lldb/unittests/Core/DumpDataExtractorTest.cpp

   1 //===-- DataDumpExtractorTest.cpp -----------------------------------------===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8
   9 #include "lldb/Core/DumpDataExtractor.h"
  10 #include "lldb/Host/FileSystem.h"
  11 #include "lldb/Host/HostInfo.h"
  12 #include "lldb/Utility/DataBufferHeap.h"
  13 #include "lldb/Utility/DataExtractor.h"
  14 #include "lldb/Utility/Endian.h"
  15 #include "lldb/Utility/StreamString.h"
  16 #include "gtest/gtest.h"
  17 #include <complex>
  18 #include <limits>
  19
  20 using namespace lldb;
  21 using namespace lldb_private;
  22
  23 // This is needed for the tests because they rely on the Target global
  24 // properties.
  25 class DumpDataExtractorTest : public ::testing::Test {
  26 public:
  27   void SetUp() override {
  28     FileSystem::Initialize();
  29     HostInfo::Initialize();
  30   }
  31   void TearDown() override {
  32     HostInfo::Terminate();
  33     FileSystem::Terminate();
  34   }
  35 };
  36
  37 static void TestDumpWithAddress(uint64_t base_addr, size_t item_count,
  38                                 llvm::StringRef expected) {
  39   std::vector<uint8_t> data{0x11, 0x22};
  40   StreamString result;
  41   DataBufferHeap dumpbuffer(&data[0], data.size());
  42   DataExtractor extractor(dumpbuffer.GetBytes(), dumpbuffer.GetByteSize(),
  43                           endian::InlHostByteOrder(), /*addr_size=*/4);
  44
  45   DumpDataExtractor(extractor, &result, 0, lldb::Format::eFormatHex,
  46                     /*item_byte_size=*/1, item_count,
  47                     /*num_per_line=*/1, base_addr, 0, 0);
  48   ASSERT_EQ(expected, result.GetString());
  49 }
  50
  51 TEST_F(DumpDataExtractorTest, BaseAddress) {
  52   TestDumpWithAddress(0x12341234, 1, "0x12341234: 0x11");
  53   TestDumpWithAddress(LLDB_INVALID_ADDRESS, 1, "0x11");
  54   TestDumpWithAddress(0x12341234, 2, "0x12341234: 0x11\n0x12341235: 0x22");
  55   TestDumpWithAddress(LLDB_INVALID_ADDRESS, 2, "0x11\n0x22");
  56 }
  57
  58 static void TestDumpWithOffset(offset_t start_offset,
  59                                llvm::StringRef expected) {
  60   std::vector<uint8_t> data{0x11, 0x22, 0x33};
  61   StreamString result;
  62   DataBufferHeap dumpbuffer(&data[0], data.size());
  63   DataExtractor extractor(dumpbuffer.GetBytes(), dumpbuffer.GetByteSize(),
  64                           endian::InlHostByteOrder(), /*addr_size=*/4);
  65
  66   DumpDataExtractor(extractor, &result, start_offset, lldb::Format::eFormatHex,
  67                     /*item_byte_size=*/1, /*item_count=*/data.size(),
  68                     /*num_per_line=*/data.size(), /*base_addr=*/0, 0, 0);
  69   ASSERT_EQ(expected, result.GetString());
  70 }
  71
  72 TEST_F(DumpDataExtractorTest, StartOffset) {
  73   TestDumpWithOffset(0, "0x00000000: 0x11 0x22 0x33");
  74   // The offset applies to the DataExtractor, not the address used when
  75   // formatting.
  76   TestDumpWithOffset(1, "0x00000000: 0x22 0x33");
  77   // If the offset is outside the DataExtractor's range we do nothing.
  78   TestDumpWithOffset(3, "");
  79 }
  80
  81 TEST_F(DumpDataExtractorTest, NullStream) {
  82   // We don't do any work if there is no output stream.
