1 //===- llvm/unittest/Support/DataExtractorTest.cpp - DataExtractor tests --===//
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 "llvm/Support/DataExtractor.h"
10 #include "llvm/Testing/Support/Error.h"
11 #include "gtest/gtest.h"
12 using namespace llvm;
13 
14 namespace {
15 
16 const char numberData[] = "\x80\x90\xFF\xFF\x80\x00\x00\x00";
17 const char leb128data[] = "\xA6\x49";
18 const char bigleb128data[] = "\xAA\xA9\xFF\xAA\xFF\xAA\xFF\x4A";
19 
TEST(DataExtractorTest,OffsetOverflow)20 TEST(DataExtractorTest, OffsetOverflow) {
21   DataExtractor DE(StringRef(numberData, sizeof(numberData)-1), false, 8);
22   EXPECT_FALSE(DE.isValidOffsetForDataOfSize(-2U, 5));
23 }
24 
TEST(DataExtractorTest,UnsignedNumbers)25 TEST(DataExtractorTest, UnsignedNumbers) {
26   DataExtractor DE(StringRef(numberData, sizeof(numberData)-1), false, 8);
27   uint64_t offset = 0;
28 
29   EXPECT_EQ(0x80U, DE.getU8(&offset));
30   EXPECT_EQ(1U, offset);
31   offset = 0;
32   EXPECT_EQ(0x8090U, DE.getU16(&offset));
33   EXPECT_EQ(2U, offset);
34   offset = 0;
35   EXPECT_EQ(0x8090FFFFU, DE.getU32(&offset));
36   EXPECT_EQ(4U, offset);
37   offset = 0;
38   EXPECT_EQ(0x8090FFFF80000000ULL, DE.getU64(&offset));
39   EXPECT_EQ(8U, offset);
40   offset = 0;
41   EXPECT_EQ(0x8090FFFF80000000ULL, DE.getAddress(&offset));
42   EXPECT_EQ(8U, offset);
43   offset = 0;
44 
45   uint32_t data[2];
46   EXPECT_EQ(data, DE.getU32(&offset, data, 2));
47   EXPECT_EQ(0x8090FFFFU, data[0]);
48   EXPECT_EQ(0x80000000U, data[1]);
49   EXPECT_EQ(8U, offset);
50   offset = 0;
51 
52   // Now for little endian.
53   DE = DataExtractor(StringRef(numberData, sizeof(numberData)-1), true, 4);
54   EXPECT_EQ(0x9080U, DE.getU16(&offset));
55   EXPECT_EQ(2U, offset);
56   offset = 0;
57   EXPECT_EQ(0xFFFF9080U, DE.getU32(&offset));
58   EXPECT_EQ(4U, offset);
59   offset = 0;
60   EXPECT_EQ(0x80FFFF9080ULL, DE.getU64(&offset));
61   EXPECT_EQ(8U, offset);
62   offset = 0;
63   EXPECT_EQ(0xFFFF9080U, DE.getAddress(&offset));
64   EXPECT_EQ(4U, offset);
65   offset = 0;
66 
67   EXPECT_EQ(data, DE.getU32(&offset, data, 2));
68   EXPECT_EQ(0xFFFF9080U, data[0]);
69   EXPECT_EQ(0x80U, data[1]);
70   EXPECT_EQ(8U, offset);
71 }
72 
TEST(DataExtractorTest,SignedNumbers)73 TEST(DataExtractorTest, SignedNumbers) {
74   DataExtractor DE(StringRef(numberData, sizeof(numberData)-1), false, 8);
75   uint64_t offset = 0;
76 
77   EXPECT_EQ(-128, DE.getSigned(&offset, 1));
78   EXPECT_EQ(1U, offset);
79   offset = 0;
80   EXPECT_EQ(-32624, DE.getSigned(&offset, 2));
81   EXPECT_EQ(2U, offset);
82   offset = 0;
83   EXPECT_EQ(-2137980929, DE.getSigned(&offset, 4));
84   EXPECT_EQ(4U, offset);
85   offset = 0;
86   EXPECT_EQ(-9182558167379214336LL, DE.getSigned(&offset, 8));
87   EXPECT_EQ(8U, offset);
88 }
89 
TEST(DataExtractorTest,Strings)90 TEST(DataExtractorTest, Strings) {
91   const char stringData[] = "hellohello\0hello";
92   DataExtractor DE(StringRef(stringData, sizeof(stringData)-1), false, 8);
93   uint64_t offset = 0;
94 
95   EXPECT_EQ(stringData, DE.getCStr(&offset));
96   EXPECT_EQ(11U, offset);
97   EXPECT_EQ(nullptr, DE.getCStr(&offset));
98   EXPECT_EQ(11U, offset);
99 
100   DataExtractor::Cursor C(0);
101   EXPECT_EQ(stringData, DE.getCStr(C));
102   EXPECT_EQ(11U, C.tell());
