Home | History | Annotate | Download | only in CodeGen
      1 // RUN: %clang_cc1 -triple x86_64-unknown-unknown -emit-llvm -o - %s| FileCheck %s
      2 // RUN: %clang_cc1 -triple x86_64-unknown-unknown -emit-llvm -o - %s -target-feature +avx | FileCheck %s -check-prefix=AVX
      3 #include <stdarg.h>
      4 
      5 // CHECK: define signext i8 @f0()
      6 char f0(void) {
      7   return 0;
      8 }
      9 
     10 // CHECK: define signext i16 @f1()
     11 short f1(void) {
     12   return 0;
     13 }
     14 
     15 // CHECK: define i32 @f2()
     16 int f2(void) {
     17   return 0;
     18 }
     19 
     20 // CHECK: define float @f3()
     21 float f3(void) {
     22   return 0;
     23 }
     24 
     25 // CHECK: define double @f4()
     26 double f4(void) {
     27   return 0;
     28 }
     29 
     30 // CHECK: define x86_fp80 @f5()
     31 long double f5(void) {
     32   return 0;
     33 }
     34 
     35 // CHECK: define void @f6(i8 signext %a0, i16 signext %a1, i32 %a2, i64 %a3, i8* %a4)
     36 void f6(char a0, short a1, int a2, long long a3, void *a4) {
     37 }
     38 
     39 // CHECK: define void @f7(i32 %a0)
     40 typedef enum { A, B, C } e7;
     41 void f7(e7 a0) {
     42 }
     43 
     44 // Test merging/passing of upper eightbyte with X87 class.
     45 //
     46 // CHECK: define void @f8_1(%union.u8* noalias sret %agg.result)
     47 // CHECK: define void @f8_2(%union.u8* byval align 16 %a0)
     48 union u8 {
     49   long double a;
     50   int b;
     51 };
     52 union u8 f8_1() { while (1) {} }
     53 void f8_2(union u8 a0) {}
     54 
     55 // CHECK: define i64 @f9()
     56 struct s9 { int a; int b; int : 0; } f9(void) { while (1) {} }
     57 
     58 // CHECK: define void @f10(i64 %a0.coerce)
     59 struct s10 { int a; int b; int : 0; };
     60 void f10(struct s10 a0) {}
     61 
     62 // CHECK: define void @f11(%union.anon* noalias sret %agg.result)
     63 union { long double a; float b; } f11() { while (1) {} }
     64 
     65 // CHECK: define i32 @f12_0()
     66 // CHECK: define void @f12_1(i32 %a0.coerce)
     67 struct s12 { int a __attribute__((aligned(16))); };
     68 struct s12 f12_0(void) { while (1) {} }
     69 void f12_1(struct s12 a0) {}
     70 
     71 // Check that sret parameter is accounted for when checking available integer
     72 // registers.
     73 // CHECK: define void @f13(%struct.s13_0* noalias sret %agg.result, i32 %a, i32 %b, i32 %c, i32 %d, {{.*}}* byval align 8 %e, i32 %f)
     74 
     75 struct s13_0 { long long f0[3]; };
     76 struct s13_1 { long long f0[2]; };
     77 struct s13_0 f13(int a, int b, int c, int d,
     78                  struct s13_1 e, int f) { while (1) {} }
     79 
     80 // CHECK: define void @f14({{.*}}, i8 signext %X)
     81 void f14(int a, int b, int c, int d, int e, int f, char X) {}
     82 
     83 // CHECK: define void @f15({{.*}}, i8* %X)
     84 void f15(int a, int b, int c, int d, int e, int f, void *X) {}
     85 
     86 // CHECK: define void @f16({{.*}}, float %X)
     87 void f16(float a, float b, float c, float d, float e, float f, float g, float h,
     88          float X) {}
     89 
     90 // CHECK: define void @f17({{.*}}, x86_fp80 %X)
     91 void f17(float a, float b, float c, float d, float e, float f, float g, float h,
     92          long double X) {}
     93 
     94 // Check for valid coercion.  The struct should be passed/returned as i32, not
     95 // as i64 for better code quality.
     96 // rdar://8135035
     97 // CHECK: define void @f18(i32 %a, i32 %f18_arg1.coerce)
     98 struct f18_s0 { int f0; };
     99 void f18(int a, struct f18_s0 f18_arg1) { while (1) {} }
    100 
    101 // Check byval alignment.
