• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 //===-- DataExtractor.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/Utility/DataExtractor.h"
10 
11 #include "lldb/lldb-defines.h"
12 #include "lldb/lldb-enumerations.h"
13 #include "lldb/lldb-forward.h"
14 #include "lldb/lldb-types.h"
15 
16 #include "lldb/Utility/DataBuffer.h"
17 #include "lldb/Utility/DataBufferHeap.h"
18 #include "lldb/Utility/LLDBAssert.h"
19 #include "lldb/Utility/Log.h"
20 #include "lldb/Utility/Stream.h"
21 #include "lldb/Utility/StreamString.h"
22 #include "lldb/Utility/UUID.h"
23 
24 #include "llvm/ADT/ArrayRef.h"
25 #include "llvm/ADT/SmallVector.h"
26 #include "llvm/Support/LEB128.h"
27 #include "llvm/Support/MD5.h"
28 #include "llvm/Support/MathExtras.h"
29 
30 #include <algorithm>
31 #include <array>
32 #include <cassert>
33 #include <cstdint>
34 #include <string>
35 
36 #include <ctype.h>
37 #include <inttypes.h>
38 #include <string.h>
39 
40 using namespace lldb;
41 using namespace lldb_private;
42 
ReadInt16(const unsigned char * ptr,offset_t offset)43 static inline uint16_t ReadInt16(const unsigned char *ptr, offset_t offset) {
44   uint16_t value;
45   memcpy(&value, ptr + offset, 2);
46   return value;
47 }
48 
ReadInt32(const unsigned char * ptr,offset_t offset=0)49 static inline uint32_t ReadInt32(const unsigned char *ptr,
50                                  offset_t offset = 0) {
51   uint32_t value;
52   memcpy(&value, ptr + offset, 4);
53   return value;
54 }
55 
ReadInt64(const unsigned char * ptr,offset_t offset=0)56 static inline uint64_t ReadInt64(const unsigned char *ptr,
57                                  offset_t offset = 0) {
58   uint64_t value;
59   memcpy(&value, ptr + offset, 8);
60   return value;
61 }
62 
ReadInt16(const void * ptr)63 static inline uint16_t ReadInt16(const void *ptr) {
64   uint16_t value;
65   memcpy(&value, ptr, 2);
66   return value;
67 }
68 
ReadSwapInt16(const unsigned char * ptr,offset_t offset)69 static inline uint16_t ReadSwapInt16(const unsigned char *ptr,
70                                      offset_t offset) {
71   uint16_t value;
72   memcpy(&value, ptr + offset, 2);
73   return llvm::ByteSwap_16(value);
74 }
75 
ReadSwapInt32(const unsigned char * ptr,offset_t offset)76 static inline uint32_t ReadSwapInt32(const unsigned char *ptr,
77                                      offset_t offset) {
78   uint32_t value;
79   memcpy(&value, ptr + offset, 4);
80   return llvm::ByteSwap_32(value);
81 }
82 
ReadSwapInt64(const unsigned char * ptr,offset_t offset)83 static inline uint64_t ReadSwapInt64(const unsigned char *ptr,
84                                      offset_t offset) {
85   uint64_t value;
86   memcpy(&value, ptr + offset, 8);
87   return llvm::ByteSwap_64(value);
88 }
89 
ReadSwapInt16(const void * ptr)90 static inline uint16_t ReadSwapInt16(const void *ptr) {
91   uint16_t value;
92   memcpy(&value, ptr, 2);
93   return llvm::ByteSwap_16(value);
94 }
95 
ReadSwapInt32(const void * ptr)96 static inline uint32_t ReadSwapInt32(const void *ptr) {
97   uint32_t value;
98   memcpy(&value, ptr, 4);
99   return llvm::ByteSwap_32(value);
100 }
101 
ReadSwapInt64(const void * ptr)102 static inline uint64_t ReadSwapInt64(const void *ptr) {
103   uint64_t value;
104   memcpy(&value, ptr, 8);
105   return llvm::ByteSwap_64(value);
106 }
107 
ReadMaxInt64(const uint8_t * data,size_t byte_size,ByteOrder byte_order)108 static inline uint64_t ReadMaxInt64(const uint8_t *data, size_t byte_size,
109                                     ByteOrder byte_order) {
110   uint64_t res = 0;
111   if (byte_order == eByteOrderBig)
112     for (size_t i = 0; i < byte_size; ++i)
113       res = (res << 8) | data[i];
114   else {
115     assert(byte_order == eByteOrderLittle);
116     for (size_t i = 0; i < byte_size; ++i)
117       res = (res << 8) | data[byte_size - 1 - i];
118   }
119   return res;
120 }
121 
DataExtractor()122 DataExtractor::DataExtractor()
123     : m_start(nullptr), m_end(nullptr),
124       m_byte_order(endian::InlHostByteOrder()), m_addr_size(sizeof(void *)),
125       m_data_sp(), m_target_byte_size(1) {}
126 
127 // This constructor allows us to use data that is owned by someone else. The
128 // data must stay around as long as this object is valid.
DataExtractor(const void * data,offset_t length,ByteOrder endian,uint32_t addr_size,uint32_t target_byte_size)129 DataExtractor::DataExtractor(const void *data, offset_t length,
130                              ByteOrder endian, uint32_t addr_size,
131                              uint32_t target_byte_size /*=1*/)
132     : m_start(const_cast<uint8_t *>(static_cast<const uint8_t *>(data))),
133       m_end(const_cast<uint8_t *>(static_cast<const uint8_t *>(data)) + length),
134       m_byte_order(endian), m_addr_size(addr_size), m_data_sp(),
135       m_target_byte_size(target_byte_size) {
136   assert(addr_size >= 1 && addr_size <= 8);
137 }
138 
139 // Make a shared pointer reference to the shared data in "data_sp" and set the
140 // endian swapping setting to "swap", and the address size to "addr_size". The
141 // shared data reference will ensure the data lives as long as any
142 // DataExtractor objects exist that have a reference to this data.
