test-macro-assembler-arm.cc 12.8 KB
Newer Older
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
// Copyright 2013 the V8 project authors. All rights reserved.
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
//     * Redistributions of source code must retain the above copyright
//       notice, this list of conditions and the following disclaimer.
//     * Redistributions in binary form must reproduce the above
//       copyright notice, this list of conditions and the following
//       disclaimer in the documentation and/or other materials provided
//       with the distribution.
//     * Neither the name of Google Inc. nor the names of its
//       contributors may be used to endorse or promote products derived
//       from this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

#include <stdlib.h>

30 31 32
#include "src/assembler-inl.h"
#include "src/macro-assembler.h"
#include "src/objects-inl.h"
33
#include "src/simulator.h"
34
#include "src/v8.h"
35
#include "test/cctest/cctest.h"
36

37 38
namespace v8 {
namespace internal {
39
namespace test_macro_assembler_arm {
40

41
using F = void*(int x, int y, int p2, int p3, int p4);
42 43 44

#define __ masm->

45 46
using F3 = Object*(void* p0, int p1, int p2, int p3, int p4);
using F5 = int(void*, void*, void*, void*, void*);
47 48 49 50

TEST(LoadAndStoreWithRepresentation) {
  Isolate* isolate = CcTest::i_isolate();
  HandleScope handles(isolate);
51 52 53 54

  size_t allocated;
  byte* buffer = AllocateAssemblerBuffer(&allocated);
  MacroAssembler assembler(isolate, buffer, static_cast<int>(allocated),
55
                           v8::internal::CodeObjectRequired::kYes);
56
  MacroAssembler* masm = &assembler;  // Create a pointer for the __ macro.
57

58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126
  __ sub(sp, sp, Operand(1 * kPointerSize));
  Label exit;

  // Test 1.
  __ mov(r0, Operand(1));  // Test number.
  __ mov(r1, Operand(0));
  __ str(r1, MemOperand(sp, 0 * kPointerSize));
  __ mov(r2, Operand(-1));
  __ Store(r2, MemOperand(sp, 0 * kPointerSize), Representation::UInteger8());
  __ ldr(r3, MemOperand(sp, 0 * kPointerSize));
  __ mov(r2, Operand(255));
  __ cmp(r3, r2);
  __ b(ne, &exit);
  __ mov(r2, Operand(255));
  __ Load(r3, MemOperand(sp, 0 * kPointerSize), Representation::UInteger8());
  __ cmp(r3, r2);
  __ b(ne, &exit);

  // Test 2.
  __ mov(r0, Operand(2));  // Test number.
  __ mov(r1, Operand(0));
  __ str(r1, MemOperand(sp, 0 * kPointerSize));
  __ mov(r2, Operand(-1));
  __ Store(r2, MemOperand(sp, 0 * kPointerSize), Representation::Integer8());
  __ ldr(r3, MemOperand(sp, 0 * kPointerSize));
  __ mov(r2, Operand(255));
  __ cmp(r3, r2);
  __ b(ne, &exit);
  __ mov(r2, Operand(-1));
  __ Load(r3, MemOperand(sp, 0 * kPointerSize), Representation::Integer8());
  __ cmp(r3, r2);
  __ b(ne, &exit);

  // Test 3.
  __ mov(r0, Operand(3));  // Test number.
  __ mov(r1, Operand(0));
  __ str(r1, MemOperand(sp, 0 * kPointerSize));
  __ mov(r2, Operand(-1));
  __ Store(r2, MemOperand(sp, 0 * kPointerSize), Representation::UInteger16());
  __ ldr(r3, MemOperand(sp, 0 * kPointerSize));
  __ mov(r2, Operand(65535));
  __ cmp(r3, r2);
  __ b(ne, &exit);
  __ mov(r2, Operand(65535));
  __ Load(r3, MemOperand(sp, 0 * kPointerSize), Representation::UInteger16());
  __ cmp(r3, r2);
  __ b(ne, &exit);

