On Wed, Apr 04, 2018 at 19:24:52 +0100, Alex Bennée wrote: > > Emilio G. Cota <c...@braap.org> writes: > > > On Wed, Feb 07, 2018 at 14:55:36 -0800, Richard Henderson wrote: > >> Reviewed-by: Alex Bennée <alex.ben...@linaro.org> > >> Reviewed-by: Peter Maydell <peter.mayd...@linaro.org> > >> Signed-off-by: Richard Henderson <richard.hender...@linaro.org> > > > > Just bisected a segfault for aarch64 nbench on a PowerPC host > > to this commit (79d61de6bdc398). I've also tested on x86_64 and > > aarch64 hosts, and these do not seem to be affected. > > I wonder about other hosts without TCG_vec backend support (powerpc is > one)? I suspect it must be the fall-back for the vector expanders is > wrong.
I just tried on two additional machines: - Reproduced on a different PowerPC host with different endianness (gcc110, which is power7 and therefore big endian) - Could not reproduce on a mips64 host (gcc23) So it might be a Power-only thing. Thanks, Emilio