  83   uint8_t c = 0x11;
  84   StreamString result;
  85   DataBufferHeap dumpbuffer(&c, 0);
  86   DataExtractor extractor(dumpbuffer.GetBytes(), dumpbuffer.GetByteSize(),
  87                           endian::InlHostByteOrder(), /*addr_size=*/4);
  88
  89   DumpDataExtractor(extractor, nullptr, 0, lldb::Format::eFormatHex,
  90                     /*item_byte_size=*/1, /*item_count=*/1,
  91                     /*num_per_line=*/1, /*base_addr=*/0, 0, 0);
  92   ASSERT_EQ("", result.GetString());
  93 }
  94
  95 static void TestDumpImpl(const void *data, size_t data_size,
  96                          size_t item_byte_size, size_t item_count,
  97                          size_t num_per_line, uint64_t base_addr,
  98                          lldb::Format format, llvm::StringRef expected) {
  99   StreamString result;
 100   DataBufferHeap dumpbuffer(data, data_size);
 101   DataExtractor extractor(dumpbuffer.GetBytes(), dumpbuffer.GetByteSize(),
 102                           endian::InlHostByteOrder(),
 103                           /*addr_size=*/4);
 104   DumpDataExtractor(extractor, &result, 0, format, item_byte_size, item_count,
 105                     num_per_line, base_addr, 0, 0);
 106   ASSERT_EQ(expected, result.GetString());
 107 }
 108
 109 template <typename T>
 110 static void TestDump(T data, lldb::Format format, llvm::StringRef expected) {
 111   TestDumpImpl(&data, sizeof(T), sizeof(T), 1, 1, LLDB_INVALID_ADDRESS, format,
 112                expected);
 113 }
 114
 115 static void TestDump(llvm::StringRef str, lldb::Format format,
 116                      llvm::StringRef expected) {
 117   TestDumpImpl(str.bytes_begin(),
 118                // +1 to include the NULL char as the last byte
 119                str.size() + 1, str.size() + 1, 1, 1, LLDB_INVALID_ADDRESS,
 120                format, expected);
 121 }
 122
 123 template <typename T>
 124 static void TestDump(const std::vector<T> data, lldb::Format format,
 125                      llvm::StringRef expected) {
 126   size_t sz_bytes = data.size() * sizeof(T);
 127   TestDumpImpl(&data[0], sz_bytes, sz_bytes, data.size(), 1,
 128                LLDB_INVALID_ADDRESS, format, expected);
 129 }
 130
 131 TEST_F(DumpDataExtractorTest, Formats) {
 132   TestDump<uint8_t>(1, lldb::eFormatDefault, "0x01");
 133   TestDump<uint8_t>(1, lldb::eFormatBoolean, "true");
 134   TestDump<uint8_t>(0xAA, lldb::eFormatBinary, "0b10101010");
 135   TestDump<uint8_t>(1, lldb::eFormatBytes, "01");
 136   TestDump<uint8_t>(1, lldb::eFormatBytesWithASCII, "01  .");
 137   TestDump('?', lldb::eFormatChar, "'?'");
 138   TestDump('\x1A', lldb::eFormatCharPrintable, ".");
 139   TestDump('#', lldb::eFormatCharPrintable, "#");
 140   TestDump(std::complex<float>(1.2, 3.4), lldb::eFormatComplex, "1.2 + 3.4i");
 141   TestDump(std::complex<double>(4.5, 6.7), lldb::eFormatComplex, "4.5 + 6.7i");
 142
 143   // long double is not tested here because for some platforms we treat it as 10
 144   // bytes when the compiler allocates 16 bytes of space for it. (see
 145   // DataExtractor::GetLongDouble) Meaning that when we extract the second one,
 146   // it gets the wrong value (it's 6 bytes off). You could manually construct a
 147   // set of bytes to match the 10 byte format but then if the test runs on a
 148   // machine where we don't use 10 it'll break.
 149
 150   // Test printable characters.