  // Test 4.
  __ mov(r0, Operand(4));  // Test number.
  __ mov(r1, Operand(0));
  __ str(r1, MemOperand(sp, 0 * kPointerSize));
  __ mov(r2, Operand(-1));
  __ Store(r2, MemOperand(sp, 0 * kPointerSize), Representation::Integer16());
  __ ldr(r3, MemOperand(sp, 0 * kPointerSize));
  __ mov(r2, Operand(65535));
  __ cmp(r3, r2);
  __ b(ne, &exit);
  __ mov(r2, Operand(-1));
  __ Load(r3, MemOperand(sp, 0 * kPointerSize), Representation::Integer16());
  __ cmp(r3, r2);
  __ b(ne, &exit);

  __ mov(r0, Operand(0));  // Success.
  __ bind(&exit);
  __ add(sp, sp, Operand(1 * kPointerSize));
  __ bx(lr);

  CodeDesc desc;
127
  masm->GetCode(isolate, &desc);
128 129
  Handle<Code> code =
      isolate->factory()->NewCode(desc, Code::STUB, Handle<Code>());
130 131

  // Call the function from C++.
132 133
  auto f = GeneratedCode<F5>::FromCode(*code);
  CHECK(!f.Call(0, 0, 0, 0, 0));
134
}
135

136 137 138 139 140
TEST(ExtractLane) {
  if (!CpuFeatures::IsSupported(NEON)) return;

  Isolate* isolate = CcTest::i_isolate();
  HandleScope handles(isolate);
141 142 143 144

  size_t allocated;
  byte* buffer = AllocateAssemblerBuffer(&allocated);
  MacroAssembler assembler(isolate, buffer, static_cast<int>(allocated),
145 146 147 148 149 150 151 152 153 154 155 156
                           v8::internal::CodeObjectRequired::kYes);
  MacroAssembler* masm = &assembler;  // Create a pointer for the __ macro.

  typedef struct {
    int32_t i32x4_low[4];
    int32_t i32x4_high[4];
    int32_t i16x8_low[8];
    int32_t i16x8_high[8];
    int32_t i8x16_low[16];
    int32_t i8x16_high[16];
    int32_t f32x4_low[4];
    int32_t f32x4_high[4];
157 158
    int32_t i8x16_low_d[16];
    int32_t i8x16_high_d[16];
159 160 161 162 163 164 165 166 167 168 169
  } T;
  T t;

  __ stm(db_w, sp, r4.bit() | r5.bit() | lr.bit());

  for (int i = 0; i < 4; i++) {
    __ mov(r4, Operand(i));
    __ vdup(Neon32, q1, r4);
    __ ExtractLane(r5, q1, NeonS32, i);
    __ str(r5, MemOperand(r0, offsetof(T, i32x4_low) + 4 * i));
    SwVfpRegister si = SwVfpRegister::from_code(i);
170
    __ ExtractLane(si, q1, i);
171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187
    __ vstr(si, r0, offsetof(T, f32x4_low) + 4 * i);
  }

  for (int i = 0; i < 8; i++) {
    __ mov(r4, Operand(i));
    __ vdup(Neon16, q1, r4);
    __ ExtractLane(r5, q1, NeonS16, i);
    __ str(r5, MemOperand(r0, offsetof(T, i16x8_low) + 4 * i));
  }

  for (int i = 0; i < 16; i++) {
    __ mov(r4, Operand(i));
    __ vdup(Neon8, q1, r4);
    __ ExtractLane(r5, q1, NeonS8, i);
    __ str(r5, MemOperand(r0, offsetof(T, i8x16_low) + 4 * i));
  }

188 189 190 191 192 193 194 195 196
  for (int i = 0; i < 8; i++) {
    __ mov(r4, Operand(i));
    __ vdup(Neon8, q1, r4);  // q1 = d2,d3
    __ ExtractLane(r5, d2, NeonS8, i);
    __ str(r5, MemOperand(r0, offsetof(T, i8x16_low_d) + 4 * i));
    __ ExtractLane(r5, d3, NeonS8, i);
    __ str(r5, MemOperand(r0, offsetof(T, i8x16_low_d) + 4 * (i + 8)));
  }