DataExtractor(const DataBufferSP & data_sp,ByteOrder endian,uint32_t addr_size,uint32_t target_byte_size)143 DataExtractor::DataExtractor(const DataBufferSP &data_sp, ByteOrder endian,
144                              uint32_t addr_size,
145                              uint32_t target_byte_size /*=1*/)
146     : m_start(nullptr), m_end(nullptr), m_byte_order(endian),
147       m_addr_size(addr_size), m_data_sp(),
148       m_target_byte_size(target_byte_size) {
149   assert(addr_size >= 1 && addr_size <= 8);
150   SetData(data_sp);
151 }
152 
153 // Initialize this object with a subset of the data bytes in "data". If "data"
154 // contains shared data, then a reference to this shared data will added and
155 // the shared data will stay around as long as any object contains a reference
156 // to that data. The endian swap and address size settings are copied from
157 // "data".
DataExtractor(const DataExtractor & data,offset_t offset,offset_t length,uint32_t target_byte_size)158 DataExtractor::DataExtractor(const DataExtractor &data, offset_t offset,
159                              offset_t length, uint32_t target_byte_size /*=1*/)
160     : m_start(nullptr), m_end(nullptr), m_byte_order(data.m_byte_order),
161       m_addr_size(data.m_addr_size), m_data_sp(),
162       m_target_byte_size(target_byte_size) {
163   assert(m_addr_size >= 1 && m_addr_size <= 8);
164   if (data.ValidOffset(offset)) {
165     offset_t bytes_available = data.GetByteSize() - offset;
166     if (length > bytes_available)
167       length = bytes_available;
168     SetData(data, offset, length);
169   }
170 }
171 
DataExtractor(const DataExtractor & rhs)172 DataExtractor::DataExtractor(const DataExtractor &rhs)
173     : m_start(rhs.m_start), m_end(rhs.m_end), m_byte_order(rhs.m_byte_order),
174       m_addr_size(rhs.m_addr_size), m_data_sp(rhs.m_data_sp),
175       m_target_byte_size(rhs.m_target_byte_size) {
176   assert(m_addr_size >= 1 && m_addr_size <= 8);
177 }
178 
179 // Assignment operator
operator =(const DataExtractor & rhs)180 const DataExtractor &DataExtractor::operator=(const DataExtractor &rhs) {
181   if (this != &rhs) {
182     m_start = rhs.m_start;
183     m_end = rhs.m_end;
184     m_byte_order = rhs.m_byte_order;
185     m_addr_size = rhs.m_addr_size;
186     m_data_sp = rhs.m_data_sp;
187   }
188   return *this;
189 }
190 
191 DataExtractor::~DataExtractor() = default;
192 
193 // Clears the object contents back to a default invalid state, and release any
194 // references to shared data that this object may contain.
Clear()195 void DataExtractor::Clear() {
196   m_start = nullptr;
197   m_end = nullptr;
198   m_byte_order = endian::InlHostByteOrder();
199   m_addr_size = sizeof(void *);
200   m_data_sp.reset();
201 }
202 
203 // If this object contains shared data, this function returns the offset into
204 // that shared data. Else zero is returned.
GetSharedDataOffset() const205 size_t DataExtractor::GetSharedDataOffset() const {
206   if (m_start != nullptr) {
207     const DataBuffer *data = m_data_sp.get();
208     if (data != nullptr) {
209       const uint8_t *data_bytes = data->GetBytes();
210       if (data_bytes != nullptr) {
211         assert(m_start >= data_bytes);
212         return m_start - data_bytes;
213       }
214     }
215   }
216   return 0;
217 }
218 
219 // Set the data with which this object will extract from to data starting at
220 // BYTES and set the length of the data to LENGTH bytes long. The data is
221 // externally owned must be around at least as long as this object points to
222 // the data. No copy of the data is made, this object just refers to this data
223 // and can extract from it. If this object refers to any shared data upon
224 // entry, the reference to that data will be released. Is SWAP is set to true,
225 // any data extracted will be endian swapped.
SetData(const void * bytes,offset_t length,ByteOrder endian)226 lldb::offset_t DataExtractor::SetData(const void *bytes, offset_t length,
227                                       ByteOrder endian) {
228   m_byte_order = endian;
229   m_data_sp.reset();
230   if (bytes == nullptr || length == 0) {
231     m_start = nullptr;
232     m_end = nullptr;
233   } else {
234     m_start = const_cast<uint8_t *>(static_cast<const uint8_t *>(bytes));
235     m_end = m_start + length;
236   }
237   return GetByteSize();
238 }
239 
240 // Assign the data for this object to be a subrange in "data" starting
241 // "data_offset" bytes into "data" and ending "data_length" bytes later. If
242 // "data_offset" is not a valid offset into "data", then this object will
243 // contain no bytes. If "data_offset" is within "data" yet "data_length" is too
244 // large, the length will be capped at the number of bytes remaining in "data".
245 // If "data" contains a shared pointer to other data, then a ref counted
246 // pointer to that data will be made in this object. If "data" doesn't contain
247 // a shared pointer to data, then the bytes referred to in "data" will need to
248 // exist at least as long as this object refers to those bytes. The address
249 // size and endian swap settings are copied from the current values in "data".