103   EXPECT_EQ(nullptr, DE.getCStr(C));
104   EXPECT_EQ(11U, C.tell());
105   EXPECT_THAT_ERROR(
106       C.takeError(),
107       FailedWithMessage("no null terminated string at offset 0xb"));
108 }
109 
TEST(DataExtractorTest,LEB128)110 TEST(DataExtractorTest, LEB128) {
111   DataExtractor DE(StringRef(leb128data, sizeof(leb128data)-1), false, 8);
112   uint64_t offset = 0;
113 
114   EXPECT_EQ(9382ULL, DE.getULEB128(&offset));
115   EXPECT_EQ(2U, offset);
116   offset = 0;
117   EXPECT_EQ(-7002LL, DE.getSLEB128(&offset));
118   EXPECT_EQ(2U, offset);
119 
120   DataExtractor BDE(StringRef(bigleb128data, sizeof(bigleb128data)-1), false,8);
121   offset = 0;
122   EXPECT_EQ(42218325750568106ULL, BDE.getULEB128(&offset));
123   EXPECT_EQ(8U, offset);
124   offset = 0;
125   EXPECT_EQ(-29839268287359830LL, BDE.getSLEB128(&offset));
126   EXPECT_EQ(8U, offset);
127 }
128 
TEST(DataExtractorTest,LEB128_error)129 TEST(DataExtractorTest, LEB128_error) {
130   DataExtractor DE(StringRef("\x81"), false, 8);
131   uint64_t Offset = 0;
132   EXPECT_EQ(0U, DE.getULEB128(&Offset));
133   EXPECT_EQ(0U, Offset);
134 
135   Offset = 0;
136   EXPECT_EQ(0U, DE.getSLEB128(&Offset));
137   EXPECT_EQ(0U, Offset);
138 
139   DataExtractor::Cursor C(0);
140   EXPECT_EQ(0U, DE.getULEB128(C));
141   EXPECT_THAT_ERROR(
142       C.takeError(),
143       FailedWithMessage("unable to decode LEB128 at offset 0x00000000: "
144                         "malformed uleb128, extends past end"));
145 
146   C = DataExtractor::Cursor(0);
147   EXPECT_EQ(0U, DE.getSLEB128(C));
148   EXPECT_THAT_ERROR(
149       C.takeError(),
150       FailedWithMessage("unable to decode LEB128 at offset 0x00000000: "
151                         "malformed sleb128, extends past end"));
152 
153   // Show non-zero offsets are reported appropriately.
154   C = DataExtractor::Cursor(1);
155   EXPECT_EQ(0U, DE.getULEB128(C));
156   EXPECT_THAT_ERROR(
157       C.takeError(),
158       FailedWithMessage("unable to decode LEB128 at offset 0x00000001: "
159                         "malformed uleb128, extends past end"));
160 }
161 
TEST(DataExtractorTest,Cursor_tell)162 TEST(DataExtractorTest, Cursor_tell) {
163   DataExtractor DE(StringRef("AB"), false, 8);
164   DataExtractor::Cursor C(0);
165   // A successful read operation advances the cursor
166   EXPECT_EQ('A', DE.getU8(C));
167   EXPECT_EQ(1u, C.tell());
168 
169   // An unsuccessful one doesn't.
170   EXPECT_EQ(0u, DE.getU16(C));
171   EXPECT_EQ(1u, C.tell());
172 
173   // And neither do any subsequent operations.
174   EXPECT_EQ(0, DE.getU8(C));
175   EXPECT_EQ(1u, C.tell());
176 
177   consumeError(C.takeError());
178 }
179 
TEST(DataExtractorTest,Cursor_takeError)180 TEST(DataExtractorTest, Cursor_takeError) {
181   DataExtractor DE(StringRef("AB"), false, 8);
182   DataExtractor::Cursor C(0);
183   // Initially, the cursor is in the "success" state.
184   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
185 
186   // It remains "success" after a successful read.
187   EXPECT_EQ('A', DE.getU8(C));
188   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
189 
190   // An unsuccessful read sets the error state.
191   EXPECT_EQ(0u, DE.getU32(C));
192   EXPECT_THAT_ERROR(C.takeError(), Failed());
193 
194   // Once set the error sticks until explicitly cleared.