197 198 199 200 201 202 203
  if (CpuFeatures::IsSupported(VFP32DREGS)) {
    for (int i = 0; i < 4; i++) {
      __ mov(r4, Operand(-i));
      __ vdup(Neon32, q15, r4);
      __ ExtractLane(r5, q15, NeonS32, i);
      __ str(r5, MemOperand(r0, offsetof(T, i32x4_high) + 4 * i));
      SwVfpRegister si = SwVfpRegister::from_code(i);
204
      __ ExtractLane(si, q15, i);
205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220
      __ vstr(si, r0, offsetof(T, f32x4_high) + 4 * i);
    }

    for (int i = 0; i < 8; i++) {
      __ mov(r4, Operand(-i));
      __ vdup(Neon16, q15, r4);
      __ ExtractLane(r5, q15, NeonS16, i);
      __ str(r5, MemOperand(r0, offsetof(T, i16x8_high) + 4 * i));
    }

    for (int i = 0; i < 16; i++) {
      __ mov(r4, Operand(-i));
      __ vdup(Neon8, q15, r4);
      __ ExtractLane(r5, q15, NeonS8, i);
      __ str(r5, MemOperand(r0, offsetof(T, i8x16_high) + 4 * i));
    }
221 222 223 224 225 226 227 228 229

    for (int i = 0; i < 8; i++) {
      __ mov(r4, Operand(-i));
      __ vdup(Neon8, q15, r4);  // q1 = d30,d31
      __ ExtractLane(r5, d30, NeonS8, i);
      __ str(r5, MemOperand(r0, offsetof(T, i8x16_high_d) + 4 * i));
      __ ExtractLane(r5, d31, NeonS8, i);
      __ str(r5, MemOperand(r0, offsetof(T, i8x16_high_d) + 4 * (i + 8)));
    }
230 231 232 233 234
  }

  __ ldm(ia_w, sp, r4.bit() | r5.bit() | pc.bit());

  CodeDesc desc;
235
  masm->GetCode(isolate, &desc);
236 237
  Handle<Code> code =
      isolate->factory()->NewCode(desc, Code::STUB, Handle<Code>());
238 239 240 241
#ifdef DEBUG
  OFStream os(stdout);
  code->Print(os);
#endif
242 243
  auto f = GeneratedCode<F3>::FromCode(*code);
  f.Call(&t, 0, 0, 0, 0);
244 245 246 247 248 249 250 251 252 253
  for (int i = 0; i < 4; i++) {
    CHECK_EQ(i, t.i32x4_low[i]);
    CHECK_EQ(i, t.f32x4_low[i]);
  }
  for (int i = 0; i < 8; i++) {
    CHECK_EQ(i, t.i16x8_low[i]);
  }
  for (int i = 0; i < 16; i++) {
    CHECK_EQ(i, t.i8x16_low[i]);
  }
254 255 256 257
  for (int i = 0; i < 8; i++) {
    CHECK_EQ(i, t.i8x16_low_d[i]);
    CHECK_EQ(i, t.i8x16_low_d[i + 8]);
  }
258 259 260 261 262 263 264 265 266 267 268
  if (CpuFeatures::IsSupported(VFP32DREGS)) {
    for (int i = 0; i < 4; i++) {
      CHECK_EQ(-i, t.i32x4_high[i]);
      CHECK_EQ(-i, t.f32x4_high[i]);
    }
    for (int i = 0; i < 8; i++) {
      CHECK_EQ(-i, t.i16x8_high[i]);
    }
    for (int i = 0; i < 16; i++) {
      CHECK_EQ(-i, t.i8x16_high[i]);
    }
269 270 271 272
    for (int i = 0; i < 8; i++) {
      CHECK_EQ(-i, t.i8x16_high_d[i]);
      CHECK_EQ(-i, t.i8x16_high_d[i + 8]);
    }
273 274 275 276 277 278 279 280
  }
}

TEST(ReplaceLane) {
  if (!CpuFeatures::IsSupported(NEON)) return;

  Isolate* isolate = CcTest::i_isolate();
  HandleScope handles(isolate);
281 282 283 284

  size_t allocated;
  byte* buffer = AllocateAssemblerBuffer(&allocated);
  MacroAssembler assembler(isolate, buffer, static_cast<int>(allocated),
285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
                           v8::internal::CodeObjectRequired::kYes);
  MacroAssembler* masm = &assembler;  // Create a pointer for the __ macro.

  typedef struct {
    int32_t i32x4_low[4];
    int32_t i32x4_high[4];
    int16_t i16x8_low[8];
    int16_t i16x8_high[8];
    int8_t i8x16_low[16];
    int8_t i8x16_high[16];
    int32_t f32x4_low[4];
    int32_t f32x4_high[4];
  } T;
  T t;

