There was a bug in GCC 4.9.4 that could cause miscompilation, such that the "compiler was freeing the function's stack frame prior to the end of the function itself." [0]
The bug was observed to cause corruption of ext4 filesystems on the aarch64 platform. The upstream report is here: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63293 Apparently this was fixed in GCC 5, but never backported to the 4.9 series. Should we fix it? We use GCC 4.9 to bootstrap the system in (gnu packages commencement), and a handful of packages depend on it. [0] https://lwn.net/SubscriberLink/842122/40442a015222c028/
