INTERNAL_SYSCALL and function calls in arguments

[Ulrich Drepper]

Andreas Schwab wrote:

> Since the change only introduces temporary variables which the compiler
> would create internally anyway, the compiler is perfectly capable to
> optimize them away, and in all cases except sched_setaffinity the
> generated code is virtually identical.

I haven't asked about theoretical results.  You must assume I'm stupid
and don't know that this is possible.  What are the concrete results?

-- 
➧ Ulrich Drepper ➧ Red Hat, Inc. ➧ 444 Castro St ➧ Mountain View, CA ❖