SetData(const DataExtractor & data,offset_t data_offset,offset_t data_length)250 lldb::offset_t DataExtractor::SetData(const DataExtractor &data,
251                                       offset_t data_offset,
252                                       offset_t data_length) {
253   m_addr_size = data.m_addr_size;
254   assert(m_addr_size >= 1 && m_addr_size <= 8);
255   // If "data" contains shared pointer to data, then we can use that
256   if (data.m_data_sp) {
257     m_byte_order = data.m_byte_order;
258     return SetData(data.m_data_sp, data.GetSharedDataOffset() + data_offset,
259                    data_length);
260   }
261 
262   // We have a DataExtractor object that just has a pointer to bytes
263   if (data.ValidOffset(data_offset)) {
264     if (data_length > data.GetByteSize() - data_offset)
265       data_length = data.GetByteSize() - data_offset;
266     return SetData(data.GetDataStart() + data_offset, data_length,
267                    data.GetByteOrder());
268   }
269   return 0;
270 }
271 
272 // Assign the data for this object to be a subrange of the shared data in
273 // "data_sp" starting "data_offset" bytes into "data_sp" and ending
274 // "data_length" bytes later. If "data_offset" is not a valid offset into
275 // "data_sp", then this object will contain no bytes. If "data_offset" is
276 // within "data_sp" yet "data_length" is too large, the length will be capped
277 // at the number of bytes remaining in "data_sp". A ref counted pointer to the
278 // data in "data_sp" will be made in this object IF the number of bytes this
279 // object refers to in greater than zero (if at least one byte was available
280 // starting at "data_offset") to ensure the data stays around as long as it is
281 // needed. The address size and endian swap settings will remain unchanged from
282 // their current settings.
SetData(const DataBufferSP & data_sp,offset_t data_offset,offset_t data_length)283 lldb::offset_t DataExtractor::SetData(const DataBufferSP &data_sp,
284                                       offset_t data_offset,
285                                       offset_t data_length) {
286   m_start = m_end = nullptr;
287 
288   if (data_length > 0) {
289     m_data_sp = data_sp;
290     if (data_sp) {
291       const size_t data_size = data_sp->GetByteSize();
292       if (data_offset < data_size) {
293         m_start = data_sp->GetBytes() + data_offset;
294         const size_t bytes_left = data_size - data_offset;
295         // Cap the length of we asked for too many
296         if (data_length <= bytes_left)
297           m_end = m_start + data_length; // We got all the bytes we wanted
298         else
299           m_end = m_start + bytes_left; // Not all the bytes requested were
300                                         // available in the shared data
301       }
302     }
303   }
304 
305   size_t new_size = GetByteSize();
306 
307   // Don't hold a shared pointer to the data buffer if we don't share any valid
308   // bytes in the shared buffer.
309   if (new_size == 0)
310     m_data_sp.reset();
311 
312   return new_size;
313 }
314 
315 // Extract a single unsigned char from the binary data and update the offset
316 // pointed to by "offset_ptr".
317 //
318 // RETURNS the byte that was extracted, or zero on failure.
GetU8(offset_t * offset_ptr) const319 uint8_t DataExtractor::GetU8(offset_t *offset_ptr) const {
320   const uint8_t *data = static_cast<const uint8_t *>(GetData(offset_ptr, 1));
321   if (data)
322     return *data;
323   return 0;
324 }
325 
326 // Extract "count" unsigned chars from the binary data and update the offset
327 // pointed to by "offset_ptr". The extracted data is copied into "dst".
328 //
329 // RETURNS the non-nullptr buffer pointer upon successful extraction of
330 // all the requested bytes, or nullptr when the data is not available in the
331 // buffer due to being out of bounds, or insufficient data.
GetU8(offset_t * offset_ptr,void * dst,uint32_t count) const332 void *DataExtractor::GetU8(offset_t *offset_ptr, void *dst,
333                            uint32_t count) const {
334   const uint8_t *data =
335       static_cast<const uint8_t *>(GetData(offset_ptr, count));
336   if (data) {
337     // Copy the data into the buffer
338     memcpy(dst, data, count);
339     // Return a non-nullptr pointer to the converted data as an indicator of
340     // success
341     return dst;
342   }
343   return nullptr;
344 }
345 
346 // Extract a single uint16_t from the data and update the offset pointed to by
347 // "offset_ptr".
348 //
349 // RETURNS the uint16_t that was extracted, or zero on failure.
GetU16(offset_t * offset_ptr) const350 uint16_t DataExtractor::GetU16(offset_t *offset_ptr) const {
351   uint16_t val = 0;
352   const uint8_t *data =
353       static_cast<const uint8_t *>(GetData(offset_ptr, sizeof(val)));
354   if (data) {
355     if (m_byte_order != endian::InlHostByteOrder())
356       val = ReadSwapInt16(data);
357     else
358       val = ReadInt16(data);
359   }
360   return val;
361 }
362 
GetU16_unchecked(offset_t * offset_ptr) const363 uint16_t DataExtractor::GetU16_unchecked(offset_t *offset_ptr) const {
364   uint16_t val;
365   if (m_byte_order == endian::InlHostByteOrder())
366     val = ReadInt16(m_start, *offset_ptr);
367   else
368     val = ReadSwapInt16(m_start, *offset_ptr);
369   *offset_ptr += sizeof(val);
370   return val;
371 }
372 
GetU32_unchecked(offset_t * offset_ptr) const373 uint32_t DataExtractor::GetU32_unchecked(offset_t *offset_ptr) const {
374   uint32_t val;
375   if (m_byte_order == endian::InlHostByteOrder())
376     val = ReadInt32(m_start, *offset_ptr);
377   else
378     val = ReadSwapInt32(m_start, *offset_ptr);
379   *offset_ptr += sizeof(val);
380   return val;
381 }
382 
GetU64_unchecked(offset_t * offset_ptr) const383 uint64_t DataExtractor::GetU64_unchecked(offset_t *offset_ptr) const {
384   uint64_t val;
385   if (m_byte_order == endian::InlHostByteOrder())
386     val = ReadInt64(m_start, *offset_ptr);
387   else
388     val = ReadSwapInt64(m_start, *offset_ptr);
389   *offset_ptr += sizeof(val);
390   return val;
391 }
392 
393 // Extract "count" uint16_t values from the binary data and update the offset
394 // pointed to by "offset_ptr". The extracted data is copied into "dst".
