maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/clang/test/CodeGen/alignment.c

62 lines
1.6 KiB
C
Raw Normal View History

Correctly allign arrays on 32 bit systems. Before this patch we would align long long int big[1024]; to 4 bytes on 32 bit systems. The problem is that we were only looking at the element type when getLargeArrayMinWidth returned non zero. llvm-svn: 187897
2013-08-08 02:08:19 +08:00
// RUN: %clang_cc1 -triple i386-unknown-linux-gnu -emit-llvm %s -o - | FileCheck %s
Add CodeGen support for alignment on globals, both for unusual natural alignment and alignment attributes. llvm-svn: 51676
2008-05-29 19:10:27 +08:00
__attribute((aligned(16))) float a[128];
filecheckize. llvm-svn: 146907
2011-12-20 05:11:59 +08:00
union {int a[4]; __attribute((aligned(16))) float b[4];} b;
// CHECK: @a = {{.*}}zeroinitializer, align 16
// CHECK: @b = {{.*}}zeroinitializer, align 16
Fix PR5279 - don't sliently drop alignment information on stores of types that have alignment less than the natural alignment of the type when it comes from a typedef. llvm-svn: 146908
2011-12-20 05:16:08 +08:00
Correctly allign arrays on 32 bit systems. Before this patch we would align long long int big[1024]; to 4 bytes on 32 bit systems. The problem is that we were only looking at the element type when getLargeArrayMinWidth returned non zero. llvm-svn: 187897
2013-08-08 02:08:19 +08:00
long long int test5[1024];
// CHECK-DAG: @test5 = common global [1024 x i64] zeroinitializer, align 8
Fix PR5279 - don't sliently drop alignment information on stores of types that have alignment less than the natural alignment of the type when it comes from a typedef. llvm-svn: 146908
2011-12-20 05:16:08 +08:00
// PR5279 - Reduced alignment on typedef.
typedef int myint __attribute__((aligned(1)));
void test1(myint *p) {
*p = 0;
}
// CHECK: @test1(
// CHECK: store i32 0, i32* {{.*}}, align 1
// CHECK: ret void
loads also handled of course. llvm-svn: 146910
2011-12-20 05:17:33 +08:00
int test1a(myint *p) {
return *p;
}
// CHECK: @test1a(
// CHECK: load i32* {{.*}}, align 1
// CHECK: ret i32
Fix PR5279 - don't sliently drop alignment information on stores of types that have alignment less than the natural alignment of the type when it comes from a typedef. llvm-svn: 146908
2011-12-20 05:16:08 +08:00
// PR5279 - Reduced alignment on typedef.
typedef float __attribute__((vector_size(16), aligned(4))) packedfloat4;
void test2(packedfloat4 *p) {
*p = (packedfloat4) { 3.2f, 2.3f, 0.1f, 0.0f };
}
// CHECK: @test2(
// CHECK: store <4 x float> {{.*}}, align 4
// CHECK: ret void
// PR5279 - Reduced alignment on typedef.
typedef float __attribute__((ext_vector_type(3), aligned(4))) packedfloat3;
void test3(packedfloat3 *p) {
*p = (packedfloat3) { 3.2f, 2.3f, 0.1f };
}
// CHECK: @test3(
Remove names from the CHECK lines. llvm-svn: 162003
2012-08-16 08:22:16 +08:00
// CHECK: %{{.*}} = bitcast <3 x float>* %{{.*}} to <4 x float>*
Convert loads and stores of vec3 to vec4 to achieve better code generation. Add test case. llvm-svn: 162002
2012-08-16 08:10:13 +08:00
// CHECK: store <4 x float> {{.*}}, align 4
Fix PR5279 - don't sliently drop alignment information on stores of types that have alignment less than the natural alignment of the type when it comes from a typedef. llvm-svn: 146908
2011-12-20 05:16:08 +08:00
// CHECK: ret void
loads also handled of course. llvm-svn: 146910
2011-12-20 05:17:33 +08:00
implement rdar://10639962 by keeping track of increased alignment information even in subscripting operations. llvm-svn: 147557
2012-01-05 06:35:55 +08:00
typedef float __attribute__((vector_size(16), aligned(64))) float4align64;
// rdar://10639962 - Typedef alignment lost in p[]-style dereferencing
void test4(float4align64 *p) {
p[0] = (float4align64){ 3.2f, 2.3f, 0.1f, 0.0f };
}
// CHECK: @test4(
fix test on -Asserts builds. llvm-svn: 147558
2012-01-05 06:51:21 +08:00
// CHECK: store <4 x float> {{.*}}, <4 x float>* {{.*}}, align 64
implement rdar://10639962 by keeping track of increased alignment information even in subscripting operations. llvm-svn: 147557
2012-01-05 06:35:55 +08:00