[PATCH]: Fix PR c/12372

[Eric Botcazou]

> I tend to think that line is in the wrong place, but I'm not completely
> sure.

No, it appears to be a collision between two patches

Tue Jul 23 12:16:58 2002  J"orn Rennecke <joern.rennecke@superh.com>

	* calls.c (emit_library_call_value_1): If
	FUNCTION_ARG_PASS_BY_REFERENCE is true, pretend this is neither
	libcall, const call nor pure call.

and

2002-09-26  David S. Miller  <davem@redhat.com>

	PR optimization/7335
	* calls.c (emit_library_call_value_1): Passing args by reference
	converts a CONST function into a PURE one.


David's patch was needed on the 3.2 branch, but a no-op on the (then) 
mainline.

> I'm still pondering whether or not to clobber the slots.  I believe
> that functions are generally allowed to clobber their arguments.  The
> most serious problem that I have noticed with the patch is that reload
> doesn't eliminate the occurences of the frame pointer in the call usage
> information.  So, the information is wrong after reload ;(

Yes, and this MEM in the uses list is a bit weird.

> You might see if this resolves the sparc failure.

It does.

-- 
Eric Botcazou