395 //
396 // RETURNS the non-nullptr buffer pointer upon successful extraction of
397 // all the requested bytes, or nullptr when the data is not available in the
398 // buffer due to being out of bounds, or insufficient data.
GetU16(offset_t * offset_ptr,void * void_dst,uint32_t count) const399 void *DataExtractor::GetU16(offset_t *offset_ptr, void *void_dst,
400                             uint32_t count) const {
401   const size_t src_size = sizeof(uint16_t) * count;
402   const uint16_t *src =
403       static_cast<const uint16_t *>(GetData(offset_ptr, src_size));
404   if (src) {
405     if (m_byte_order != endian::InlHostByteOrder()) {
406       uint16_t *dst_pos = static_cast<uint16_t *>(void_dst);
407       uint16_t *dst_end = dst_pos + count;
408       const uint16_t *src_pos = src;
409       while (dst_pos < dst_end) {
410         *dst_pos = ReadSwapInt16(src_pos);
411         ++dst_pos;
412         ++src_pos;
413       }
414     } else {
415       memcpy(void_dst, src, src_size);
416     }
417     // Return a non-nullptr pointer to the converted data as an indicator of
418     // success
419     return void_dst;
420   }
421   return nullptr;
422 }
423 
424 // Extract a single uint32_t from the data and update the offset pointed to by
425 // "offset_ptr".
426 //
427 // RETURNS the uint32_t that was extracted, or zero on failure.
GetU32(offset_t * offset_ptr) const428 uint32_t DataExtractor::GetU32(offset_t *offset_ptr) const {
429   uint32_t val = 0;
430   const uint8_t *data =
431       static_cast<const uint8_t *>(GetData(offset_ptr, sizeof(val)));
432   if (data) {
433     if (m_byte_order != endian::InlHostByteOrder()) {
434       val = ReadSwapInt32(data);
435     } else {
436       memcpy(&val, data, 4);
437     }
438   }
439   return val;
440 }
441 
442 // Extract "count" uint32_t values from the binary data and update the offset
443 // pointed to by "offset_ptr". The extracted data is copied into "dst".
444 //
445 // RETURNS the non-nullptr buffer pointer upon successful extraction of
446 // all the requested bytes, or nullptr when the data is not available in the
447 // buffer due to being out of bounds, or insufficient data.
GetU32(offset_t * offset_ptr,void * void_dst,uint32_t count) const448 void *DataExtractor::GetU32(offset_t *offset_ptr, void *void_dst,
449                             uint32_t count) const {
450   const size_t src_size = sizeof(uint32_t) * count;
451   const uint32_t *src =
452       static_cast<const uint32_t *>(GetData(offset_ptr, src_size));
453   if (src) {
454     if (m_byte_order != endian::InlHostByteOrder()) {
455       uint32_t *dst_pos = static_cast<uint32_t *>(void_dst);
456       uint32_t *dst_end = dst_pos + count;
457       const uint32_t *src_pos = src;
458       while (dst_pos < dst_end) {
459         *dst_pos = ReadSwapInt32(src_pos);
460         ++dst_pos;
461         ++src_pos;
462       }
463     } else {
464       memcpy(void_dst, src, src_size);
465     }
466     // Return a non-nullptr pointer to the converted data as an indicator of
467     // success
468     return void_dst;
469   }
470   return nullptr;
471 }
472 
473 // Extract a single uint64_t from the data and update the offset pointed to by
474 // "offset_ptr".
475 //
476 // RETURNS the uint64_t that was extracted, or zero on failure.
GetU64(offset_t * offset_ptr) const477 uint64_t DataExtractor::GetU64(offset_t *offset_ptr) const {
478   uint64_t val = 0;
479   const uint8_t *data =
480       static_cast<const uint8_t *>(GetData(offset_ptr, sizeof(val)));
481   if (data) {
482     if (m_byte_order != endian::InlHostByteOrder()) {
483       val = ReadSwapInt64(data);
484     } else {
485       memcpy(&val, data, 8);
486     }
487   }
488   return val;
489 }
490 
491 // GetU64
492 //
493 // Get multiple consecutive 64 bit values. Return true if the entire read
494 // succeeds and increment the offset pointed to by offset_ptr, else return
495 // false and leave the offset pointed to by offset_ptr unchanged.
GetU64(offset_t * offset_ptr,void * void_dst,uint32_t count) const496 void *DataExtractor::GetU64(offset_t *offset_ptr, void *void_dst,
497                             uint32_t count) const {
498   const size_t src_size = sizeof(uint64_t) * count;
499   const uint64_t *src =
500       static_cast<const uint64_t *>(GetData(offset_ptr, src_size));
501   if (src) {
502     if (m_byte_order != endian::InlHostByteOrder()) {
503       uint64_t *dst_pos = static_cast<uint64_t *>(void_dst);
504       uint64_t *dst_end = dst_pos + count;
505       const uint64_t *src_pos = src;
506       while (dst_pos < dst_end) {
507         *dst_pos = ReadSwapInt64(src_pos);
508         ++dst_pos;
509         ++src_pos;
510       }
511     } else {
512       memcpy(void_dst, src, src_size);
513     }
514     // Return a non-nullptr pointer to the converted data as an indicator of
515     // success
516     return void_dst;
517   }
518   return nullptr;
519 }
520 
GetMaxU32(offset_t * offset_ptr,size_t byte_size) const521 uint32_t DataExtractor::GetMaxU32(offset_t *offset_ptr,
522                                   size_t byte_size) const {
523   lldbassert(byte_size > 0 && byte_size <= 4 && "GetMaxU32 invalid byte_size!");
524   return GetMaxU64(offset_ptr, byte_size);
525 }
526 
GetMaxU64(offset_t * offset_ptr,size_t byte_size) const527 uint64_t DataExtractor::GetMaxU64(offset_t *offset_ptr,
528                                   size_t byte_size) const {
529   lldbassert(byte_size > 0 && byte_size <= 8 && "GetMaxU64 invalid byte_size!");
530   switch (byte_size) {
531   case 1:
532     return GetU8(offset_ptr);
533   case 2:
534     return GetU16(offset_ptr);
535   case 4:
536     return GetU32(offset_ptr);
537   case 8:
538     return GetU64(offset_ptr);
539   default: {
540     // General case.
