🚀 go-pugleaf

Good day,

This morning, I sent over a bug report regarding GCC's handling
of variable-length argument lists.  Since then I have dug into the
source code and come up with a fix to the problem.  Turns out
the function `expand_builtin()' was moving the instruction preceding
the __builtin_saveregs call also when rearranging the instructions.

Below find a patch for this bug.  Hopefully, it will get incorporated
into future releases of GCC.

	1) GCC Version:		1.34
	2) Processor Type:	SPARC (SUN-4)
	3) Operating System:    SUNOS 4.0.3
	4) Command Line:	gcc -O	(optimized compilation)
	5) md:			sparc.md
	6) tm.h:		tm-sparc.h
	7) Problem:		GCC occasionally corrupts variable length
				arg lists by placing an instruction before
				the __builtin_saveregs call which steps
				on `i0' through `i5'.

---------------------------------------------------------------------------
*** expr.c-old	Tue Jul 11 13:34:32 1989
--- expr.c	Tue Jul 11 13:32:23 1989
***************
*** 3184,3190 ****
  	/* Now really call the function.  `expand_call' does not call
  	   expand_builtin, so there is no danger of infinite recursion here.  */
  	expand_call (exp, target, 1);
! 	reorder_insns (last, get_last_insn (), get_insns ());
        }
        
      case BUILT_IN_ALLOCA:
--- 3184,3192 ----
  	/* Now really call the function.  `expand_call' does not call
  	   expand_builtin, so there is no danger of infinite recursion here.  */
  	expand_call (exp, target, 1);
! 	last = NEXT_INSN(last);
! 	if( last )
! 	    reorder_insns (last, get_last_insn (), get_insns ());
        }
        
      case BUILT_IN_ALLOCA:
---------------------------------------------------------------------------

Jon Olson, Modular Mining Systems
	   1210 E. Pennsylvania St.
	   Tucson, AZ 85714

USENET:     {allegra,cmcl2,hao!noao}!arizona!modular!olson
INTERNET:   modular!olson@arizona.edu