    102 
    103 // CHECK: define void @f19(%struct.s19* byval align 16 %x)
    104 struct s19 {
    105   long double a;
    106 };
    107 void f19(struct s19 x) {}
    108 
    109 // CHECK: define void @f20(%struct.s20* byval align 32 %x)
    110 struct __attribute__((aligned(32))) s20 {
    111   int x;
    112   int y;
    113 };
    114 void f20(struct s20 x) {}
    115 
    116 struct StringRef {
    117   long x;
    118   const char *Ptr;
    119 };
    120 
    121 // rdar://7375902
    122 // CHECK: define i8* @f21(i64 %S.coerce0, i8* %S.coerce1)
    123 const char *f21(struct StringRef S) { return S.x+S.Ptr; }
    124 
    125 // PR7567
    126 typedef __attribute__ ((aligned(16))) struct f22s { unsigned long long x[2]; } L;
    127 void f22(L x, L y) { }
    128 // CHECK: @f22
    129 // CHECK: %x = alloca{{.*}}, align 16
    130 // CHECK: %y = alloca{{.*}}, align 16
    131 
    132 
    133 
    134 // PR7714
    135 struct f23S {
    136   short f0;
    137   unsigned f1;
    138   int f2;
    139 };
    140 
    141 
    142 void f23(int A, struct f23S B) {
    143   // CHECK: define void @f23(i32 %A, i64 %B.coerce0, i32 %B.coerce1)
    144 }
    145 
    146 struct f24s { long a; int b; };
    147 
    148 struct f23S f24(struct f23S *X, struct f24s *P2) {
    149   return *X;
    150 
    151   // CHECK: define { i64, i32 } @f24(%struct.f23S* %X, %struct.f24s* %P2)
    152 }
    153 
    154 // rdar://8248065
    155 typedef float v4f32 __attribute__((__vector_size__(16)));
    156 v4f32 f25(v4f32 X) {
    157   // CHECK: define <4 x float> @f25(<4 x float> %X)
    158   // CHECK-NOT: alloca
    159   // CHECK: alloca <4 x float>
    160   // CHECK-NOT: alloca
    161   // CHECK: store <4 x float> %X, <4 x float>*
    162   // CHECK-NOT: store
    163   // CHECK: ret <4 x float>
    164   return X+X;
    165 }
    166 
    167 struct foo26 {
    168   int *X;
    169   float *Y;
    170 };
    171 
    172 struct foo26 f26(struct foo26 *P) {
    173   // CHECK: define { i32*, float* } @f26(%struct.foo26* %P)
    174   return *P;
    175 }
    176 
    177 
    178 struct v4f32wrapper {
    179   v4f32 v;
    180 };
    181 
    182 struct v4f32wrapper f27(struct v4f32wrapper X) {
    183   // CHECK: define <4 x float> @f27(<4 x float> %X.coerce)
    184   return X;
    185 }
    186 
    187 // rdar://5711709
    188 struct f28c {
    189   double x;
    190   int y;
    191 };
    192 void f28(struct f28c C) {
    193   // CHECK: define void @f28(double %C.coerce0, i32 %C.coerce1)
    194 }
    195 
    196 struct f29a {
    197   struct c {
    198     double x;
    199     int y;
    200   } x[1];
    201 };
    202 
    203 void f29a(struct f29a A) {
    204   // CHECK: define void @f29a(double %A.coerce0, i32 %A.coerce1)
    205 }
    206 
    207 // rdar://8249586
    208 struct S0 { char f0[8]; char f2; char f3; char f4; };
    209 void f30(struct S0 p_4) {
    210   // CHECK: define void @f30(i64 %p_4.coerce0, i24 %p_4.coerce1)
    211 }
    212 
    213 // Pass the third element as a float when followed by tail padding.