541     const uint8_t *data =
542         static_cast<const uint8_t *>(GetData(offset_ptr, byte_size));
543     if (data == nullptr)
544       return 0;
545     return ReadMaxInt64(data, byte_size, m_byte_order);
546   }
547   }
548   return 0;
549 }
550 
GetMaxU64_unchecked(offset_t * offset_ptr,size_t byte_size) const551 uint64_t DataExtractor::GetMaxU64_unchecked(offset_t *offset_ptr,
552                                             size_t byte_size) const {
553   switch (byte_size) {
554   case 1:
555     return GetU8_unchecked(offset_ptr);
556   case 2:
557     return GetU16_unchecked(offset_ptr);
558   case 4:
559     return GetU32_unchecked(offset_ptr);
560   case 8:
561     return GetU64_unchecked(offset_ptr);
562   default: {
563     uint64_t res = ReadMaxInt64(&m_start[*offset_ptr], byte_size, m_byte_order);
564     *offset_ptr += byte_size;
565     return res;
566   }
567   }
568   return 0;
569 }
570 
GetMaxS64(offset_t * offset_ptr,size_t byte_size) const571 int64_t DataExtractor::GetMaxS64(offset_t *offset_ptr, size_t byte_size) const {
572   uint64_t u64 = GetMaxU64(offset_ptr, byte_size);
573   return llvm::SignExtend64(u64, 8 * byte_size);
574 }
575 
GetMaxU64Bitfield(offset_t * offset_ptr,size_t size,uint32_t bitfield_bit_size,uint32_t bitfield_bit_offset) const576 uint64_t DataExtractor::GetMaxU64Bitfield(offset_t *offset_ptr, size_t size,
577                                           uint32_t bitfield_bit_size,
578                                           uint32_t bitfield_bit_offset) const {
579   assert(bitfield_bit_size <= 64);
580   uint64_t uval64 = GetMaxU64(offset_ptr, size);
581 
582   if (bitfield_bit_size == 0)
583     return uval64;
584 
585   int32_t lsbcount = bitfield_bit_offset;
586   if (m_byte_order == eByteOrderBig)
587     lsbcount = size * 8 - bitfield_bit_offset - bitfield_bit_size;
588 
589   if (lsbcount > 0)
590     uval64 >>= lsbcount;
591 
592   uint64_t bitfield_mask =
593       (bitfield_bit_size == 64
594            ? std::numeric_limits<uint64_t>::max()
595            : ((static_cast<uint64_t>(1) << bitfield_bit_size) - 1));
596   if (!bitfield_mask && bitfield_bit_offset == 0 && bitfield_bit_size == 64)
597     return uval64;
598 
599   uval64 &= bitfield_mask;
600 
601   return uval64;
602 }
603 
GetMaxS64Bitfield(offset_t * offset_ptr,size_t size,uint32_t bitfield_bit_size,uint32_t bitfield_bit_offset) const604 int64_t DataExtractor::GetMaxS64Bitfield(offset_t *offset_ptr, size_t size,
605                                          uint32_t bitfield_bit_size,
606                                          uint32_t bitfield_bit_offset) const {
607   assert(size >= 1 && "GetMaxS64Bitfield size must be >= 1");
608   assert(size <= 8 && "GetMaxS64Bitfield size must be <= 8");
609   int64_t sval64 = GetMaxS64(offset_ptr, size);
610   if (bitfield_bit_size == 0)
611     return sval64;
612   int32_t lsbcount = bitfield_bit_offset;
613   if (m_byte_order == eByteOrderBig)
614     lsbcount = size * 8 - bitfield_bit_offset - bitfield_bit_size;
615   if (lsbcount > 0)
616     sval64 >>= lsbcount;
617   uint64_t bitfield_mask = llvm::maskTrailingOnes<uint64_t>(bitfield_bit_size);
618   sval64 &= bitfield_mask;
619   // sign extend if needed
620   if (sval64 & ((static_cast<uint64_t>(1)) << (bitfield_bit_size - 1)))
621     sval64 |= ~bitfield_mask;
622   return sval64;
623 }
624 
GetFloat(offset_t * offset_ptr) const625 float DataExtractor::GetFloat(offset_t *offset_ptr) const {
626   return Get<float>(offset_ptr, 0.0f);
627 }
628 
GetDouble(offset_t * offset_ptr) const629 double DataExtractor::GetDouble(offset_t *offset_ptr) const {
630   return Get<double>(offset_ptr, 0.0);
631 }
632 
GetLongDouble(offset_t * offset_ptr) const633 long double DataExtractor::GetLongDouble(offset_t *offset_ptr) const {
634   long double val = 0.0;
635 #if defined(__i386__) || defined(__amd64__) || defined(__x86_64__) ||          \
636     defined(_M_IX86) || defined(_M_IA64) || defined(_M_X64)
637   *offset_ptr += CopyByteOrderedData(*offset_ptr, 10, &val, sizeof(val),
638                                      endian::InlHostByteOrder());
639 #else
640   *offset_ptr += CopyByteOrderedData(*offset_ptr, sizeof(val), &val,
641                                      sizeof(val), endian::InlHostByteOrder());
642 #endif
643   return val;
644 }
645 
646 // Extract a single address from the data and update the offset pointed to by
647 // "offset_ptr". The size of the extracted address comes from the
648 // "this->m_addr_size" member variable and should be set correctly prior to
649 // extracting any address values.
