Hi Guys,

  I am applying the patch below to fix a problem with the RL78's popping
  of pushed registers in G10 mode.  The problem was that the pop uses a
  two instruction sequence and dead code elimination was deleting the
  second instruction (the move of the popped value from register A to
  the correct register) because the popped values were never used.

Cheers
  Nick

gcc/ChangeLog
2014-09-19  Nick Clifton  <ni...@redhat.com>

        * config/rl78/rl78.c (rl78_expand_epilogue): Generate a USE of the
        pop'ed registers so that DCE does not eliminate them.

Index: gcc/config/rl78/rl78.c
===================================================================
RCS file: /cvs/cvsfiles/gnupro/gcc/config/rl78/rl78.c,v
retrieving revision 1.7.4.2
diff -u -3 -p -r1.7.4.2 rl78.c
@@ -1219,10 +1219,19 @@ rl78_expand_epilogue (void)
   for (i = 15; i >= 0; i--)
     if (cfun->machine->need_to_push [i])
       {
+       rtx dest = gen_rtx_REG (HImode, i * 2);
+
        if (TARGET_G10)
          {
-           emit_insn (gen_pop (gen_rtx_REG (HImode, 0)));
-           emit_move_insn (gen_rtx_REG (HImode, i*2), gen_rtx_REG (HImode, 0));
+           rtx ax = gen_rtx_REG (HImode, 0);
+
+           emit_insn (gen_pop (ax));
+           if (i != 0)
+             {
+               emit_move_insn (dest, ax);
+               /* Generate a USE of the pop'd register so that DCE will not 
eliminate the move.  */
+               emit_insn (gen_use (dest));
+             }
          }
        else
          {
@@ -1233,7 +1242,7 @@ rl78_expand_epilogue (void)
                emit_insn (gen_sel_rb (GEN_INT (need_bank)));
                rb = need_bank;
              }
-           emit_insn (gen_pop (gen_rtx_REG (HImode, i * 2)));
+           emit_insn (gen_pop (dest));
          }
       }
 

Reply via email to