Re: Fix EM_S390 value

[Andreas Jaeger <aj at suse dot de>]

Hi Martin,

schwidefsky@de.ibm.com writes:

> Hi Andreas,
> >Newer binutils use 22 as value for EM_S390.  I hope this is the
> >official value.
> It is the value I got from Dave Prosser.
> 
> >Programs compiled with the new binutils will fail since dl-machine
> >checks only for the old value.  The appended patch corrects the value
> >and glibc checks now for both values so that glibc will work with old
> >and new binutils.
> >
> >Martin, is this really the official value?  Is the appended patch ok?

> The patch looks ok. To get the glibc-2.2.2 working you'll want to have
> another patch for pt-initfini.c. The alignment at the end of the function
> prologs adds zeros to the instructions stream. Needless to say that this
> kills every program that uses the new pthreads.

Ok, I've added both patches now.

> By the way I made a glibc-2.2.2 patch for 31bit and 64bit linux/390. It
> is big (595818 bytes). What do we need to get it into the official source
> tree? A new FSF assignment or can we live with the old one?

AFAIK we can live with the old assignment but you need to CC your
patch to the special gnu.org address for those assignments.

Please send the changes (together with ChangeLog entries) in several,
smaller chunks and check first that all patches conform to the GNU
coding standards.  This time I will not reformat the patches like I
did last time.

I suggest to split the patch in at least the following thunks:
- Changes for generic code (this is a must, they should come
  separatly).  Please explain in detail why each patch is needed.
- Changes for the existing 31bit S390.
- The new 64 bit port.

Andreas
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.inka.de
    http://www.suse.de/~aj