650 //
651 // RETURNS the address that was extracted, or zero on failure.
GetAddress(offset_t * offset_ptr) const652 uint64_t DataExtractor::GetAddress(offset_t *offset_ptr) const {
653   assert(m_addr_size >= 1 && m_addr_size <= 8);
654   return GetMaxU64(offset_ptr, m_addr_size);
655 }
656 
GetAddress_unchecked(offset_t * offset_ptr) const657 uint64_t DataExtractor::GetAddress_unchecked(offset_t *offset_ptr) const {
658   assert(m_addr_size >= 1 && m_addr_size <= 8);
659   return GetMaxU64_unchecked(offset_ptr, m_addr_size);
660 }
661 
ExtractBytes(offset_t offset,offset_t length,ByteOrder dst_byte_order,void * dst) const662 size_t DataExtractor::ExtractBytes(offset_t offset, offset_t length,
663                                    ByteOrder dst_byte_order, void *dst) const {
664   const uint8_t *src = PeekData(offset, length);
665   if (src) {
666     if (dst_byte_order != GetByteOrder()) {
667       // Validate that only a word- or register-sized dst is byte swapped
668       assert(length == 1 || length == 2 || length == 4 || length == 8 ||
669              length == 10 || length == 16 || length == 32);
670 
671       for (uint32_t i = 0; i < length; ++i)
672         (static_cast<uint8_t *>(dst))[i] = src[length - i - 1];
673     } else
674       ::memcpy(dst, src, length);
675     return length;
676   }
677   return 0;
678 }
679 
680 // Extract data as it exists in target memory
CopyData(offset_t offset,offset_t length,void * dst) const681 lldb::offset_t DataExtractor::CopyData(offset_t offset, offset_t length,
682                                        void *dst) const {
683   const uint8_t *src = PeekData(offset, length);
684   if (src) {
685     ::memcpy(dst, src, length);
686     return length;
687   }
688   return 0;
689 }
690 
691 // Extract data and swap if needed when doing the copy
692 lldb::offset_t
CopyByteOrderedData(offset_t src_offset,offset_t src_len,void * dst_void_ptr,offset_t dst_len,ByteOrder dst_byte_order) const693 DataExtractor::CopyByteOrderedData(offset_t src_offset, offset_t src_len,
694                                    void *dst_void_ptr, offset_t dst_len,
695                                    ByteOrder dst_byte_order) const {
696   // Validate the source info
697   if (!ValidOffsetForDataOfSize(src_offset, src_len))
698     assert(ValidOffsetForDataOfSize(src_offset, src_len));
699   assert(src_len > 0);
700   assert(m_byte_order == eByteOrderBig || m_byte_order == eByteOrderLittle);
701 
702   // Validate the destination info
703   assert(dst_void_ptr != nullptr);
704   assert(dst_len > 0);
705   assert(dst_byte_order == eByteOrderBig || dst_byte_order == eByteOrderLittle);
706 
707   // Validate that only a word- or register-sized dst is byte swapped
708   assert(dst_byte_order == m_byte_order || dst_len == 1 || dst_len == 2 ||
709          dst_len == 4 || dst_len == 8 || dst_len == 10 || dst_len == 16 ||
710          dst_len == 32);
711 
712   // Must have valid byte orders set in this object and for destination
713   if (!(dst_byte_order == eByteOrderBig ||
714         dst_byte_order == eByteOrderLittle) ||
715       !(m_byte_order == eByteOrderBig || m_byte_order == eByteOrderLittle))
716     return 0;
717 
718   uint8_t *dst = static_cast<uint8_t *>(dst_void_ptr);
719   const uint8_t *src = PeekData(src_offset, src_len);
720   if (src) {
721     if (dst_len >= src_len) {
722       // We are copying the entire value from src into dst. Calculate how many,
723       // if any, zeroes we need for the most significant bytes if "dst_len" is
724       // greater than "src_len"...
725       const size_t num_zeroes = dst_len - src_len;
726       if (dst_byte_order == eByteOrderBig) {
727         // Big endian, so we lead with zeroes...
728         if (num_zeroes > 0)
729           ::memset(dst, 0, num_zeroes);
730         // Then either copy or swap the rest
731         if (m_byte_order == eByteOrderBig) {
732           ::memcpy(dst + num_zeroes, src, src_len);
733         } else {
734           for (uint32_t i = 0; i < src_len; ++i)
735             dst[i + num_zeroes] = src[src_len - 1 - i];
736         }
737       } else {
738         // Little endian destination, so we lead the value bytes
739         if (m_byte_order == eByteOrderBig) {
740           for (uint32_t i = 0; i < src_len; ++i)
741             dst[i] = src[src_len - 1 - i];
742         } else {
743           ::memcpy(dst, src, src_len);
744         }
745         // And zero the rest...
746         if (num_zeroes > 0)
747           ::memset(dst + src_len, 0, num_zeroes);
748       }
749       return src_len;
750     } else {
751       // We are only copying some of the value from src into dst..
752 
753       if (dst_byte_order == eByteOrderBig) {
754         // Big endian dst
755         if (m_byte_order == eByteOrderBig) {
756           // Big endian dst, with big endian src
757           ::memcpy(dst, src + (src_len - dst_len), dst_len);
758         } else {
759           // Big endian dst, with little endian src
760           for (uint32_t i = 0; i < dst_len; ++i)
761             dst[i] = src[dst_len - 1 - i];
762         }
763       } else {
764         // Little endian dst
765         if (m_byte_order == eByteOrderBig) {
766           // Little endian dst, with big endian src
767           for (uint32_t i = 0; i < dst_len; ++i)
768             dst[i] = src[src_len - 1 - i];
769         } else {
770           // Little endian dst, with big endian src
771           ::memcpy(dst, src, dst_len);
772         }
773       }
774       return dst_len;
775     }
776   }
777   return 0;
778 }
779 
780 // Extracts a variable length NULL terminated C string from the data at the
781 // offset pointed to by "offset_ptr".  The "offset_ptr" will be updated with
782 // the offset of the byte that follows the NULL terminator byte.