195   EXPECT_EQ(0u, DE.getU32(C));
196   EXPECT_EQ(0, DE.getU8(C));
197   EXPECT_THAT_ERROR(C.takeError(), Failed());
198 
199   // At which point reads can be succeed again.
200   EXPECT_EQ('B', DE.getU8(C));
201   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
202 }
203 
TEST(DataExtractorTest,Cursor_chaining)204 TEST(DataExtractorTest, Cursor_chaining) {
205   DataExtractor DE(StringRef("ABCD"), false, 8);
206   DataExtractor::Cursor C(0);
207 
208   // Multiple reads can be chained without trigerring any assertions.
209   EXPECT_EQ('A', DE.getU8(C));
210   EXPECT_EQ('B', DE.getU8(C));
211   EXPECT_EQ('C', DE.getU8(C));
212   EXPECT_EQ('D', DE.getU8(C));
213   // And the error checked at the end.
214   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
215 }
216 
217 #if defined(GTEST_HAS_DEATH_TEST) && defined(_DEBUG) &&                        \
218     LLVM_ENABLE_ABI_BREAKING_CHECKS
TEST(DataExtractorDeathTest,Cursor)219 TEST(DataExtractorDeathTest, Cursor) {
220   DataExtractor DE(StringRef("AB"), false, 8);
221 
222   // Even an unused cursor must be checked for errors:
223   EXPECT_DEATH(DataExtractor::Cursor(0),
224                "Success values must still be checked prior to being destroyed");
225 
226   {
227     auto C = std::make_unique<DataExtractor::Cursor>(0);
228     EXPECT_EQ(0u, DE.getU32(*C));
229     // It must also be checked after an unsuccessful operation.
230     // destruction.
231     EXPECT_DEATH(C.reset(), "unexpected end of data");
232     EXPECT_THAT_ERROR(C->takeError(), Failed());
233   }
234   {
235     auto C = std::make_unique<DataExtractor::Cursor>(0);
236     EXPECT_EQ('A', DE.getU8(*C));
237     // Same goes for a successful one.
238     EXPECT_DEATH(
239         C.reset(),
240         "Success values must still be checked prior to being destroyed");
241     EXPECT_THAT_ERROR(C->takeError(), Succeeded());
242   }
243   {
244     auto C = std::make_unique<DataExtractor::Cursor>(0);
245     EXPECT_EQ('A', DE.getU8(*C));
246     EXPECT_EQ(0u, DE.getU32(*C));
247     // Even if a successful operation is followed by an unsuccessful one.
248     EXPECT_DEATH(C.reset(), "unexpected end of data");
249     EXPECT_THAT_ERROR(C->takeError(), Failed());
250   }
251   {
252     auto C = std::make_unique<DataExtractor::Cursor>(0);
253     EXPECT_EQ(0u, DE.getU32(*C));
254     EXPECT_EQ(0, DE.getU8(*C));
255     // Even if an unsuccessful operation is followed by one that would normally
256     // succeed.
257     EXPECT_DEATH(C.reset(), "unexpected end of data");
258     EXPECT_THAT_ERROR(C->takeError(), Failed());
259   }
260 }
261 #endif
262 
TEST(DataExtractorTest,getU8_vector)263 TEST(DataExtractorTest, getU8_vector) {
264   DataExtractor DE(StringRef("AB"), false, 8);
265   DataExtractor::Cursor C(0);
266   SmallVector<uint8_t, 2> S;
267 
268   DE.getU8(C, S, 4);
269   EXPECT_THAT_ERROR(C.takeError(), Failed());
270   EXPECT_EQ("", toStringRef(S));
271 
272   DE.getU8(C, S, 2);
273   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
274   EXPECT_EQ("AB", toStringRef(S));
275 
276   C = DataExtractor::Cursor(0x47);
277   DE.getU8(C, S, 2);
278   EXPECT_THAT_ERROR(
279       C.takeError(),
280       FailedWithMessage("offset 0x47 is beyond the end of data at 0x2"));
281 }
282 
TEST(DataExtractorTest,getU24)283 TEST(DataExtractorTest, getU24) {
284   DataExtractor DE(StringRef("ABCD"), false, 8);
285   DataExtractor::Cursor C(0);
286 
287   EXPECT_EQ(0x414243u, DE.getU24(C));