    214 // rdar://8251384
    215 struct f31foo { float a, b, c; };
    216 float f31(struct f31foo X) {
    217   // CHECK: define float @f31(<2 x float> %X.coerce0, float %X.coerce1)
    218   return X.c;
    219 }
    220 
    221 _Complex float f32(_Complex float A, _Complex float B) {
    222   // rdar://6379669
    223   // CHECK: define <2 x float> @f32(<2 x float> %A.coerce, <2 x float> %B.coerce)
    224   return A+B;
    225 }
    226 
    227 
    228 // rdar://8357396
    229 struct f33s { long x; float c,d; };
    230 
    231 void f33(va_list X) {
    232   va_arg(X, struct f33s);
    233 }
    234 
    235 typedef unsigned long long v1i64 __attribute__((__vector_size__(8)));
    236 
    237 // rdar://8359248
    238 // CHECK: define i64 @f34(i64 %arg.coerce)
    239 v1i64 f34(v1i64 arg) { return arg; }
    240 
    241 
    242 // rdar://8358475
    243 // CHECK: define i64 @f35(i64 %arg.coerce)
    244 typedef unsigned long v1i64_2 __attribute__((__vector_size__(8)));
    245 v1i64_2 f35(v1i64_2 arg) { return arg+arg; }
    246 
    247 // rdar://9122143
    248 // CHECK: declare void @func(%struct._str* byval align 16)
    249 typedef struct _str {
    250   union {
    251     long double a;
    252     long c;
    253   };
    254 } str;
    255 
    256 void func(str s);
    257 str ss;
    258 void f9122143()
    259 {
    260   func(ss);
    261 }
    262 
    263 // CHECK: define double @f36(double %arg.coerce)
    264 typedef unsigned v2i32 __attribute((__vector_size__(8)));
    265 v2i32 f36(v2i32 arg) { return arg; }
    266 
    267 // AVX: declare void @f38(<8 x float>)
    268 // AVX: declare void @f37(<8 x float>)
    269 // CHECK: declare void @f38(%struct.s256* byval align 32)
    270 // CHECK: declare void @f37(<8 x float>* byval align 32)
    271 typedef float __m256 __attribute__ ((__vector_size__ (32)));
    272 typedef struct {
    273   __m256 m;
    274 } s256;
    275 
    276 s256 x38;
    277 __m256 x37;
    278 
    279 void f38(s256 x);
    280 void f37(__m256 x);
    281 void f39() { f38(x38); f37(x37); }
    282 
    283 // The two next tests make sure that the struct below is passed
    284 // in the same way regardless of avx being used
    285 
    286 // CHECK: declare void @func40(%struct.t128* byval align 16)
    287 typedef float __m128 __attribute__ ((__vector_size__ (16)));
    288 typedef struct t128 {
    289   __m128 m;
    290   __m128 n;
    291 } two128;
    292 
    293 extern void func40(two128 s);
    294 void func41(two128 s) {
    295   func40(s);
    296 }
    297 
    298 // CHECK: declare void @func42(%struct.t128_2* byval align 16)
    299 typedef struct xxx {
    300   __m128 array[2];
    301 } Atwo128;
    302 typedef struct t128_2 {
    303   Atwo128 x;
    304 } SA;
    305 
    306 extern void func42(SA s);
    307 void func43(SA s) {
    308   func42(s);
    309 }
    310 
    311 // CHECK: define i32 @f44
    312 // CHECK: ptrtoint
    313 // CHECK-NEXT: and {{.*}}, -32
    314 // CHECK-NEXT: inttoptr
    315 typedef int T44 __attribute((vector_size(32)));
    316 struct s44 { T44 x; int y; };
    317 int f44(int i, ...) {
    318   __builtin_va_list ap;
    319   __builtin_va_start(ap, i);
    320   struct s44 s = __builtin_va_arg(ap, struct s44);
    321   __builtin_va_end(ap);
    322   return s.y;
    323 }
    324 
    325 // Text that vec3 returns the correct LLVM IR type.
    326 // AVX: define i32 @foo(<3 x i64> %X)
    327 typedef long long3 __attribute((ext_vector_type(3)));
    328 int foo(long3 X)
    329 {
    330   return 0;
    331 }
    332 
    333 // Make sure we don't use a varargs convention for a function without a
    334 // prototype where AVX types are involved.
    335 // AVX: @test45
    336 // AVX: call i32 bitcast (i32 (...)* @f45 to i32 (<8 x float>)*)
    337 int f45();
    338 __m256 x45;
    339 void test45() { f45(x45); }
    340 
    341 // Make sure we use byval to pass 64-bit vectors in memory; the LLVM call
    342 // lowering can't handle this case correctly because it runs after legalization.
    343 // CHECK: @test46
    344 // CHECK: call void @f46({{.*}}<2 x float>* byval align 8 {{.*}}, <2 x float>* byval align 8 {{.*}})
    345 typedef float v46 __attribute((vector_size(8)));
    346 void f46(v46,v46,v46,v46,v46,v46,v46,v46,v46,v46);
    347 void test46() { v46 x = {1,2}; f46(x,x,x,x,x,x,x,x,x,x); }
    348 
    349 // Check that we pass the struct below without using byval, which helps out
    350 // codegen.
    351 //
    352 // CHECK: @test47
    353 // CHECK: call void @f47(i32 {{.*}}, i32 {{.*}}, i32 {{.*}}, i32 {{.*}}, i32 {{.*}}, i32 {{.*}}, i32 {{.*}})
    354 struct s47 { unsigned a; };
    355 void f47(int,int,int,int,int,int,struct s47);
    356 void test47(int a, struct s47 b) { f47(a, a, a, a, a, a, b); }
    357 
    358 // rdar://12723368
    359 // In the following example, there are holes in T4 at the 3rd byte and the 4th
    360 // byte, however, T2 does not have those holes. T4 is chosen to be the
    361 // representing type for union T1, but we can't use load or store of T4 since
    362 // it will skip the 3rd byte and the 4th byte.