783 //
784 // If the offset pointed to by "offset_ptr" is out of bounds, or if "length" is
785 // non-zero and there aren't enough available bytes, nullptr will be returned
786 // and "offset_ptr" will not be updated.
GetCStr(offset_t * offset_ptr) const787 const char *DataExtractor::GetCStr(offset_t *offset_ptr) const {
788   const char *start = reinterpret_cast<const char *>(PeekData(*offset_ptr, 1));
789   // Already at the end of the data.
790   if (!start)
791     return nullptr;
792 
793   const char *end = reinterpret_cast<const char *>(m_end);
794 
795   // Check all bytes for a null terminator that terminates a C string.
796   const char *terminator_or_end = std::find(start, end, '\0');
797 
798   // We didn't find a null terminator, so return nullptr to indicate that there
799   // is no valid C string at that offset.
800   if (terminator_or_end == end)
801     return nullptr;
802 
803   // Update offset_ptr for the caller to point to the data behind the
804   // terminator (which is 1 byte long).
805   *offset_ptr += (terminator_or_end - start + 1UL);
806   return start;
807 }
808 
809 // Extracts a NULL terminated C string from the fixed length field of length
810 // "len" at the offset pointed to by "offset_ptr". The "offset_ptr" will be
811 // updated with the offset of the byte that follows the fixed length field.
812 //
813 // If the offset pointed to by "offset_ptr" is out of bounds, or if the offset
814 // plus the length of the field is out of bounds, or if the field does not
815 // contain a NULL terminator byte, nullptr will be returned and "offset_ptr"
816 // will not be updated.
GetCStr(offset_t * offset_ptr,offset_t len) const817 const char *DataExtractor::GetCStr(offset_t *offset_ptr, offset_t len) const {
818   const char *cstr = reinterpret_cast<const char *>(PeekData(*offset_ptr, len));
819   if (cstr != nullptr) {
820     if (memchr(cstr, '\0', len) == nullptr) {
821       return nullptr;
822     }
823     *offset_ptr += len;
824     return cstr;
825   }
826   return nullptr;
827 }
828 
829 // Peeks at a string in the contained data. No verification is done to make
830 // sure the entire string lies within the bounds of this object's data, only
831 // "offset" is verified to be a valid offset.
832 //
833 // Returns a valid C string pointer if "offset" is a valid offset in this
834 // object's data, else nullptr is returned.
PeekCStr(offset_t offset) const835 const char *DataExtractor::PeekCStr(offset_t offset) const {
836   return reinterpret_cast<const char *>(PeekData(offset, 1));
837 }
838 
839 // Extracts an unsigned LEB128 number from this object's data starting at the
840 // offset pointed to by "offset_ptr". The offset pointed to by "offset_ptr"
841 // will be updated with the offset of the byte following the last extracted
842 // byte.
843 //
844 // Returned the extracted integer value.
GetULEB128(offset_t * offset_ptr) const845 uint64_t DataExtractor::GetULEB128(offset_t *offset_ptr) const {
846   const uint8_t *src = PeekData(*offset_ptr, 1);
847   if (src == nullptr)
848     return 0;
849 
850   unsigned byte_count = 0;
851   uint64_t result = llvm::decodeULEB128(src, &byte_count, m_end);
852   *offset_ptr += byte_count;
853   return result;
854 }
855 
856 // Extracts an signed LEB128 number from this object's data starting at the
857 // offset pointed to by "offset_ptr". The offset pointed to by "offset_ptr"
858 // will be updated with the offset of the byte following the last extracted
859 // byte.
860 //
861 // Returned the extracted integer value.
GetSLEB128(offset_t * offset_ptr) const862 int64_t DataExtractor::GetSLEB128(offset_t *offset_ptr) const {
863   const uint8_t *src = PeekData(*offset_ptr, 1);
864   if (src == nullptr)
865     return 0;
866 
867   unsigned byte_count = 0;
868   int64_t result = llvm::decodeSLEB128(src, &byte_count, m_end);
869   *offset_ptr += byte_count;
870   return result;
871 }
872 
873 // Skips a ULEB128 number (signed or unsigned) from this object's data starting
874 // at the offset pointed to by "offset_ptr". The offset pointed to by
875 // "offset_ptr" will be updated with the offset of the byte following the last
876 // extracted byte.
877 //
878 // Returns the number of bytes consumed during the extraction.
Skip_LEB128(offset_t * offset_ptr) const879 uint32_t DataExtractor::Skip_LEB128(offset_t *offset_ptr) const {
880   uint32_t bytes_consumed = 0;
881   const uint8_t *src = PeekData(*offset_ptr, 1);
882   if (src == nullptr)
883     return 0;
884 
885   const uint8_t *end = m_end;
886 
887   if (src < end) {
888     const uint8_t *src_pos = src;
889     while ((src_pos < end) && (*src_pos++ & 0x80))
890       ++bytes_consumed;
891     *offset_ptr += src_pos - src;
892   }
893   return bytes_consumed;
894 }
895 
896 // Dumps bytes from this object's data to the stream "s" starting
897 // "start_offset" bytes into this data, and ending with the byte before
898 // "end_offset". "base_addr" will be added to the offset into the dumped data
899 // when showing the offset into the data in the output information.
900 // "num_per_line" objects of type "type" will be dumped with the option to
901 // override the format for each object with "type_format". "type_format" is a
902 // printf style formatting string. If "type_format" is nullptr, then an
903 // appropriate format string will be used for the supplied "type". If the
904 // stream "s" is nullptr, then the output will be send to Log().
