[OpenAFS] SuSE 10.1 kernel 2.6.16.13-4/64Bit compile error - OpenAFS 1.5.2

[Karsten Künne]

On Tuesday 11 July 2006 19:48, Jeffrey Hutzelman wrote:
> On Monday, July 10, 2006 02:08:14 PM -0700 ted creedon
>
> <tcreedon@easystreet.com> wrote:
> > Is there any particular branch for 64 bit?
> > Tedc
> > -----Original Message-----
> > From: openafs-info-admin@openafs.org
> > [mailto:openafs-info-admin@openafs.org] On Behalf Of Derek Atkins
> > Sent: Monday, July 10, 2006 11:34 AM
> > To: tcreedon@easystreet.com
> > Cc: openafs-info@openafs.org
> > Subject: Re: [OpenAFS] SuSE 10.1 kernel 2.6.16.13-4/64Bit compile error -
> > OpenAFS 1.5.2
> >
> > "ted creedon" <tcreedon@easystreet.com> writes:
> >> I'm not at all sure what this error is when building OpenAFS 1.5.2.
> >>
> >> Any clues?
> >>
> >> Tedc
> >
> > [snip]
> >
> >> /data/openafs-1.5.2/src/afs/LINUX/osi_machdep.h:55:2: error: #error Not
> >
> > sure
> >
> >> what to do about rlim (should be in the Linux task struct
> >
> > [snip]
> >
> > This has been discussed on the -devel list a number of times.
> > The kernel tests don't work for certain bleeding-edge kernels.
>
> It's true that for bleeding-edge kernels, sometimes there's something new
> that we need to test for.  However, OpenAFS does build on 2.6.16 kernels,
> and a failure in the rlim test almost always means you've done something
> wrong which prevents _any_ of the compile-time kernel tests from compiling.
> It constantly amazes me that whenever anyone reports this, the first
> response of people on this list who really ought to know better is to try
> to figure out what has changed in the kernel.
>
> Instead, the _first_ things you should be looking for are
> - is configure using the same compiler that was used for your kernel?
> - do you have the linux kernel headers installed, and in the right place?
> - if so _why_ did configure fail to compile the tests?
>
> Look in config.log, find the first occurrance of the string "rlim", and
> look at the compiler errors around there.  Not the messages from the
> configure script, and not the test program that failed.  What you're
> interested in is the compiler errors.
>

I had the same or at least a similar error on SUSE 10.1 with 1.4.1 and it 
turned out that SUSE requires a small change so that OpenAFS can find the 
kernel headers. In the directory /lib/modules/2.6.16.13-4-smp (or -default if 
you don't run the smp kernel) remove the link "build" and do "ln -s source 
build" and also "mkdir extra" (otherwise "make install" will bomb). That 
fixed it for me. And don't forget to run "make cloneconfig; make 
modules_prepare" in the kernel source directory.

It had nothing to do with bleeding-edge kernels.


Karsten.
-- 
A CONS is an object which cares.
		-- Bernie Greenberg.