Re: bring mips*-linux back to a buildable state

[Andreas Jaeger <aj at suse dot de>]

Alexandre Oliva <aoliva@redhat.com> writes:

> On Jul 20, 2004, Andreas Jaeger <aj@suse.de> wrote:
>
>> Alexandre Oliva <aoliva@redhat.com> writes:
>>> On Jul 20, 2004, Andreas Jaeger <aj@suse.de> wrote:
>>> 
>>>> Alexandre Oliva <aoliva@redhat.com> writes:
>>>>> On Jul 19, 2004, Andreas Jaeger <aj@suse.de> wrote:
>>>>> 
>>>>>> Alexandre Oliva <aoliva@redhat.com> writes:
>>>>>>> These patches gets all 3 ABIs of mips64-linux to build and work again.
>>>>> 
>>>>>> Can you send them separtly, so that it's easier to comment on them?
>>>>> 
>>>>> The kernel tree I was using didn't define __NR_semtimedop nor
>>>>> __NR_ipc, so linux/semtimedop.c didn't work.  I figured it would be
>>> 
>>>> The current 2.6 kernel has those in asm-mips/unistd.h.  Or where is it
>>>> missing right now?
>>> 
>>> How about asm-mips64?  IIRC the syscalls were missing for n32 and n64,
>>> but not for o32.
>
>> There's no such directory in my sources,
>
> Aah, I didn't realize the mips64 headers had been merged into mips in
> the kernel.  I was still using an old kernel/libc ABI headers tree.
>
> Anyhow, I see __NR_semtimedop defined for n32 and n64, but not for
> o32, that still uses __NR_ipc.  And __NR_ipc is not defined for n32
> nor n64, so the default linux/semtimedop.c won't work for them.  We
> could just add an entry to mips/mips64/syscalls.list to fix the
> problem in glibc mainline, but this means we won't support kernels
> that still didn't have semtimedop defined for the 64-bit mips64 ABIs.
> I suppose this is not a big deal, and we could go with that instead of
> this patch.  Any preferences?

Let's go with current 2.6 kernels, so a patch to syscalls.list would
be ok,

Andreas
-- 
 Andreas Jaeger, aj@suse.de, http://www.suse.de/~aj
  SUSE Linux AG, Maxfeldstr. 5, 90409 Nürnberg, Germany
   GPG fingerprint = 93A3 365E CE47 B889 DF7F  FED1 389A 563C C272 A126

Attachment: pgp00000.pgp
Description: PGP signature