Fix for PR 48988

[Xinliang David Li <davidxl@google.com>]

[-- Attachment #1: Type: text/plain, Size: 370 bytes --]

Hi, the following trial patch fixed PR 48988 which is a regression.

Bootstrap and tested on x86/linux. Verified the reported failure is fixed.

Ok for trunk?

David


2011-05-22  David Li  <davidxl@google.com>

        PR tree-optimization/48988
        * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
        Initialize has_valid_pred for each pred chain.

[-- Attachment #2: u.p --]
[-- Type: application/octet-stream, Size: 644 bytes --]

2011-05-22  David Li  <davidxl@google.com>

	PR tree-optimization/48988
	* tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
	Initialize has_valid_pred for each pred chain.

Index: tree-ssa-uninit.c
===================================================================
--- tree-ssa-uninit.c	(revision 173837)
+++ tree-ssa-uninit.c	(working copy)
@@ -358,6 +358,8 @@ convert_control_dep_chain_into_preds (VE
   for (i = 0; i < num_chains; i++)
     {
       VEC(edge, heap) *one_cd_chain = dep_chains[i];
+
+      has_valid_pred = false;
       for (j = 0; j < VEC_length (edge, one_cd_chain); j++)
         {
           gimple cond_stmt;