Re: [PATCH] Fix cond-expr handling in genmatch

[Richard Biener]

On Mon, 17 Oct 2016, Richard Biener wrote:

> 
> This fixes matching of toplevel (cond (lt @1 @2) ...) as reported by
> Bin to me privately.
> 
> Bootstrapped on x86_64-unknown-linux-gnu, testing in progress.

This is what I applied.

Richard.

2016-10-18  Richard Biener  <rguent...@suse.de>

        * genmatch.c (dt_operand::gen_gimple_expr): Use get_name to
        get at the operand to look at with TREE_OPERAND for generic
        sub-nodes.

Index: gcc/genmatch.c
===================================================================
--- gcc/genmatch.c      (revision 241228)
+++ gcc/genmatch.c      (working copy)
@@ -2644,9 +2644,19 @@ dt_operand::gen_gimple_expr (FILE *f, in
              /* ???  If this is a memory operation we can't (and should not)
                 match this.  The only sensible operand types are
                 SSA names and invariants.  */
-             fprintf_indent (f, indent,
-                             "tree %s = TREE_OPERAND (gimple_assign_rhs1 
(def), %i);\n",
-                             child_opname, i);
+             if (e->is_generic)
+               {
+                 char opname[20];
+                 get_name (opname);
+                 fprintf_indent (f, indent,
+                                 "tree %s = TREE_OPERAND (%s, %i);\n",
+                                 child_opname, opname, i);
+               }
+             else
+               fprintf_indent (f, indent,
+                               "tree %s = TREE_OPERAND "
+                               "(gimple_assign_rhs1 (def), %i);\n",
+                               child_opname, i);
              fprintf_indent (f, indent,
                              "if ((TREE_CODE (%s) == SSA_NAME\n",
                              child_opname);