[Bug tree-optimization/51519] [4.7 Regression] ICE: in inline_small_functions, at ipa-inline.c:1410 with -O -fno-guess-branch-probability -findirect-inlining

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51519

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2011-12-13
                 CC|                            |hubicka at gcc dot gnu.org
         AssignedTo|unassigned at gcc dot       |rguenth at gcc dot gnu.org
                   |gnu.org                     |
   Target Milestone|---                         |4.7.0
     Ever Confirmed|0                           |1

--- Comment #1 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-12-13 
11:15:58 UTC ---
Confirmed.  The -fno-guess-branch-probability case still uses the overall
growth in edge_badness.  The following will fix it.

Index: gcc/ipa-inline.c
===================================================================
--- gcc/ipa-inline.c    (revision 182220)
+++ gcc/ipa-inline.c    (working copy)
@@ -861,7 +861,7 @@ edge_badness (struct cgraph_edge *edge,
   else
     {
       int nest = MIN (inline_edge_summary (edge)->loop_depth, 8);
-      badness = estimate_growth (callee) * 256;
+      badness = growth * 256;

       /* Decrease badness if call is nested.  */
       if (badness > 0)