User account creation filtered due to spam.
Since r241991, gcc.dg/vect/vect-strided-a-u8-i2-gap.c fails on armeb: gcc.dg/vect/vect-strided-a-u8-i2-gap.c -flto -ffat-lto-objects scan-tree-dump-times vect "vectorized 2 loops" 1 gcc.dg/vect/vect-strided-a-u8-i2-gap.c scan-tree-dump-times vect "vectorized 2 loops" 1 This was discussed in the thread starting at: https://gcc.gnu.org/ml/gcc-patches/2016-11/msg00550.html
Hi Guys, I have proposed an extended version of Richard's patch to fix this regression: https://gcc.gnu.org/ml/gcc-patches/2017-01/msg01615.html Cheers Nick
Author: nickc Date: Mon Jan 23 12:24:35 2017 New Revision: 244796 URL: https://gcc.gnu.org/viewcvs?rev=244796&root=gcc&view=rev Log: PR testsuite/78421 * lib/target-supports.exp (check_effective_target_vect_hw_misalign): If the target is ARM return the result of the check_effective_target_arm_vect_no_misalign proc. * gcc.dg/vect/vect-strided-a-u8-i2-gap.c: If the target does not support unaligned vectors then only expect one of the loops to be unrolled. Modified: trunk/gcc/testsuite/ChangeLog trunk/gcc/testsuite/gcc.dg/vect/vect-strided-a-u8-i2-gap.c trunk/gcc/testsuite/lib/target-supports.exp
Patch applied.
I still see the test failing on arm-none-eabi when testing with, for example: -mthumb/-march=armv8-a/-mfpu=crypto-neon-fp-armv8/-mfloat-abi=hard Christophe also has a report: https://gcc.gnu.org/ml/gcc-patches/2017-01/msg01806.html
(In reply to ktkachov from comment #4) > I still see the test failing on arm-none-eabi when testing with, for example: > -mthumb/-march=armv8-a/-mfpu=crypto-neon-fp-armv8/-mfloat-abi=hard > > Christophe also has a report: > https://gcc.gnu.org/ml/gcc-patches/2017-01/msg01806.html Please have a look at the vectorizer logs to see why it fails. armeb is not primary or secondary target, demoting.
GCC 7.1 has been released.