On Sun, Oct 06, 2013 at 05:38:15PM +0200, Jan Kratochvil wrote: > On Sun, 06 Oct 2013 17:25:20 +0200, Mark Wielaard wrote: > > But on ia64 the abi_cfi data and actual frame data were placed > > too far apart and caused a crash in tests/run-addrcfi.sh for ppc32. > > OK, it seems right to me, thanks for the simplification. > I sure was not testing it on an ia64 host.
Embarrassingly I did test on ia64 before the 0.157 release and did see the addrcfi failure. But I had assumed that was because we didn't have an ia64_abi_cfi ebl hook and ignored it. Only now when I wanted to implement it, did I notice it crashed on the ppc32 testcase (and it took me some time to figure out why). And that ia64 doesn't support CFI in the way we support it. It doesn't have an .eh_frame or .debug_frame but its own .IA_64.unwind section format we don't currently support. But the fix seems good anyway since theoretically it could also crash on other arches. Cheers, Mark
