File: arm-vector-align.c

package info (click to toggle)
llvm-toolchain-19 1%3A19.1.7-3
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid, trixie
  • size: 1,998,520 kB
  • sloc: cpp: 6,951,680; ansic: 1,486,157; asm: 913,598; python: 232,024; f90: 80,126; objc: 75,281; lisp: 37,276; pascal: 16,990; sh: 10,009; ml: 5,058; perl: 4,724; awk: 3,523; makefile: 3,167; javascript: 2,504; xml: 892; fortran: 664; cs: 573
file content (29 lines) | stat: -rw-r--r-- 1,003 bytes parent folder | download | duplicates (8) 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
 
// REQUIRES: arm-registered-target
// RUN: %clang_cc1 -triple thumbv7-apple-darwin \
// RUN:   -target-abi apcs-gnu \
// RUN:   -target-cpu cortex-a8 \
// RUN:   -mfloat-abi soft \
// RUN:   -target-feature +soft-float-abi \
// RUN:   -ffreestanding \
// RUN:   -emit-llvm -w -o - %s | FileCheck %s

#include <arm_neon.h>

// Check that alignment specifier is used in Neon load/store intrinsics.
typedef float AlignedAddr __attribute__ ((aligned (16)));
void t1(AlignedAddr *addr1, AlignedAddr *addr2) {
// CHECK: @t1
// CHECK: call <4 x float> @llvm.arm.neon.vld1.v4f32.p0(ptr %{{.*}}, i32 16)
  float32x4_t a = vld1q_f32(addr1);
// CHECK: call void @llvm.arm.neon.vst1.p0.v4f32(ptr %{{.*}}, <4 x float> %{{.*}}, i32 16)
  vst1q_f32(addr2, a);
}

// Make sure unaligned load/stores do not gain alignment.
void t2(char *addr) {
// CHECK: @t2
// CHECK: load i32, ptr %{{.*}}, align 1
  int32x2_t vec = vld1_dup_s32(addr);
// CHECK: store i32 %{{.*}}, ptr {{.*}}, align 1
  vst1_lane_s32(addr, vec, 1);
}