Xtensa syscalls (Was: Re: 2.6.12-rc5-mm1)

[Christian Zankel <chris@zankel.net>]

Arnd Bergmann wrote:
>>>Chris, are there any existing binaries that rely on your implementations
>>>of old_mmap, sys_fork, sys_vfork, sys_olduname or sys_ipc and need to
>>>work with future kernels? Otherwise, you should probably drop these.
>>>For sys_ipc, you would need to add the subcalls directly to the table,
>>>like parisc does.
> Hmm, xtensa is now in -rc1, with the obsolete syscalls still in there,
> so I guess this about the last chance to correct the ABI. Applying the
> patch obviously breaks all sorts of user space binaries and probably
> also requires the appropriate changes to be made to libc.

I have to admit, the -rc1 caught me a bit by surprise; I have a few 
patches pending that I want to send out today.

The question is, if we had to break glibc compatibility, shouldn't we 
use the opportunity to clean-up the syscall list? It was copied from 
MIPS and, thus, has inherited a lot of legacy from there. As a new 
architecture, maybe we should even go as far as removing all ni-syscalls 
and start fresh?

> On the other hand, if a decision is made to keep the broken interface,
> it should at least be a conscious one instead of an oversight.

I will try out your patch and see if there are any obvious problems.

Thanks,
~Chris