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