diff options
| author | Andrew Pinski <apinski@marvell.com> | 2021-09-02 07:08:22 +0000 |
|---|---|---|
| committer | Andrew Pinski <apinski@marvell.com> | 2021-09-13 15:19:05 +0000 |
| commit | 03312cbd54f337dfb25be356a1d1abc9925c6c03 (patch) | |
| tree | 88711fbf822c544cb5629a2d343e6a672bd89e28 /gcc/testsuite | |
| parent | 20f3c168205cc7b0a97ecd54ffc54bed7637be74 (diff) | |
| download | gcc-03312cbd54f337dfb25be356a1d1abc9925c6c03.zip gcc-03312cbd54f337dfb25be356a1d1abc9925c6c03.tar.gz gcc-03312cbd54f337dfb25be356a1d1abc9925c6c03.tar.bz2 | |
[aarch64] Fix target/95969: __builtin_aarch64_im_lane_boundsi interferes with gimple
This patch adds simple folding of __builtin_aarch64_im_lane_boundsi where
we are not going to error out. It fixes the problem by the removal
of the function from the IR.
OK? Bootstrapped and tested on aarch64-linux-gnu with no regressions.
gcc/ChangeLog:
PR target/95969
* config/aarch64/aarch64-builtins.c (aarch64_fold_builtin_lane_check):
New function.
(aarch64_general_fold_builtin): Handle AARCH64_SIMD_BUILTIN_LANE_CHECK.
(aarch64_general_gimple_fold_builtin): Likewise.
gcc/testsuite/ChangeLog:
PR target/95969
* gcc.target/aarch64/lane-bound-1.c: New test.
* gcc.target/aarch64/lane-bound-2.c: New test.
Diffstat (limited to 'gcc/testsuite')
| -rw-r--r-- | gcc/testsuite/gcc.target/aarch64/lane-bound-1.c | 14 | ||||
| -rw-r--r-- | gcc/testsuite/gcc.target/aarch64/lane-bound-2.c | 10 |
2 files changed, 24 insertions, 0 deletions
diff --git a/gcc/testsuite/gcc.target/aarch64/lane-bound-1.c b/gcc/testsuite/gcc.target/aarch64/lane-bound-1.c new file mode 100644 index 0000000..bbbe679 --- /dev/null +++ b/gcc/testsuite/gcc.target/aarch64/lane-bound-1.c @@ -0,0 +1,14 @@ +/* { dg-do compile } */ +/* { dg-options "-O2 -fdump-tree-optimized" } */ +#include <arm_neon.h> + +void +f (float32x4_t **ptr) +{ + float32x4_t res = vsetq_lane_f32 (0.0f, **ptr, 0); + **ptr = res; +} +/* GCC should be able to remove the call to "__builtin_aarch64_im_lane_boundsi" + and optimize out the second load from *ptr. */ +/* { dg-final { scan-tree-dump-times "__builtin_aarch64_im_lane_boundsi" 0 "optimized" } } */ +/* { dg-final { scan-tree-dump-times " = \\\*ptr_" 1 "optimized" } } */ diff --git a/gcc/testsuite/gcc.target/aarch64/lane-bound-2.c b/gcc/testsuite/gcc.target/aarch64/lane-bound-2.c new file mode 100644 index 0000000..923c946 --- /dev/null +++ b/gcc/testsuite/gcc.target/aarch64/lane-bound-2.c @@ -0,0 +1,10 @@ +/* { dg-do compile } */ +/* { dg-options "-O2 -fdump-tree-original" } */ +void +f (void) +{ + __builtin_aarch64_im_lane_boundsi (16, 4, 0); + __builtin_aarch64_im_lane_boundsi (8, 8, 0); +} +/* GCC should be able to optimize these out before gimplification. */ +/* { dg-final { scan-tree-dump-times "__builtin_aarch64_im_lane_boundsi" 0 "original" } } */ |