This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH/AARCH64] Fix 64893: ICE with vget_lane_u32 with C++ front-end at -O0
- From: James Greenhalgh <james dot greenhalgh at arm dot com>
- To: Christophe Lyon <christophe dot lyon at linaro dot org>
- Cc: Andrew Pinski <pinskia at gmail dot com>, Jakub Jelinek <jakub at redhat dot com>, GCC Patches <gcc-patches at gcc dot gnu dot org>, Alan Lawrence <Alan dot Lawrence at arm dot com>
- Date: Fri, 6 Mar 2015 09:45:34 +0000
- Subject: Re: [PATCH/AARCH64] Fix 64893: ICE with vget_lane_u32 with C++ front-end at -O0
- Authentication-results: sourceware.org; auth=none
- References: <CA+=Sn1nmq5gaB7G5B5x7QTEt4v0wVCKki15nC6oXJodzuDR-TA at mail dot gmail dot com> <20150203073744 dot GV1746 at tucnak dot redhat dot com> <CA+=Sn1=Z7wYY06pyrTCo31J3JTR=HAv7Vz-1bwmykm18isuLpA at mail dot gmail dot com> <CA+=Sn1k0P3t1r65R_8E1Z3_pr3oosgZWpkEEFC=nmF9fni6PrA at mail dot gmail dot com> <CAKdteOZx+3y8H6R05pnjmkGpCmboGzHZ02Aqmd1vEfpV2qUz2g at mail dot gmail dot com>
On Thu, Feb 12, 2015 at 03:37:33PM +0000, Christophe Lyon wrote:
> On 8 February 2015 at 03:24, Andrew Pinski <pinskia@gmail.com> wrote:
> > On Fri, Feb 6, 2015 at 5:02 PM, Andrew Pinski <pinskia@gmail.com> wrote:
> > PR target/64893
> > * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
> > Change the first argument type to size_type_node and add another
> > size_type_node.
> > (aarch64_simd_expand_builtin): Handle the new argument to
> > AARCH64_SIMD_BUILTIN_LANE_CHECK and don't ICE but rather
> > print an out when the first two arguments are not
> > nonzero integer constants.
> > * config/aarch64/arm_neon.h (__AARCH64_LANE_CHECK):
> > Pass the sizeof directly to __builtin_aarch64_im_lane_boundsi.
> >
> > testsuite/ChangeLog:
> >
> > * c-c++-common/torture/aarch64-vect-lane-1.c: New testcase.
> > * c-c++-common/torture/aarch64-vect-lane-2.c: New testcase.
> In case you haven't noticed, aarch64-vect-lane-2.c FAILs when compiled at
> -O2 -flto -fuse-linker-plugin -fno-fat-lto-objects
> but PASSes with the other optimization levels.
Hi Andrew,
Did you get a chance to look at this? Given the way that your error
checking works, the test probably just needs a dg-skip-if -flto
directive. Did you intend to come back to this and make it fail earlier
than link time with -flto?
If I don't hear otherwise, I'll propose the patch adding dg-skip-if on
Monday.
Cheers,
James