Re: 2.6.0-test1-mm1

[Andrew Morton <akpm@osdl.org>]

"Barry K. Nathan" <barryn@pobox.com> wrote:
>
> >  Make ppc build
> 
>  Really? ;)
> 
>  More seriously, that patch is good and necessary, but I think something
>  else in -mm is breaking the compile (this is gcc 2.95.3):
> 
>    CC      arch/ppc/kernel/irq.o
>  In file included from include/linux/fs.h:14,
>                   from include/linux/mm.h:14,
>                   from include/asm/pci.h:8,
>                   from include/linux/pci.h:672,
>                   from arch/ppc/kernel/irq.c:41:
>  include/linux/kdev_t.h: In function `to_kdev_t':
>  include/linux/kdev_t.h:101: warning: right shift count >= width of type

Well you would appear to be the first person who has tested a -mm kernel on
ppc since mid-April or earlier.

As soon as Viro returns, that code hits Linus's tree.  Oh well, can't say
they weren't warned.

Try this:

--- 25/include/asm-ppc/posix_types.h~a	2003-07-16 03:55:34.000000000 -0700
+++ 25-akpm/include/asm-ppc/posix_types.h	2003-07-16 03:55:51.000000000 -0700
@@ -7,7 +7,7 @@
  * assume GCC is being used.
  */
 
-typedef unsigned int	__kernel_dev_t;
+typedef unsigned long long	__kernel_dev_t;
 typedef unsigned long	__kernel_ino_t;
 typedef unsigned int	__kernel_mode_t;
 typedef unsigned short	__kernel_nlink_t;



>  arch/ppc/kernel/irq.c: At top level:  
>  arch/ppc/kernel/irq.c:575: braced-group within expression allowed only
>  inside a function

Bill?