288   EXPECT_EQ(0u, DE.getU24(C));
289   EXPECT_EQ(3u, C.tell());
290   EXPECT_THAT_ERROR(C.takeError(), Failed());
291 }
292 
TEST(DataExtractorTest,skip)293 TEST(DataExtractorTest, skip) {
294   DataExtractor DE(StringRef("AB"), false, 8);
295   DataExtractor::Cursor C(0);
296 
297   DE.skip(C, 4);
298   EXPECT_THAT_ERROR(C.takeError(), Failed());
299   EXPECT_EQ(0u, C.tell());
300 
301   DE.skip(C, 2);
302   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
303   EXPECT_EQ(2u, C.tell());
304 }
305 
TEST(DataExtractorTest,eof)306 TEST(DataExtractorTest, eof) {
307   DataExtractor DE(StringRef("A"), false, 8);
308   DataExtractor::Cursor C(0);
309 
310   EXPECT_FALSE(DE.eof(C));
311 
312   EXPECT_EQ(0, DE.getU16(C));
313   EXPECT_FALSE(DE.eof(C));
314   EXPECT_THAT_ERROR(C.takeError(), Failed());
315 
316   EXPECT_EQ('A', DE.getU8(C));
317   EXPECT_TRUE(DE.eof(C));
318   EXPECT_THAT_ERROR(C.takeError(), Succeeded());
319 }
320 
TEST(DataExtractorTest,size)321 TEST(DataExtractorTest, size) {
322   uint8_t Data[] = {'A', 'B', 'C', 'D'};
323   DataExtractor DE1(StringRef(reinterpret_cast<char *>(Data), sizeof(Data)),
324                     false, 8);
325   EXPECT_EQ(DE1.size(), sizeof(Data));
326   DataExtractor DE2(ArrayRef<uint8_t>(Data), false, 8);
327   EXPECT_EQ(DE2.size(), sizeof(Data));
328 }
329 
TEST(DataExtractorTest,FixedLengthString)330 TEST(DataExtractorTest, FixedLengthString) {
331   const char Data[] = "hello\x00\x00\x00world  \thola\x00";
332   DataExtractor DE(StringRef(Data, sizeof(Data)-1), false, 8);
333   uint64_t Offset = 0;
334   StringRef Str;
335   // Test extracting too many bytes doesn't modify Offset and returns None.
336   Str = DE.getFixedLengthString(&Offset, sizeof(Data));
337   EXPECT_TRUE(Str.empty());
338   EXPECT_EQ(Offset, 0u);
339 
340   // Test extracting a fixed width C string with trailing NULL characters.
341   Str = DE.getFixedLengthString(&Offset, 8);
342   EXPECT_EQ(Offset, 8u);
343   EXPECT_EQ(Str.size(), 5u);
344   EXPECT_EQ(Str, "hello");
345   // Test extracting a fixed width C string with trailing space and tab
346   // characters.
347   Str = DE.getFixedLengthString(&Offset, 8, " \t");
348   EXPECT_EQ(Offset, 16u);
349   EXPECT_EQ(Str.size(), 5u);
350   EXPECT_EQ(Str, "world");
351   // Now extract a normal C string.
352   Str = DE.getCStrRef(&Offset);
353   EXPECT_EQ(Str.size(), 4u);
354   EXPECT_EQ(Str, "hola");
355 }
356 
357 
TEST(DataExtractorTest,GetBytes)358 TEST(DataExtractorTest, GetBytes) {
359   // Use data with an embedded NULL character for good measure.
360   const char Data[] = "\x01\x02\x00\x04";
361   StringRef Bytes(Data, sizeof(Data)-1);
362   DataExtractor DE(Bytes, false, 8);
363   uint64_t Offset = 0;
364   StringRef Str;
365   // Test extracting too many bytes doesn't modify Offset and returns None.
366   Str = DE.getBytes(&Offset, sizeof(Data));
367   EXPECT_TRUE(Str.empty());
368   EXPECT_EQ(Offset, 0u);
369   // Test extracting 4 bytes from the stream.
370   Str = DE.getBytes(&Offset, 4);
371   EXPECT_EQ(Offset, 4u);
372   EXPECT_EQ(Str.size(), 4u);
373   EXPECT_EQ(Str, Bytes);
374 
375   DataExtractor::Cursor C(0);
376   EXPECT_EQ(StringRef("\x01\x02"), DE.getBytes(C, 2));
377   EXPECT_EQ(StringRef("\x00\x04", 2), DE.getBytes(C, 2));
378   EXPECT_EQ(StringRef(), DE.getBytes(C, 2));
379   EXPECT_EQ(StringRef(), DE.getBytes(C, 2));
380   EXPECT_EQ(4u, C.tell());
381   EXPECT_THAT_ERROR(C.takeError(), Failed());
382 }
383 
384 }
385