Fix pr/60004

Richard Henderson rth@redhat.com
Fri Jan 31 20:50:00 GMT 2014


It seems that lower_try_finally_switch was not properly updated for the
addition of EH_ELSE.  We tried to lower the finally construct before we
actually handled the EH_ELSE construct specially.

I examined the other lower_try_finally_* alternatives and they all seem to do
the right thing.

Testing in progress; I'll commit if things go well.


r~
-------------- next part --------------
diff --git a/gcc/testsuite/g++.dg/tm/pr60004.C b/gcc/testsuite/g++.dg/tm/pr60004.C
new file mode 100644
index 0000000..b8c2c0e
--- /dev/null
+++ b/gcc/testsuite/g++.dg/tm/pr60004.C
@@ -0,0 +1,10 @@
+// { dg-do compile }
+// { dg-options "-fgnu-tm" }
+
+int a;
+int f() {
+    __transaction_atomic {
+        if (a == 5)
+            return 1;
+    }
+}
diff --git a/gcc/tree-eh.c b/gcc/tree-eh.c
index 0c8282e..e9c714c 100644
--- a/gcc/tree-eh.c
+++ b/gcc/tree-eh.c
@@ -1388,9 +1388,6 @@ lower_try_finally_switch (struct leh_state *state, struct leh_tf_state *tf)
   x = gimple_seq_last_stmt (finally);
   finally_loc = x ? gimple_location (x) : tf_loc;
 
-  /* Lower the finally block itself.  */
-  lower_eh_constructs_1 (state, &finally);
-
   /* Prepare for switch statement generation.  */
   nlabels = tf->dest_array.length ();
   return_index = nlabels;
@@ -1476,6 +1473,7 @@ lower_try_finally_switch (struct leh_state *state, struct leh_tf_state *tf)
   x = gimple_build_label (finally_label);
   gimple_seq_add_stmt (&tf->top_p_seq, x);
 
+  lower_eh_constructs_1 (state, &finally);
   gimple_seq_add_seq (&tf->top_p_seq, finally);
 
   /* Redirect each incoming goto edge.  */


More information about the Gcc-patches mailing list