Reviewers: Jakob,
Description:
Ensure expected behavior for transition hosting tests by flushing ICs
[email protected]
Please review this at https://chromiumcodereview.appspot.com/9373027/
SVN Base: https://v8.googlecode.com/svn/branches/bleeding_edge
Affected files:
M test/mjsunit/elements-transition-hoisting.js
Index: test/mjsunit/elements-transition-hoisting.js
diff --git a/test/mjsunit/elements-transition-hoisting.js
b/test/mjsunit/elements-transition-hoisting.js
index
53dc9409199eba2e9af02ed9abaa8975e48801f3..76f7b82b5ae4d15711364ccbf0e3db1c86ae6e36
100644
--- a/test/mjsunit/elements-transition-hoisting.js
+++ b/test/mjsunit/elements-transition-hoisting.js
@@ -25,7 +25,7 @@
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-// Flags: --allow-natives-syntax --smi-only-arrays
+// Flags: --allow-natives-syntax --smi-only-arrays --expose-gc
// Ensure that ElementsKind transitions in various situations are hoisted
(or
// not hoisted) correctly, don't change the semantics programs and don't
trigger
@@ -39,6 +39,11 @@ if (support_smi_only_arrays) {
print("Tests do NOT include smi-only arrays.");
}
+// Force existing ICs from previous stress runs to be flushed, otherwise
the
+// assumptions in this test about when deoptimizations get triggered are
not
+// valid.
+gc();
+
if (support_smi_only_arrays) {
// Make sure that a simple elements array transitions inside a loop
before
// stores to an array gets hoisted in a way that doesn't generate a
deopt in
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