 151   TestDump(llvm::StringRef("aardvark"), lldb::Format::eFormatCString,
 152            "\"aardvark\"");
 153   // Test unprintable characters.
 154   TestDump(llvm::StringRef("\xcf\xfa\xed\xfe\f"), lldb::Format::eFormatCString,
 155            "\"\\xcf\\xfa\\xed\\xfe\\f\"");
 156   // Test a mix of printable and unprintable characters.
 157   TestDump(llvm::StringRef("\xcf\xfa\ffoo"), lldb::Format::eFormatCString,
 158            "\"\\xcf\\xfa\\ffoo\"");
 159
 160   TestDump<uint16_t>(99, lldb::Format::eFormatDecimal, "99");
 161   // Just prints as a signed integer.
 162   TestDump(-1, lldb::Format::eFormatEnum, "-1");
 163   TestDump(0xcafef00d, lldb::Format::eFormatHex, "0xcafef00d");
 164   TestDump(0xcafef00d, lldb::Format::eFormatHexUppercase, "0xCAFEF00D");
 165   TestDump(0.456, lldb::Format::eFormatFloat, "0.45600000000000002");
 166   TestDump(9, lldb::Format::eFormatOctal, "011");
 167   // Chars packed into an integer.
 168   TestDump<uint32_t>(0x4C4C4442, lldb::Format::eFormatOSType, "'LLDB'");
 169   // Unicode8 doesn't have a specific formatter.
 170   TestDump<uint8_t>(0x34, lldb::Format::eFormatUnicode8, "0x34");
 171   TestDump<uint16_t>(0x1122, lldb::Format::eFormatUnicode16, "U+1122");
 172   TestDump<uint32_t>(0x12345678, lldb::Format::eFormatUnicode32,
 173                      "U+0x12345678");
 174   TestDump<unsigned int>(654321, lldb::Format::eFormatUnsigned, "654321");
 175   // This pointer is printed based on the size of uint64_t, so the test is the
 176   // same for 32/64 bit host.
 177   TestDump<uint64_t>(0x4444555566667777, lldb::Format::eFormatPointer,
 178                      "0x4444555566667777");
 179
 180   TestDump(std::vector<char>{'A', '\x01', 'C'},
 181            lldb::Format::eFormatVectorOfChar, "{A\\x01C}");
 182   TestDump(std::vector<int8_t>{0, -1, std::numeric_limits<int8_t>::max()},
 183            lldb::Format::eFormatVectorOfSInt8, "{0 -1 127}");
 184   TestDump(std::vector<uint8_t>{12, 0xFF, 34},
 185            lldb::Format::eFormatVectorOfUInt8, "{0x0c 0xff 0x22}");
 186   TestDump(std::vector<int16_t>{-1, 1234, std::numeric_limits<int16_t>::max()},
 187            lldb::Format::eFormatVectorOfSInt16, "{-1 1234 32767}");
 188   TestDump(std::vector<uint16_t>{0xffff, 0xabcd, 0x1234},
 189            lldb::Format::eFormatVectorOfUInt16, "{0xffff 0xabcd 0x1234}");
 190   TestDump(std::vector<int32_t>{0, -1, std::numeric_limits<int32_t>::max()},
 191            lldb::Format::eFormatVectorOfSInt32, "{0 -1 2147483647}");
 192   TestDump(std::vector<uint32_t>{0, 0xffffffff, 0x1234abcd},
 193            lldb::Format::eFormatVectorOfUInt32,
 194            "{0x00000000 0xffffffff 0x1234abcd}");
 195   TestDump(std::vector<int64_t>{0, -1, std::numeric_limits<int64_t>::max()},
 196            lldb::Format::eFormatVectorOfSInt64, "{0 -1 9223372036854775807}");
 197   TestDump(std::vector<uint64_t>{0, 0xaaaabbbbccccdddd},
 198            lldb::Format::eFormatVectorOfUInt64,
 199            "{0x0000000000000000 0xaaaabbbbccccdddd}");
 200
 201   // See half2float for format details.
