On Mon, Jul 07, 2014 at 05:23:28PM +0100, James Greenhalgh wrote: > On Fri, Jul 04, 2014 at 03:45:10PM +0100, Marcus Shawcroft wrote: > > On 24 June 2014 09:45, James Greenhalgh <james.greenha...@arm.com> wrote: > > > > > 2014-06-20 James Greenhalgh <james.greenha...@arm.com> > > > > > > * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>): > > > New. > > > (move_lo_quad_internal_be_<mode>): Likewise. > > > (move_lo_quad_<mode>): Convert to define_expand. > > > (aarch64_simd_move_hi_quad_<mode>): Gate on BYTES_BIG_ENDIAN. > > > (aarch64_simd_move_hi_quad_be_<mode>): New. > > > (move_hi_quad_<mode>): Use appropriate insn for BYTES_BIG_ENDIAN. > > > (aarch64_combinez<mode>): Gate on BYTES_BIG_ENDIAN. > > > (aarch64_combinez_be<mode>): New. > > > (aarch64_combine<mode>): Convert to define_expand. > > > (aarch64_combine_internal<mode>): New. > > > (aarch64_simd_combine<mode>): Remove bogus RTL description. > > > > OK... and back port to 4.9 please? > > > > The backport applies cleanly and there are no regressions for aarch64-none-elf > or aarch64_be-none-elf in the testsuite. > > Jakub, I know you were planning a 4.9.1 release soon, is this patch OK to > port to gcc-4_9-branch now, or would you prefer I wait? >
Jakub on IRC said: <jakub> generally, unless the release branch is frozen, target maintainers can ack backports to the release branches, there is no need for an extra ack from RMs So, I've committed the clean backport as r212359 as this patch is a correctness issue acked by a maintainer which does not cause any ABI changes. Thanks, James
