Linux/mipsel is now broken.

[H . J . Lu]

On Sat, Jan 12, 2002 at 05:28:35PM -0800, Richard Henderson wrote:
> On Sat, Jan 12, 2002 at 04:56:38PM -0800, H . J . Lu wrote:
> > ... is your change safe for o32 when long long is involved?
> 
> As far as I know, yes.  How can I possibly guess at what
> is going on?  If you build just the stage1 cc1, do you 
> see any check-gcc regressions?

I will give it a try. It will take a while.

If I applied the patch enclosed here, Linux/mipsel bootstrap fine.


H.J.
----
--- gcc/regrename.c.mode	Fri Jan 11 14:13:10 2002
+++ gcc/regrename.c	Sat Jan 12 16:58:30 2002
@@ -1600,8 +1600,10 @@ copyprop_hardreg_forward_1 (bb, vd)
 	  /* Otherwise, try all valid registers and see if its valid.  */
 	  for (i = vd->e[regno].oldest_regno; i != regno;
 	       i = vd->e[i].next_regno)
-	    if (vd->e[i].mode == mode
+	    if (vd->e[i].mode == mode)
+#if 0
 		|| mode_change_ok (vd->e[i].mode, mode, i))
+#endif
 	      {
 		new = gen_rtx_raw_REG (mode, i);
 		if (validate_change (insn, &SET_SRC (set), new, 0))