 202   // Test zeroes.
 203   TestDump(std::vector<uint16_t>{0x0000, 0x8000},
 204            lldb::Format::eFormatVectorOfFloat16, "{0 -0}");
 205   // Some subnormal numbers.
 206   TestDump(std::vector<uint16_t>{0x0001, 0x8001},
 207            lldb::Format::eFormatVectorOfFloat16, "{5.9605E-8 -5.9605E-8}");
 208   // A full mantisse and empty expontent.
 209   TestDump(std::vector<uint16_t>{0x83ff, 0x03ff},
 210            lldb::Format::eFormatVectorOfFloat16, "{-6.0976E-5 6.0976E-5}");
 211   // Some normal numbers.
 212   TestDump(std::vector<uint16_t>{0b0100001001001000},
 213            lldb::Format::eFormatVectorOfFloat16, "{3.1406}");
 214   // Largest and smallest normal number.
 215   TestDump(std::vector<uint16_t>{0x0400, 0x7bff},
 216            lldb::Format::eFormatVectorOfFloat16, "{6.1035E-5 65504}");
 217   TestDump(std::vector<uint16_t>{0xabcd, 0x1234},
 218            lldb::Format::eFormatVectorOfFloat16, "{-0.060944 7.5722E-4}");
 219
 220   // quiet/signaling NaNs.
 221   TestDump(std::vector<uint16_t>{0xffff, 0xffc0, 0x7fff, 0x7fc0},
 222            lldb::Format::eFormatVectorOfFloat16, "{NaN NaN NaN NaN}");
 223   // +/-Inf.
 224   TestDump(std::vector<uint16_t>{0xfc00, 0x7c00},
 225            lldb::Format::eFormatVectorOfFloat16, "{-Inf +Inf}");
 226
 227   TestDump(std::vector<float>{std::numeric_limits<float>::min(),
 228                               std::numeric_limits<float>::max()},
 229            lldb::Format::eFormatVectorOfFloat32,
 230            "{1.17549435E-38 3.40282347E+38}");
 231   TestDump(std::vector<float>{std::numeric_limits<float>::quiet_NaN(),
 232                               std::numeric_limits<float>::signaling_NaN(),
 233                               -std::numeric_limits<float>::quiet_NaN(),
 234                               -std::numeric_limits<float>::signaling_NaN()},
 235            lldb::Format::eFormatVectorOfFloat32, "{NaN NaN NaN NaN}");
 236   TestDump(std::vector<double>{std::numeric_limits<double>::min(),
 237                                std::numeric_limits<double>::max()},
 238            lldb::Format::eFormatVectorOfFloat64,
 239            "{2.2250738585072014E-308 1.7976931348623157E+308}");
 240   TestDump(
 241       std::vector<double>{
 242           std::numeric_limits<double>::quiet_NaN(),
 243           std::numeric_limits<double>::signaling_NaN(),
 244           -std::numeric_limits<double>::quiet_NaN(),
 245           -std::numeric_limits<double>::signaling_NaN(),
 246       },
 247       lldb::Format::eFormatVectorOfFloat64, "{NaN NaN NaN NaN}");
 248
 249   // Not sure we can rely on having uint128_t everywhere so emulate with
 250   // uint64_t.
 251   TestDump(
 252       std::vector<uint64_t>{0x1, 0x1111222233334444, 0xaaaabbbbccccdddd, 0x0},
 253       lldb::Format::eFormatVectorOfUInt128,
 254       "{0x11112222333344440000000000000001 "
 255       "0x0000000000000000aaaabbbbccccdddd}");
 256
 257   TestDump(std::vector<int>{2, 4}, lldb::Format::eFormatComplexInteger,
 258            "2 + 4i");
 259
 260   // Without an execution context this just prints the pointer on its own.
 261   TestDump<uint32_t>(0x11223344, lldb::Format::eFormatAddressInfo,
 262                      "0x11223344");
 263
 264   // Input not written in hex form because that requires C++17.
