Hi,
profiledbootstrap shows bug in edge_badness. With inline hints the growth is
no longer
limited by small constant and thus multiplying it with overall growth may
overlfow.
Bootstrapped/regtested & comitted.
Honza
PR bootstrap/55051
* ipa-inline.c (edge_badness): Improve dumping; fix overflow.
Index: ipa-inline.c
===================================================================
--- ipa-inline.c (revision 193505)
+++ ipa-inline.c (working copy)
@@ -850,9 +850,11 @@ edge_badness (struct cgraph_edge *edge,
if (dump)
{
- fprintf (dump_file, " Badness calculation for %s -> %s\n",
+ fprintf (dump_file, " Badness calculation for %s/%i -> %s/%i\n",
xstrdup (cgraph_node_name (edge->caller)),
- xstrdup (cgraph_node_name (callee)));
+ edge->caller->uid,
+ xstrdup (cgraph_node_name (callee)),
+ edge->callee->uid);
fprintf (dump_file, " size growth %i, time %i ",
growth,
edge_time);
@@ -917,7 +919,7 @@ edge_badness (struct cgraph_edge *edge,
{
badness = (relative_time_benefit (callee_info, edge, edge_time)
* (INT_MIN / 16 / RELATIVE_TIME_BENEFIT_RANGE));
- badness /= (growth * MAX (1, callee_info->growth));
+ badness /= (MIN (65536/2, growth) * MIN (65536/2, MAX (1,
callee_info->growth)));
gcc_checking_assert (badness <=0 && badness >= INT_MIN / 16);
if ((hints & (INLINE_HINT_indirect_call
| INLINE_HINT_loop_iterations
