[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: gcc-2.95.1?



The not-building he was referring to was from the debian source
package, which still has six or so patches to the source, even after
some were applied upstream.

And the binutils hack is not a good idea; I'm not the person to ask why
not, though.

On Mon, Sep 06, 1999 at 01:50:22PM -0700, C.M. Connelly wrote:
> "GJ" == Greg Johnson <gjohnson@physics.clarku.edu>
> "DJ" == Daniel Jacobowitz <dan@debian.org> writes:
> 
> GJ> Is someone working on building gcc-2.95.1 for powerpc?  It
> GJ> doesn't build out of the box.  It looks like some of the
> GJ> patches in the source package have been applied upstream.
> GJ> After removing these from debian/rules.patch, it starts to
> GJ> build.  However, my ld doesn't like "-m elf32ppclinux".  Is
> GJ> binutils out of date, or should it be changed back to elf32ppc?
> 
> DJ> Binutils is out of date.  You need to use an experimental one;
> DJ> I uploaded a working one today, and it should be available
> DJ> tomorrow afternoon, in project/experimental/.
> 
> Hmm, gcc v. 2.95.1 built fine on our machine with binutils
> 2.9.1.0.25-2 and one change to one line:
> 
> --- gcc-2.95.1/gcc/config/rs6000/linux.h.orig	Mon May 31 02:37:25 1999
> +++ gcc-2.95.1/gcc/config/rs6000/linux.h	Thu Aug 19 11:53:55 1999
> @@ -35,9 +35,9 @@
>  #undef	CPP_OS_DEFAULT_SPEC
>  #define CPP_OS_DEFAULT_SPEC "%(cpp_os_linux)"
>  
>  #undef LINK_SPEC
> -#define LINK_SPEC "-m elf32ppclinux %{G*} %{shared:-shared} \
> +#define LINK_SPEC "-m elf32ppc %{G*} %{shared:-shared} \
>    %{!shared: \
>      %{!static: \
>        %{rdynamic:-export-dynamic} \
>        %{!dynamic-linker:-dynamic-linker /lib/ld.so.1}} \
> 
> 
> You might also want to make another change that fixes a big with
> -ffixed support on the PowerPC -- without it, -ffixed is *totally*
> broken.  Not many people use this feature, but those who do *need*
> this patch.  (For various reasons, this patch never made it into the
> main source tree, despite having been posted two years ago.  It's just
> been resubmitted and will (hopefully) make it in soon.)
> 
> 
> --- gcc-2.95.1/gcc/config/rs6000/rs6000.c.orig	Thu Aug 19 01:21:45 1999
> +++ gcc-2.95.1/gcc/config/rs6000/rs6000.c	Fri Sep  3 00:54:28 1999
> @@ -3240,10 +3240,12 @@
>  {
>    int first_reg;
>  
>    /* Find lowest numbered live register.  */
> -  for (first_reg = 13; first_reg <= 31; first_reg++)
> -    if (regs_ever_live[first_reg])
> +  for (first_reg = 13 + FIXED_R13; first_reg <= 31; first_reg++)
> +    if (regs_ever_live[first_reg]
> +	&& !call_used_regs[first_reg]
> +	&& !fixed_regs[first_reg])
>        break;
>  
>    if (profile_flag)
>      {
> 
> +=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
>  Behind the counter a boy with a shaven head stared vacantly into space, 
>  a dozen spikes of microsoft protruding from the socket behind his ear.
> +=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
>    C.M. Connelly               c@eskimo.com                   SHC, DS
> +=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+ 
> 
> 
> 
> 
> --  
> To UNSUBSCRIBE, email to debian-powerpc-request@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
> 


Dan

/--------------------------------\  /--------------------------------\
|       Daniel Jacobowitz        |__|        SCS Class of 2002       |
|   Debian GNU/Linux Developer    __    Carnegie Mellon University   |
|         dan@debian.org         |  |       dmj+@andrew.cmu.edu      |
\--------------------------------/  \--------------------------------/


Reply to: