gcc/config/i386/i386.c patch

[Marcus Meissner (CIP 91 - Stud) <Marcus dot Meissner at informatik dot uni-erlangen dot de>]

Hi,

I define this to be a small change, which does not require the copyright
buerocratics.

Bug in egcs/gcc:
	gcc could not detect mismatch of presence of 
	__attribute__(("stdcall")) (or "cdecl") between function declaration
	and definition.
How the patch fixes the bug:
	it looks up the attribute which is not the default (as checked by
	TARGET_RTD) and compares presence indicators between definition
	and declaration. 

Changelog:
	Handle mismatched attributes "stdcall" and "cdecl" between 
	function definition and declaration (which leads to mysterious
	stackcorruption in programs using this mixed, like WINE).

Ciao, Marcus
PS: I do not read any mailing list related to egcs.

Index: i386.c
===================================================================
RCS file: /egcs/carton/cvsfiles/egcs/gcc/config/i386/i386.c,v
retrieving revision 1.43
diff -c -3 -p -r1.43 i386.c
*** i386.c	1998/10/12 10:06:49	1.43
--- i386.c	1998/10/16 21:40:51
*************** i386_valid_type_attribute_p (type, attri
*** 622,630 ****
  
  int
  i386_comp_type_attributes (type1, type2)
!      tree type1 ATTRIBUTE_UNUSED;
!      tree type2 ATTRIBUTE_UNUSED;
  {
    return 1;
  }
  
--- 622,642 ----
  
  int
  i386_comp_type_attributes (type1, type2)
!      tree type1;
!      tree type2;
  {
+   /* Check for mismatch of non-default calling convention. */
+   int rtd = TARGET_RTD;
+   char *rtdstr = rtd ? "cdecl" : "stdcall";
+ 
+   if ((TREE_CODE(type1)!=FUNCTION_TYPE))
+     return 1;
+ 
+   /* Check for mismatched returntypes (cdecl vs stdcall). */ 
+   if ((0!=lookup_attribute (rtdstr, TYPE_ATTRIBUTES (type1))) !=
+       (0!=lookup_attribute (rtdstr, TYPE_ATTRIBUTES (type2)))
+   )
+     return 0;
    return 1;
  }