  __ stm(db_w, sp, r4.bit() | r5.bit() | r6.bit() | r7.bit() | lr.bit());

  __ veor(q0, q0, q0);  // Zero
  __ veor(q1, q1, q1);  // Zero
  for (int i = 0; i < 4; i++) {
    __ mov(r4, Operand(i));
    __ ReplaceLane(q0, q0, r4, NeonS32, i);
    SwVfpRegister si = SwVfpRegister::from_code(i);
    __ vmov(si, r4);
309
    __ ReplaceLane(q1, q1, si, i);
310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339
  }
  __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, i32x4_low))));
  __ vst1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
  __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, f32x4_low))));
  __ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));

  __ veor(q0, q0, q0);  // Zero
  for (int i = 0; i < 8; i++) {
    __ mov(r4, Operand(i));
    __ ReplaceLane(q0, q0, r4, NeonS16, i);
  }
  __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, i16x8_low))));
  __ vst1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));

  __ veor(q0, q0, q0);  // Zero
  for (int i = 0; i < 16; i++) {
    __ mov(r4, Operand(i));
    __ ReplaceLane(q0, q0, r4, NeonS8, i);
  }
  __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, i8x16_low))));
  __ vst1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));

  if (CpuFeatures::IsSupported(VFP32DREGS)) {
    __ veor(q14, q14, q14);  // Zero
    __ veor(q15, q15, q15);  // Zero
    for (int i = 0; i < 4; i++) {
      __ mov(r4, Operand(-i));
      __ ReplaceLane(q14, q14, r4, NeonS32, i);
      SwVfpRegister si = SwVfpRegister::from_code(i);
      __ vmov(si, r4);
340
      __ ReplaceLane(q15, q15, si, i);
341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366
    }
    __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, i32x4_high))));
    __ vst1(Neon8, NeonListOperand(q14), NeonMemOperand(r4));
    __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, f32x4_high))));
    __ vst1(Neon8, NeonListOperand(q15), NeonMemOperand(r4));

    __ veor(q14, q14, q14);  // Zero
    for (int i = 0; i < 8; i++) {
      __ mov(r4, Operand(-i));
      __ ReplaceLane(q14, q14, r4, NeonS16, i);
    }
    __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, i16x8_high))));
    __ vst1(Neon8, NeonListOperand(q14), NeonMemOperand(r4));

    __ veor(q14, q14, q14);  // Zero
    for (int i = 0; i < 16; i++) {
      __ mov(r4, Operand(-i));
      __ ReplaceLane(q14, q14, r4, NeonS8, i);
    }
    __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, i8x16_high))));
    __ vst1(Neon8, NeonListOperand(q14), NeonMemOperand(r4));
  }

  __ ldm(ia_w, sp, r4.bit() | r5.bit() | r6.bit() | r7.bit() | pc.bit());

  CodeDesc desc;
367
  masm->GetCode(isolate, &desc);
368 369
  Handle<Code> code =
      isolate->factory()->NewCode(desc, Code::STUB, Handle<Code>());
370 371 372 373
#ifdef DEBUG
  OFStream os(stdout);
  code->Print(os);
#endif
374 375
  auto f = GeneratedCode<F3>::FromCode(*code);
  f.Call(&t, 0, 0, 0, 0);
376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399
  for (int i = 0; i < 4; i++) {
    CHECK_EQ(i, t.i32x4_low[i]);
    CHECK_EQ(i, t.f32x4_low[i]);
  }
  for (int i = 0; i < 8; i++) {
    CHECK_EQ(i, t.i16x8_low[i]);
  }
  for (int i = 0; i < 16; i++) {
    CHECK_EQ(i, t.i8x16_low[i]);
  }
  if (CpuFeatures::IsSupported(VFP32DREGS)) {
    for (int i = 0; i < 4; i++) {
      CHECK_EQ(-i, t.i32x4_high[i]);
      CHECK_EQ(-i, t.f32x4_high[i]);
    }
    for (int i = 0; i < 8; i++) {
      CHECK_EQ(-i, t.i16x8_high[i]);
    }
    for (int i = 0; i < 16; i++) {
      CHECK_EQ(-i, t.i8x16_high[i]);
    }
  }
}

400
#undef __
401

402
}  // namespace test_macro_assembler_arm
403 404
}  // namespace internal
}  // namespace v8