xref: /llvm-project/lldb/unittests/Core/DumpDataExtractorTest.cpp (revision ae58cf5f45a9c159afbf86e93c0c257a22c4ee02)
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   TestDump(llvm::StringRef("aardvark"), lldb::Format::eFormatCString,
135            "\"aardvark\"");
136   TestDump<uint16_t>(99, lldb::Format::eFormatDecimal, "99");
137   // Just prints as a signed integer.
138   TestDump(-1, lldb::Format::eFormatEnum, "-1");
139   TestDump(0xcafef00d, lldb::Format::eFormatHex, "0xcafef00d");
140   TestDump(0xcafef00d, lldb::Format::eFormatHexUppercase, "0xCAFEF00D");
141   TestDump(0.456, lldb::Format::eFormatFloat, "0.456");
142   TestDump(9, lldb::Format::eFormatOctal, "011");
143   // Chars packed into an integer.
144   TestDump<uint32_t>(0x4C4C4442, lldb::Format::eFormatOSType, "'LLDB'");
145   // Unicode8 doesn't have a specific formatter.
146   TestDump<uint8_t>(0x34, lldb::Format::eFormatUnicode8, "0x34");
147   TestDump<uint16_t>(0x1122, lldb::Format::eFormatUnicode16, "U+1122");
148   TestDump<uint32_t>(0x12345678, lldb::Format::eFormatUnicode32,
149                      "U+0x12345678");
150   TestDump<unsigned int>(654321, lldb::Format::eFormatUnsigned, "654321");
151   // This pointer is printed based on the size of uint64_t, so the test is the
152   // same for 32/64 bit host.
153   TestDump<uint64_t>(0x4444555566667777, lldb::Format::eFormatPointer,
154                      "0x4444555566667777");
155 
156   TestDump(std::vector<char>{'A', '\x01', 'C'},
157            lldb::Format::eFormatVectorOfChar, "{A\\x01C}");
158   TestDump(std::vector<int8_t>{0, -1, std::numeric_limits<int8_t>::max()},
159            lldb::Format::eFormatVectorOfSInt8, "{0 -1 127}");
160   TestDump(std::vector<uint8_t>{12, 0xFF, 34},
161            lldb::Format::eFormatVectorOfUInt8, "{0x0c 0xff 0x22}");
162   TestDump(std::vector<int16_t>{-1, 1234, std::numeric_limits<int16_t>::max()},
163            lldb::Format::eFormatVectorOfSInt16, "{-1 1234 32767}");
164   TestDump(std::vector<uint16_t>{0xffff, 0xabcd, 0x1234},
165            lldb::Format::eFormatVectorOfUInt16, "{0xffff 0xabcd 0x1234}");
166   TestDump(std::vector<int32_t>{0, -1, std::numeric_limits<int32_t>::max()},
167            lldb::Format::eFormatVectorOfSInt32, "{0 -1 2147483647}");
168   TestDump(std::vector<uint32_t>{0, 0xffffffff, 0x1234abcd},
169            lldb::Format::eFormatVectorOfUInt32,
170            "{0x00000000 0xffffffff 0x1234abcd}");
171   TestDump(std::vector<int64_t>{0, -1, std::numeric_limits<int64_t>::max()},
172            lldb::Format::eFormatVectorOfSInt64, "{0 -1 9223372036854775807}");
173   TestDump(std::vector<uint64_t>{0, 0xaaaabbbbccccdddd},
174            lldb::Format::eFormatVectorOfUInt64,
175            "{0x0000000000000000 0xaaaabbbbccccdddd}");
176 
177   // See half2float for format details.
178   // Test zeroes.
179   TestDump(std::vector<uint16_t>{0x0000, 0x8000},
180            lldb::Format::eFormatVectorOfFloat16, "{0 -0}");
181   // Some subnormal numbers.
182   TestDump(std::vector<uint16_t>{0x0001, 0x8001},
183            lldb::Format::eFormatVectorOfFloat16, "{5.96046e-08 -5.96046e-08}");
184   // A full mantisse and empty expontent.
185   TestDump(std::vector<uint16_t>{0x83ff, 0x03ff},
186            lldb::Format::eFormatVectorOfFloat16, "{-6.09756e-05 6.09756e-05}");
187   // Some normal numbers.
188   TestDump(std::vector<uint16_t>{0b0100001001001000},
189            lldb::Format::eFormatVectorOfFloat16,
190 #ifdef _WIN32
191            // FIXME: This should print the same on all platforms.
192            "{3.14063}");
193 #else
194            "{3.14062}");
195 #endif
196   // Largest and smallest normal number.
197   TestDump(std::vector<uint16_t>{0x0400, 0x7bff},
198            lldb::Format::eFormatVectorOfFloat16, "{6.10352e-05 65504}");
199   TestDump(std::vector<uint16_t>{0xabcd, 0x1234},
200            lldb::Format::eFormatVectorOfFloat16, "{-0.0609436 0.000757217}");
201 
202   // quiet/signaling NaNs.