 265   TestDump<float>(10, lldb::Format::eFormatHexFloat, "0x1.4p3");
 266   TestDump<double>(10, lldb::Format::eFormatHexFloat, "0x1.4p3");
 267   // long double not supported, see ItemByteSizeErrors.
 268
 269   // Can't disassemble without an execution context.
 270   TestDump<uint32_t>(0xcafef00d, lldb::Format::eFormatInstruction,
 271                      "invalid target");
 272
 273   // Has no special handling, intended for use elsewhere.
 274   TestDump<int>(99, lldb::Format::eFormatVoid, "0x00000063");
 275 }
 276
 277 TEST_F(DumpDataExtractorTest, FormatCharArray) {
 278   // Unlike the other formats, charArray isn't 1 array of N chars.
 279   // It must be passed as N chars of 1 byte each.
 280   // (eFormatVectorOfChar does this swap for you)
 281   std::vector<char> data{'A', '\x01', '#'};
 282   StreamString result;
 283   DataBufferHeap dumpbuffer(&data[0], data.size());
 284   DataExtractor extractor(dumpbuffer.GetBytes(), dumpbuffer.GetByteSize(),
 285                           endian::InlHostByteOrder(), /*addr_size=*/4);
 286
 287   DumpDataExtractor(extractor, &result, 0, lldb::Format::eFormatCharArray,
 288                     /*item_byte_size=*/1,
 289                     /*item_count=*/data.size(),
 290                     /*num_per_line=*/data.size(), 0, 0, 0);
 291   ASSERT_EQ("0x00000000: A\\x01#", result.GetString());
 292
 293   result.Clear();
 294   DumpDataExtractor(extractor, &result, 0, lldb::Format::eFormatCharArray, 1,
 295                     data.size(), 1, 0, 0, 0);
 296   // ASSERT macro thinks the split strings are multiple arguments so make a var.
 297   const char *expected = "0x00000000: A\n"
 298                          "0x00000001: \\x01\n"
 299                          "0x00000002: #";
 300   ASSERT_EQ(expected, result.GetString());
 301 }
 302
 303 template <typename T>
 304 void TestDumpMultiLine(std::vector<T> data, lldb::Format format,
 305                        size_t num_per_line, llvm::StringRef expected) {
 306   size_t sz_bytes = data.size() * sizeof(T);
 307   TestDumpImpl(&data[0], sz_bytes, data.size(), sz_bytes, num_per_line,
 308                0x80000000, format, expected);
 309 }
 310
 311 template <typename T>
 312 void TestDumpMultiLine(const T *data, size_t num_items, lldb::Format format,
 313                        size_t num_per_line, llvm::StringRef expected) {
 314   TestDumpImpl(data, sizeof(T) * num_items, sizeof(T), num_items, num_per_line,
 315                0x80000000, format, expected);
 316 }
 317
 318 TEST_F(DumpDataExtractorTest, MultiLine) {
 319   // A vector counts as 1 item regardless of size.
 320   TestDumpMultiLine(std::vector<uint8_t>{0x11},
 321                     lldb::Format::eFormatVectorOfUInt8, 1,
 322                     "0x80000000: {0x11}");
 323   TestDumpMultiLine(std::vector<uint8_t>{0x11, 0x22},
 324                     lldb::Format::eFormatVectorOfUInt8, 1,
 325                     "0x80000000: {0x11 0x22}");
 326
 327   // If you have multiple vectors then that's multiple items.
 328   // Here we say that these 2 bytes are actually 2 1 byte vectors.
 329   const std::vector<uint8_t> vector_data{0x11, 0x22};
 330   TestDumpMultiLine(vector_data.data(), 2, lldb::Format::eFormatVectorOfUInt8,
 331                     1, "0x80000000: {0x11}\n0x80000001: {0x22}");
 332
 333   // Single value formats can span multiple lines.
