Hi, Jakub,
According to this discussion thread:
http://gcc.gnu.org/ml/gcc/2013-08/msg00142.html
The bug that David mentioned has been fixed on trunk but not on 4.8 branch.
IMHO it would be a good idea to backport it and Vladmir agreed with that.
The patch and a plaintext ChangeLog are as follow.
Bootstraped and regtested on x86_64-unknown-linux-gnu.
Is it OK to backport it into gcc-4_8-branch?
Index: gcc/ChangeLog
===================================================================
--- gcc/ChangeLog (revision 201718)
+++ gcc/ChangeLog (working copy)
@@ -1,3 +1,11 @@
+2013-08-14 David Given <d...@cowlark.com>
+
+ Backport from mainline
+ 2013-04-26 Vladimir Makarov <vmaka...@redhat.com>
+
+ * lra-constraints.c (process_alt_operands): Use #if HAVE_ATTR_enable
+ instead of #ifdef.
+
2013-08-13 Jakub Jelinek <ja...@redhat.com>
PR sanitizer/56417
Index: gcc/lra-constraints.c
===================================================================
--- gcc/lra-constraints.c (revision 201718)
+++ gcc/lra-constraints.c (working copy)
@@ -1388,7 +1388,7 @@
for (nalt = 0; nalt < n_alternatives; nalt++)
{
/* Loop over operands for one constraint alternative. */
-#ifdef HAVE_ATTR_enabled
+#if HAVE_ATTR_enabled
if (curr_id->alternative_enabled_p != NULL
&& ! curr_id->alternative_enabled_p[nalt])
continue;
Best regards,
jasonwucj
