[PATCH] PR33300 Skip 27_io/ios_base/storage/2.cc at -m64 on darwin
Jack Howarth
howarth@bromo.med.uc.edu
Sat Feb 7 18:50:00 GMT 2009
Currently darwin doesn't properly honor RLIMITS when making
impossible memory requests in malloc() at -m64. This can
severely destablize a darwin system after running the
libstdc++-v3/testsuite/27_io/ios_base/storage/2.cc. This
test should be skipped on Darwin. Tested on i686-apple-darwin10.
Okay for gcc trunk?
Jack
ps This issue was discussed in the thread...
http://lists.apple.com/archives/Darwin-dev/2008/Dec/msg00073.html
2009-02-07 Jack Howarth <howarth@bromo.med.uc.edu>
PR testsuite/33300
* libstdc++-v3/testsuite/27_io/ios_base/storage/2.cc: Skip
on darwin at -m64 because RLIMITS aren't honored.
Index: libstdc++-v3/testsuite/27_io/ios_base/storage/2.cc
===================================================================
--- libstdc++-v3/testsuite/27_io/ios_base/storage/2.cc (revision 143993)
+++ libstdc++-v3/testsuite/27_io/ios_base/storage/2.cc (working copy)
@@ -31,6 +31,11 @@
// NULL even if an allocation fails (filed as Radar 3884894).
// { dg-do run { xfail *-*-darwin8.[0-4].* } }
+// Skip test at -m64 on Darwin because RLIMITS are not being honored.
+// Radar 6467883: 10.4/10.5 setrlimits are not honored by memory allocators
+// Radar 6467884: 10.X systems are not robust when paging space is exceeded
+// { dg-skip-if "" { *-*-darwin* } { "-m64" } { "" } }
+
#include <sstream>
#include <iostream>
#include <limits>
More information about the Gcc-patches
mailing list