203   TestDump(std::vector<uint16_t>{0xffff, 0xffc0, 0x7fff, 0x7fc0},
204            lldb::Format::eFormatVectorOfFloat16, "{-nan -nan nan nan}");
205   // +/-Inf.
206   TestDump(std::vector<uint16_t>{0xfc00, 0x7c00},
207            lldb::Format::eFormatVectorOfFloat16, "{-inf inf}");
208 
209   TestDump(std::vector<float>{std::numeric_limits<float>::min(),
210                               std::numeric_limits<float>::max()},
211            lldb::Format::eFormatVectorOfFloat32, "{1.17549e-38 3.40282e+38}");
212   TestDump(std::vector<float>{std::numeric_limits<float>::quiet_NaN(),
213                               std::numeric_limits<float>::signaling_NaN(),
214                               -std::numeric_limits<float>::quiet_NaN(),
215                               -std::numeric_limits<float>::signaling_NaN()},
216            lldb::Format::eFormatVectorOfFloat32, "{nan nan -nan -nan}");
217   TestDump(std::vector<double>{std::numeric_limits<double>::min(),
218                                std::numeric_limits<double>::max()},
219            lldb::Format::eFormatVectorOfFloat64,
220            "{2.2250738585072e-308 1.79769313486232e+308}");
221   TestDump(
222       std::vector<double>{
223           std::numeric_limits<double>::quiet_NaN(),
224           std::numeric_limits<double>::signaling_NaN(),
225           -std::numeric_limits<double>::quiet_NaN(),
226           -std::numeric_limits<double>::signaling_NaN(),
227       },
228       lldb::Format::eFormatVectorOfFloat64, "{nan nan -nan -nan}");
229 
230   // Not sure we can rely on having uint128_t everywhere so emulate with
231   // uint64_t.
232   TestDump(
233       std::vector<uint64_t>{0x1, 0x1111222233334444, 0xaaaabbbbccccdddd, 0x0},
234       lldb::Format::eFormatVectorOfUInt128,
235       "{0x11112222333344440000000000000001 "
236       "0x0000000000000000aaaabbbbccccdddd}");
237 
238   TestDump(std::vector<int>{2, 4}, lldb::Format::eFormatComplexInteger,
239            "2 + 4i");
240 
241   // Without an execution context this just prints the pointer on its own.
242   TestDump<uint32_t>(0x11223344, lldb::Format::eFormatAddressInfo,
243                      "0x11223344");
244 
245   // Input not written in hex form because that requires C++17.
246   TestDump<float>(10, lldb::Format::eFormatHexFloat, "0x1.4p3");
247   TestDump<double>(10, lldb::Format::eFormatHexFloat, "0x1.4p3");
248   // long double not supported, see ItemByteSizeErrors.
249 
250   // Can't disassemble without an execution context.
251   TestDump<uint32_t>(0xcafef00d, lldb::Format::eFormatInstruction,
252                      "invalid target");
253 
254   // Has no special handling, intended for use elsewhere.
255   TestDump<int>(99, lldb::Format::eFormatVoid, "0x00000063");
256 }
257 
258 TEST(DumpDataExtractorTest, FormatCharArray) {
259   // Unlike the other formats, charArray isn't 1 array of N chars.
260   // It must be passed as N chars of 1 byte each.
261   // (eFormatVectorOfChar does this swap for you)
262   std::vector<char> data{'A', '\x01', '#'};
263   StreamString result;
264   DataBufferHeap dumpbuffer(&data[0], data.size());
265   DataExtractor extractor(dumpbuffer.GetBytes(), dumpbuffer.GetByteSize(),
266                           endian::InlHostByteOrder(), /*addr_size=*/4);
267 
268   DumpDataExtractor(extractor, &result, 0, lldb::Format::eFormatCharArray,
269                     /*item_byte_size=*/1,
270                     /*item_count=*/data.size(),
271                     /*num_per_line=*/data.size(), 0, 0, 0);
272   ASSERT_EQ("0x00000000: A\\x01#", result.GetString());
273 
274   result.Clear();
275   DumpDataExtractor(extractor, &result, 0, lldb::Format::eFormatCharArray, 1,
276                     data.size(), 1, 0, 0, 0);
277   // ASSERT macro thinks the split strings are multiple arguments so make a var.
278   const char *expected = "0x00000000: A\n"
279                          "0x00000001: \\x01\n"
280                          "0x00000002: #";
281   ASSERT_EQ(expected, result.GetString());
282 }
283 
284 template <typename T>
285 void TestDumpMultiLine(std::vector<T> data, lldb::Format format,
286                        size_t num_per_line, llvm::StringRef expected) {
287   size_t sz_bytes = data.size() * sizeof(T);
288   TestDumpImpl(&data[0], sz_bytes, data.size(), sz_bytes, num_per_line,
289                0x80000000, format, expected);
290 }
291 
292 template <typename T>
293 void TestDumpMultiLine(const T *data, size_t num_items, lldb::Format format,
294                        size_t num_per_line, llvm::StringRef expected) {
295   TestDumpImpl(data, sizeof(T) * num_items, sizeof(T), num_items, num_per_line,
296                0x80000000, format, expected);
297 }
298 
299 TEST(DumpDataExtractorTest, MultiLine) {
300   // A vector counts as 1 item regardless of size.
