice 
>> of a stupid bug or two in the openafs sources.
>
> Correction: I can now read from AFS, but ALL 64bit processes fail when they 
> try to do an ioctl (for example calling fopen() has that effect).

<snip>

I found the problem, I had missed a "case rs_aix5*" which led to the 
AIX4 afs.exp being applied, which obviously didn't work ;)

Look at http://rt.central.org/rt/Ticket/Display.html?id=113638 for the 
patch, use the newest one and param-rs_aix61.h.

Please commit to trunk, it seems to work even though it hasn't seen a 
lot of testing yet :)

> In any case, I include the patchset for trunk which got me this far. 
> param.rs_aix61.h goes into src/config/.
> The patch consists roughly of the following fixes:
>
> - Make build system recognise AIX 6.1
> - Convince build system to only build 64bit kernel modules on AIX 6.1,
>  there is no 32bit kernel on AIX 6.
> - Misc build fixes, trunk obviously haven't worked on AIX for a while.
> - Fix cfgexport to allocate enough memory (the realloc stuff didn't
>  work).
> - Don't ignore building aklog.
> - Fix the AIX rc.afs startup script to correctly determine 32/64-bit
>  kernels on all AIX versions.
>
> /Nikke
>


/Nikke
-- 
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
  Niklas Edmundsson, Admin @ {acc,hpc2n}.umu.se     |    nikke@hpc2n.umu.se
---------------------------------------------------------------------------
  Do Hard Drives sneeze when they catch a virus?
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=