[pypy-svn] pypy jitypes2: give up trying to properly freeing the memory allocated for the struct:

antocuni commits-noreply at bitbucket.org
Mon Dec 27 19:04:31 CET 2010


Author: Antonio Cuni <anto.cuni at gmail.com>
Branch: jitypes2
Changeset: r40243:57faf52c9d5a
Date: 2010-12-27 19:00 +0100
http://bitbucket.org/pypy/pypy/changeset/57faf52c9d5a/

Log:	give up trying to properly freeing the memory allocated for the
	struct: instead, we simply don't track allocations for this test.
	The leak happens because we call the function 10 times in the loop,
	but only free the last result. It's not too bad, as the leak in
	only in the test code anyway

diff --git a/pypy/jit/metainterp/test/test_fficall.py b/pypy/jit/metainterp/test/test_fficall.py
--- a/pypy/jit/metainterp/test/test_fficall.py
+++ b/pypy/jit/metainterp/test/test_fficall.py
@@ -9,6 +9,7 @@
 from pypy.rpython.lltypesystem import lltype, rffi
 from pypy.jit.metainterp.test.test_basic import LLJitMixin
 from pypy.rlib.objectmodel import specialize
+from pypy.tool.sourcetools import func_with_new_name
 
 class TestFfiCall(LLJitMixin, _TestLibffiCall):
 
@@ -60,3 +61,8 @@
         #
         res = self.meta_interp(f, [0], jit_ffi=True, backendopt=True)
         return res
+
+    def test_byval_result(self):
+        _TestLibffiCall.test_byval_result(self)
+    test_byval_result.__doc__ = _TestLibffiCall.test_byval_result.__doc__
+    test_byval_result.dont_track_allocations = True


More information about the Pypy-commit mailing list