 334   const std::vector<uint8_t> bytes{0x11, 0x22, 0x33};
 335   const char *expected_bytes_3_line = "0x80000000: 0x11\n"
 336                                       "0x80000001: 0x22\n"
 337                                       "0x80000002: 0x33";
 338   TestDumpMultiLine(bytes.data(), bytes.size(), lldb::Format::eFormatHex, 1,
 339                     expected_bytes_3_line);
 340
 341   // Lines may not have the full number of items.
 342   TestDumpMultiLine(bytes.data(), bytes.size(), lldb::Format::eFormatHex, 4,
 343                     "0x80000000: 0x11 0x22 0x33");
 344   const char *expected_bytes_2_line = "0x80000000: 0x11 0x22\n"
 345                                       "0x80000002: 0x33";
 346   TestDumpMultiLine(bytes.data(), bytes.size(), lldb::Format::eFormatHex, 2,
 347                     expected_bytes_2_line);
 348
 349   // The line address accounts for item sizes other than 1 byte.
 350   const std::vector<uint16_t> shorts{0x1111, 0x2222, 0x3333};
 351   const char *expected_shorts_2_line = "0x80000000: 0x1111 0x2222\n"
 352                                        "0x80000004: 0x3333";
 353   TestDumpMultiLine(shorts.data(), shorts.size(), lldb::Format::eFormatHex, 2,
 354                     expected_shorts_2_line);
 355
 356   // The ascii column is positioned using the maximum line length.
 357   const std::vector<char> chars{'L', 'L', 'D', 'B'};
 358   const char *expected_chars_2_lines = "0x80000000: 4c 4c 44  LLD\n"
 359                                        "0x80000003: 42        B";
 360   TestDumpMultiLine(chars.data(), chars.size(),
 361                     lldb::Format::eFormatBytesWithASCII, 3,
 362                     expected_chars_2_lines);
 363 }
 364
 365 void TestDumpWithItemByteSize(size_t item_byte_size, lldb::Format format,
 366                               llvm::StringRef expected) {
 367   // We won't be reading this data so anything will do.
 368   uint8_t dummy = 0;
 369   TestDumpImpl(&dummy, 1, item_byte_size, 1, 1, LLDB_INVALID_ADDRESS, format,
 370                expected);
 371 }
 372
 373 TEST_F(DumpDataExtractorTest, ItemByteSizeErrors) {
 374   TestDumpWithItemByteSize(
 375       16, lldb::Format::eFormatBoolean,
 376       "error: unsupported byte size (16) for boolean format");
 377   TestDumpWithItemByteSize(21, lldb::Format::eFormatChar,
 378                            "error: unsupported byte size (21) for char format");
 379   TestDumpWithItemByteSize(
 380       18, lldb::Format::eFormatComplexInteger,
 381       "error: unsupported byte size (18) for complex integer format");
 382
 383   // The code uses sizeof(long double) for these checks. This changes by host
 384   // but we know it won't be >16.
 385   TestDumpWithItemByteSize(
 386       34, lldb::Format::eFormatComplex,
 387       "error: unsupported byte size (34) for complex float format");
 388   TestDumpWithItemByteSize(
 389       18, lldb::Format::eFormatFloat,
 390       "error: unsupported byte size (18) for float format");
 391
 392   // We want sizes to exactly match one of float/double.
 393   TestDumpWithItemByteSize(
 394       14, lldb::Format::eFormatComplex,
 395       "error: unsupported byte size (14) for complex float format");
 396   TestDumpWithItemByteSize(3, lldb::Format::eFormatFloat,
 397                            "error: unsupported byte size (3) for float format");
 398
 399   // We only allow float and double size.
 400   TestDumpWithItemByteSize(
 401       1, lldb::Format::eFormatHexFloat,
 402       "error: unsupported byte size (1) for hex float format");
 403   TestDumpWithItemByteSize(
 404       17, lldb::Format::eFormatHexFloat,
 405       "error: unsupported byte size (17) for hex float format");
 406 }