Fix MIPS waitid build [committed]

[Joseph Myers]

On Wed, 10 Dec 2014, Rich Felker wrote:

> Wouldn't it be preferable just to fix whatever the underlying cause
> is? It sounds like some ugly and fragile hack is being used to
> position the syscall arguments on the stack and I wouldn't be
> surprised if it's only working by chance. I've seen other bad code

A single asm adjusts the stack, puts arguments there, does the syscall and 
restores the stack.  The frame pointer is needed to avoid the stack 
pointer adjustment breaking unwind info.

-- 
Joseph S. Myers
joseph@codesourcery.com