bootstrap broken: darwin, hppa ..... Re: PR 10776 II

[Jan Hubicka]

> Jan Hubicka wrote:
> 
> 
> >I apologize for the breakage.  It seems to me that the attached patch
> >will fix it.  I got fooled by fact that the code just preceeding the
> >hunk use CSElib to modify the memory, but that code is probably wrong as
> >it is dead anyway.
> >Can you please test this for me?
> 
> The build continues now, let ya know more.
> Thx,
Good.  Here is version of patch that actually doesn't kill the
optimization.  If it would pass for you, I will install it as obivous.
Honza

Index: sched-deps.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/sched-deps.c,v
retrieving revision 1.64
diff -c -3 -p -r1.64 sched-deps.c
*** sched-deps.c	14 Jan 2004 22:43:53 -0000	1.64
--- sched-deps.c	15 Jan 2004 09:30:29 -0000
*************** sched_analyze_1 (struct deps *deps, rtx 
*** 545,551 ****
  	  cselib_lookup (XEXP (t, 0), Pmode, 1);
  	  XEXP (t, 0) = cselib_subst_to_values (XEXP (t, 0));
  	}
!       XEXP (t, 0) = canon_rtx (XEXP (t, 0));
  
        if (deps->pending_lists_length > MAX_PENDING_LIST_LENGTH)
  	{
--- 545,551 ----
  	  cselib_lookup (XEXP (t, 0), Pmode, 1);
  	  XEXP (t, 0) = cselib_subst_to_values (XEXP (t, 0));
  	}
!       t = canon_rtx (t);
  
        if (deps->pending_lists_length > MAX_PENDING_LIST_LENGTH)
  	{
*************** sched_analyze_2 (struct deps *deps, rtx 
*** 687,693 ****
  	    cselib_lookup (XEXP (t, 0), Pmode, 1);
  	    XEXP (t, 0) = cselib_subst_to_values (XEXP (t, 0));
  	  }
!         XEXP (t, 0) = canon_rtx (XEXP (t, 0));
  	pending = deps->pending_read_insns;
  	pending_mem = deps->pending_read_mems;
  	while (pending)
--- 687,693 ----
  	    cselib_lookup (XEXP (t, 0), Pmode, 1);
  	    XEXP (t, 0) = cselib_subst_to_values (XEXP (t, 0));
  	  }
! 	t = canon_rtx (t);
  	pending = deps->pending_read_insns;
  	pending_mem = deps->pending_read_mems;
  	while (pending)