301   TestDumpMultiLine(std::vector<uint8_t>{0x11},
302                     lldb::Format::eFormatVectorOfUInt8, 1,
303                     "0x80000000: {0x11}");
304   TestDumpMultiLine(std::vector<uint8_t>{0x11, 0x22},
305                     lldb::Format::eFormatVectorOfUInt8, 1,
306                     "0x80000000: {0x11 0x22}");
307 
308   // If you have multiple vectors then that's multiple items.
309   // Here we say that these 2 bytes are actually 2 1 byte vectors.
310   const std::vector<uint8_t> vector_data{0x11, 0x22};
311   TestDumpMultiLine(vector_data.data(), 2, lldb::Format::eFormatVectorOfUInt8,
312                     1, "0x80000000: {0x11}\n0x80000001: {0x22}");
313 
314   // Single value formats can span multiple lines.
315   const std::vector<uint8_t> bytes{0x11, 0x22, 0x33};
316   const char *expected_bytes_3_line = "0x80000000: 0x11\n"
317                                       "0x80000001: 0x22\n"
318                                       "0x80000002: 0x33";
319   TestDumpMultiLine(bytes.data(), bytes.size(), lldb::Format::eFormatHex, 1,
320                     expected_bytes_3_line);
321 
322   // Lines may not have the full number of items.
323   TestDumpMultiLine(bytes.data(), bytes.size(), lldb::Format::eFormatHex, 4,
324                     "0x80000000: 0x11 0x22 0x33");
325   const char *expected_bytes_2_line = "0x80000000: 0x11 0x22\n"
326                                       "0x80000002: 0x33";
327   TestDumpMultiLine(bytes.data(), bytes.size(), lldb::Format::eFormatHex, 2,
328                     expected_bytes_2_line);
329 
330   // The line address accounts for item sizes other than 1 byte.
331   const std::vector<uint16_t> shorts{0x1111, 0x2222, 0x3333};
332   const char *expected_shorts_2_line = "0x80000000: 0x1111 0x2222\n"
333                                        "0x80000004: 0x3333";
334   TestDumpMultiLine(shorts.data(), shorts.size(), lldb::Format::eFormatHex, 2,
335                     expected_shorts_2_line);
336 
337   // The ascii column is positioned using the maximum line length.
338   const std::vector<char> chars{'L', 'L', 'D', 'B'};
339   const char *expected_chars_2_lines = "0x80000000: 4c 4c 44  LLD\n"
340                                        "0x80000003: 42        B";
341   TestDumpMultiLine(chars.data(), chars.size(),
342                     lldb::Format::eFormatBytesWithASCII, 3,
343                     expected_chars_2_lines);
344 }
345 
346 void TestDumpWithItemByteSize(size_t item_byte_size, lldb::Format format,
347                               llvm::StringRef expected) {
348   // We won't be reading this data so anything will do.
349   uint8_t dummy = 0;
350   TestDumpImpl(&dummy, 1, item_byte_size, 1, 1, LLDB_INVALID_ADDRESS, format,
351                expected);
352 }
353 
354 TEST(DumpDataExtractorTest, ItemByteSizeErrors) {
355   TestDumpWithItemByteSize(
356       16, lldb::Format::eFormatBoolean,
357       "error: unsupported byte size (16) for boolean format");
358   TestDumpWithItemByteSize(21, lldb::Format::eFormatChar,
359                            "error: unsupported byte size (21) for char format");
360   TestDumpWithItemByteSize(
361       18, lldb::Format::eFormatComplexInteger,
362       "error: unsupported byte size (18) for complex integer format");
363 
364   // The code uses sizeof(long double) for these checks. This changes by host
365   // but we know it won't be >16.
366   TestDumpWithItemByteSize(
367       34, lldb::Format::eFormatComplex,
368       "error: unsupported byte size (34) for complex float format");
369   TestDumpWithItemByteSize(
370       18, lldb::Format::eFormatFloat,
371       "error: unsupported byte size (18) for float format");
372 
373   // We want sizes to exactly match one of float/double.
374   TestDumpWithItemByteSize(
375       14, lldb::Format::eFormatComplex,
376       "error: unsupported byte size (14) for complex float format");
377   TestDumpWithItemByteSize(3, lldb::Format::eFormatFloat,
378                            "error: unsupported byte size (3) for float format");
379 
380   // We only allow float and double size.
381   TestDumpWithItemByteSize(
382       1, lldb::Format::eFormatHexFloat,
383       "error: unsupported byte size (1) for hex float format");
384   TestDumpWithItemByteSize(
385       17, lldb::Format::eFormatHexFloat,
386       "error: unsupported byte size (17) for hex float format");
387 }
388