PutToLog(Log * log,offset_t start_offset,offset_t length,uint64_t base_addr,uint32_t num_per_line,DataExtractor::Type type) const905 lldb::offset_t DataExtractor::PutToLog(Log *log, offset_t start_offset,
906                                        offset_t length, uint64_t base_addr,
907                                        uint32_t num_per_line,
908                                        DataExtractor::Type type) const {
909   if (log == nullptr)
910     return start_offset;
911 
912   offset_t offset;
913   offset_t end_offset;
914   uint32_t count;
915   StreamString sstr;
916   for (offset = start_offset, end_offset = offset + length, count = 0;
917        ValidOffset(offset) && offset < end_offset; ++count) {
918     if ((count % num_per_line) == 0) {
919       // Print out any previous string
920       if (sstr.GetSize() > 0) {
921         log->PutString(sstr.GetString());
922         sstr.Clear();
923       }
924       // Reset string offset and fill the current line string with address:
925       if (base_addr != LLDB_INVALID_ADDRESS)
926         sstr.Printf("0x%8.8" PRIx64 ":",
927                     static_cast<uint64_t>(base_addr + (offset - start_offset)));
928     }
929 
930     switch (type) {
931     case TypeUInt8:
932       sstr.Printf(" %2.2x", GetU8(&offset));
933       break;
934     case TypeChar: {
935       char ch = GetU8(&offset);
936       sstr.Printf(" %c", llvm::isPrint(ch) ? ch : ' ');
937     } break;
938     case TypeUInt16:
939       sstr.Printf(" %4.4x", GetU16(&offset));
940       break;
941     case TypeUInt32:
942       sstr.Printf(" %8.8x", GetU32(&offset));
943       break;
944     case TypeUInt64:
945       sstr.Printf(" %16.16" PRIx64, GetU64(&offset));
946       break;
947     case TypePointer:
948       sstr.Printf(" 0x%" PRIx64, GetAddress(&offset));
949       break;
950     case TypeULEB128:
951       sstr.Printf(" 0x%" PRIx64, GetULEB128(&offset));
952       break;
953     case TypeSLEB128:
954       sstr.Printf(" %" PRId64, GetSLEB128(&offset));
955       break;
956     }
957   }
958 
959   if (!sstr.Empty())
960     log->PutString(sstr.GetString());
961 
962   return offset; // Return the offset at which we ended up
963 }
964 
Copy(DataExtractor & dest_data) const965 size_t DataExtractor::Copy(DataExtractor &dest_data) const {
966   if (m_data_sp) {
967     // we can pass along the SP to the data
968     dest_data.SetData(m_data_sp);
969   } else {
970     const uint8_t *base_ptr = m_start;
971     size_t data_size = GetByteSize();
972     dest_data.SetData(DataBufferSP(new DataBufferHeap(base_ptr, data_size)));
973   }
974   return GetByteSize();
975 }
976 
Append(DataExtractor & rhs)977 bool DataExtractor::Append(DataExtractor &rhs) {
978   if (rhs.GetByteOrder() != GetByteOrder())
979     return false;
980 
981   if (rhs.GetByteSize() == 0)
982     return true;
983 
984   if (GetByteSize() == 0)
985     return (rhs.Copy(*this) > 0);
986 
987   size_t bytes = GetByteSize() + rhs.GetByteSize();
988 
989   DataBufferHeap *buffer_heap_ptr = nullptr;
990   DataBufferSP buffer_sp(buffer_heap_ptr = new DataBufferHeap(bytes, 0));
991 
992   if (!buffer_sp || buffer_heap_ptr == nullptr)
993     return false;
994 
995   uint8_t *bytes_ptr = buffer_heap_ptr->GetBytes();
996 
997   memcpy(bytes_ptr, GetDataStart(), GetByteSize());
998   memcpy(bytes_ptr + GetByteSize(), rhs.GetDataStart(), rhs.GetByteSize());
999 
1000   SetData(buffer_sp);
1001 
1002   return true;
1003 }
1004 
Append(void * buf,offset_t length)1005 bool DataExtractor::Append(void *buf, offset_t length) {
1006   if (buf == nullptr)
1007     return false;
1008 
1009   if (length == 0)
1010     return true;
1011 
1012   size_t bytes = GetByteSize() + length;
1013 
1014   DataBufferHeap *buffer_heap_ptr = nullptr;
1015   DataBufferSP buffer_sp(buffer_heap_ptr = new DataBufferHeap(bytes, 0));
1016 
1017   if (!buffer_sp || buffer_heap_ptr == nullptr)
1018     return false;
1019 
1020   uint8_t *bytes_ptr = buffer_heap_ptr->GetBytes();
1021 
1022   if (GetByteSize() > 0)
1023     memcpy(bytes_ptr, GetDataStart(), GetByteSize());
1024 
1025   memcpy(bytes_ptr + GetByteSize(), buf, length);
1026 
1027   SetData(buffer_sp);
1028 
1029   return true;
1030 }
1031 
Checksum(llvm::SmallVectorImpl<uint8_t> & dest,uint64_t max_data)1032 void DataExtractor::Checksum(llvm::SmallVectorImpl<uint8_t> &dest,
1033                              uint64_t max_data) {
1034   if (max_data == 0)
1035     max_data = GetByteSize();
1036   else
1037     max_data = std::min(max_data, GetByteSize());
1038 
1039   llvm::MD5 md5;
1040 
1041   const llvm::ArrayRef<uint8_t> data(GetDataStart(), max_data);
1042   md5.update(data);
1043 
1044   llvm::MD5::MD5Result result;
1045   md5.final(result);
1046 
1047   dest.clear();
1048   dest.append(result.Bytes.begin(), result.Bytes.end());
1049 }
1050