Bug 52532 – gcc-3.4.0 , binutils 2.15.91.0.1 can't compile kernel

• Product: Gentoo Linux
• Status: VERIFIED WONTFIX

[Description Philippe Trottier (RETIRED) 2004-05-31 02:02:10 UTC]

I think this is a show stopper for 2004.2, I haven't made any deep research on this yet. We could use gcc-3.3.3_(something) just to make the kernel, but having just one compiler would be nice.

make
  HOSTCC  scripts/basic/fixdep
  HOSTCC  scripts/basic/split-include
  HOSTCC  scripts/basic/docproc
  HOSTCC  scripts/genksyms/genksyms.o
  HOSTCC  scripts/genksyms/lex.o
  HOSTCC  scripts/genksyms/parse.o
  HOSTLD  scripts/genksyms/genksyms
  HOSTCC  scripts/conmakehash
  HOSTCC  scripts/kallsyms
  HOSTCC  scripts/mk_elfconfig
  MKELF   scripts/elfconfig.h
  HOSTCC  scripts/file2alias.o
  HOSTCC  scripts/modpost.o
  HOSTCC  scripts/sumversion.o
  HOSTLD  scripts/modpost
  HOSTCC  scripts/pnmtologo
  HOSTCC  scripts/bin2c
make[1]: `arch/ppc64/kernel/asm-offsets.s' is up to date.
  CHK     include/linux/compile.h
  UPD     include/linux/compile.h
  CC      init/version.o
  LD      init/built-in.o
  HOSTCC  usr/gen_init_cpio
  CPIO    usr/initramfs_data.cpio
  GZIP    usr/initramfs_data.cpio.gz
  AS      usr/initramfs_data.o
  LD      usr/built-in.o
  AS      arch/ppc64/kernel/idle_power4.o
arch/ppc64/kernel/idle_power4.S: Assembler messages:
arch/ppc64/kernel/idle_power4.S:69: Error: Unrecognized opcode: `dssall'
make[1]: *** [arch/ppc64/kernel/idle_power4.o] Error 1
make: *** [arch/ppc64/kernel] Error 2

[Comment 1 Philippe Trottier (RETIRED) 2004-05-31 02:04:32 UTC]

even with gcc-3.3.3 it fails at the same place so it must be a bintuils issue... Ill try to find at what version this problem starts.

[Comment 2 Philippe Trottier (RETIRED) 2004-05-31 05:09:50 UTC]

2.15.90.0.3-r3 does go thrue the kernel, still can't check if it makes a bootable kernel ...

[Comment 3 Tom Gall (RETIRED) 2004-05-31 20:47:20 UTC]

this is a known issue and is cause by a change that was made in both gcc and binutils. Unfortunately this requires folks like us to have to watch how far forward we go.  If you go to use binutils 2.15 you've got to be on a bleeding edge 3.4 as well. (Mainline cvs should be fine) ... Alanm did some changes for -many and other ppc/ppc64 options.

The bug below is because binutils thinks it doesn't have to emit altivec instructions and it really does.

Going to mark this effectively won't fix but I will update the bug with the exact gcc / binutils combo would should be "productive". 

[Comment 4 Tom Gall (RETIRED) 2004-06-12 20:43:14 UTC]

I've verified this does work. 3.4 + 2.15 binutils does yeild a successful kernel build. (didn't test boot it... but it's good ... it's gotta be... I have faith! :-)

[Comment 5 Tom Gall (RETIRED) 2004-06-12 20:43:35 UTC]

sticking a fork in it