On Fri, Mar 27, 2015 at 05:15:43PM +0300, Maxim Uvarov wrote: > On 03/27/15 17:11, Stuart Haslam wrote: > >It doesn't break non-jumbo tests, in fact all tests pass and I > >didn't *notice* any adverse effects, I found the issue by > >inspection. > > > >The problem is that when generating non-jumbo packets (which most > >of the tests do) the magic2 marker is written beyond the end of a > >calloc'd block into un-allocated memory. The receiver doesn't > >check magic2 for non-jumbo packets. > > > >The issue could be trivially avoided with the simpler change > >below, but I wasn't happy with that as we'd still have a pointer > >to a structure that's much larger than the memory allocated. > > > >Stuart. > > Yes this change is ok. Did that the same as > https://bugs.linaro.org/show_bug.cgi?id=1365 > ? > > Maxim.
Bug 1365 (MMSG lock up) is a different issue, I'm working on that now. -- Stuart. _______________________________________________ lng-odp mailing list [email protected] http://lists.linaro.org/mailman/listinfo/lng-odp