    363 // In general, Since we don't accurately represent the data fields of a union,
    364 // do not use load or store of the representing llvm type for the union.
    365 typedef _Complex int T2;
    366 typedef _Complex char T5;
    367 typedef _Complex int T7;
    368 typedef struct T4 { T5 field0; T7 field1; } T4;
    369 typedef union T1 { T2 field0; T4 field1; } T1;
    370 extern T1 T1_retval;
    371 T1 test48(void) {
    372 // CHECK: @test48
    373 // CHECK: memcpy
    374 // CHECK: memcpy
    375   return T1_retval;
    376 }
    377 
    378 void test49_helper(double, ...);
    379 void test49(double d, double e) {
    380   test49_helper(d, e);
    381 }
    382 // CHECK:    define void @test49(
    383 // CHECK:      [[T0:%.*]] = load double*
    384 // CHECK-NEXT: [[T1:%.*]] = load double*
    385 // CHECK-NEXT: call void (double, ...)* @test49_helper(double [[T0]], double [[T1]])
    386 
    387 void test50_helper();
    388 void test50(double d, double e) {
    389   test50_helper(d, e);
    390 }
    391 // CHECK:    define void @test50(
    392 // CHECK:      [[T0:%.*]] = load double*
    393 // CHECK-NEXT: [[T1:%.*]] = load double*
    394 // CHECK-NEXT: call void (double, double, ...)* bitcast (void (...)* @test50_helper to void (double, double, ...)*)(double [[T0]], double [[T1]])
    395 
    396 struct test51_s { __uint128_t intval; };
    397 void test51(struct test51_s *s, __builtin_va_list argList) {
    398     *s = __builtin_va_arg(argList, struct test51_s);
    399 }
    400 
    401 // CHECK: define void @test51
    402 // CHECK: [[TMP_ADDR:%.*]] = alloca [[STRUCT_TEST51:%.*]], align 16
    403 // CHECK: br i1
    404 // CHECK: [[REG_SAVE_AREA_PTR:%.*]] = getelementptr inbounds {{.*}}, i32 0, i32 3
    405 // CHECK-NEXT: [[REG_SAVE_AREA:%.*]] = load i8** [[REG_SAVE_AREA_PTR]]
    406 // CHECK-NEXT: [[VALUE_ADDR:%.*]] = getelementptr i8* [[REG_SAVE_AREA]], i32 {{.*}}
    407 // CHECK-NEXT: [[CASTED_VALUE_ADDR:%.*]] = bitcast i8* [[VALUE_ADDR]] to [[STRUCT_TEST51]]
    408 // CHECK-NEXT: [[CASTED_TMP_ADDR:%.*]] = bitcast [[STRUCT_TEST51]]* [[TMP_ADDR]] to i8*
    409 // CHECK-NEXT: [[RECASTED_VALUE_ADDR:%.*]] = bitcast [[STRUCT_TEST51]]* [[CASTED_VALUE_ADDR]] to i8*
    410 // CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[CASTED_TMP_ADDR]], i8* [[RECASTED_VALUE_ADDR]], i64 16, i32 8, i1 false)
    411 // CHECK-NEXT: add i32 {{.*}}, 16
    412 // CHECK-NEXT: store i32 {{.*}}, i32* {{.*}}
    413 // CHECK-NEXT: br label
    414 
    415 void test52_helper(int, ...);
    416 __m256 x52;
    417 void test52() {
    418   test52_helper(0, x52, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0i);
    419 }
    420 // AVX: @test52_helper(i32 0, <8 x float> {{%[a-zA-Z0-9]+}}, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double {{%[a-zA-Z0-9]+}}, double {{%[a-zA-Z0-9]+}})
    421 
    422 void test53(__m256 *m, __builtin_va_list argList) {
    423   *m = __builtin_va_arg(argList, __m256);
    424 }
    425 // AVX: define void @test53
    426 // AVX-NOT: br i1
    427 // AVX: ret void
    428 
    429 void test54_helper(__m256, ...);
    430 __m256 x54;
    431 void test54() {
    432   test54_helper(x54, x54, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0i);
    433   test54_helper(x54, x54, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0i);
    434 }
    435 // AVX: @test54_helper(<8 x float> {{%[a-zA-Z0-9]+}}, <8 x float> {{%[a-zA-Z0-9]+}}, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double {{%[a-zA-Z0-9]+}}, double {{%[a-zA-Z0-9]+}})
    436 // AVX: @test54_helper(<8 x float> {{%[a-zA-Z0-9]+}}, <8 x float> {{%[a-zA-Z0-9]+}}, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, { double, double }* byval align 8 {{%[a-zA-Z0-9]+}})
    437