blob: 63a1c54a40f88e405b4713cf5b18f9c49f1ecf33 [file] [log] [blame]
Chris Lattner333ffd42007-04-29 06:58:52 +00001//===--- MemoryBuffer.cpp - Memory Buffer implementation ------------------===//
2//
3// The LLVM Compiler Infrastructure
4//
Chris Lattner4ee451d2007-12-29 20:36:04 +00005// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
Chris Lattner333ffd42007-04-29 06:58:52 +00007//
8//===----------------------------------------------------------------------===//
9//
10// This file implements the MemoryBuffer interface.
11//
12//===----------------------------------------------------------------------===//
13
14#include "llvm/Support/MemoryBuffer.h"
Chris Lattner11da4cf2008-04-01 06:05:21 +000015#include "llvm/ADT/OwningPtr.h"
16#include "llvm/ADT/SmallString.h"
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000017#include "llvm/Support/MathExtras.h"
Michael J. Spencer1f6efa32010-11-29 18:16:10 +000018#include "llvm/Support/Errno.h"
19#include "llvm/Support/Path.h"
20#include "llvm/Support/Process.h"
21#include "llvm/Support/Program.h"
Michael J. Spencer333fb042010-12-09 17:36:48 +000022#include "llvm/Support/system_error.h"
Jeff Cohen9bc40602007-04-29 14:21:44 +000023#include <cassert>
Chris Lattner333ffd42007-04-29 06:58:52 +000024#include <cstdio>
25#include <cstring>
26#include <cerrno>
Nick Lewycky476b2422010-12-19 20:43:38 +000027#include <new>
Chris Lattner11da4cf2008-04-01 06:05:21 +000028#include <sys/types.h>
29#include <sys/stat.h>
30#if !defined(_MSC_VER) && !defined(__MINGW32__)
31#include <unistd.h>
32#include <sys/uio.h>
Chris Lattner11da4cf2008-04-01 06:05:21 +000033#else
34#include <io.h>
35#endif
Gabor Greif88110802008-04-30 08:53:22 +000036#include <fcntl.h>
Chris Lattner333ffd42007-04-29 06:58:52 +000037using namespace llvm;
38
Michael J. Spencer3ff95632010-12-16 03:29:14 +000039namespace { const llvm::error_code success; }
40
Chris Lattner333ffd42007-04-29 06:58:52 +000041//===----------------------------------------------------------------------===//
42// MemoryBuffer implementation itself.
43//===----------------------------------------------------------------------===//
44
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000045MemoryBuffer::~MemoryBuffer() { }
Chris Lattner333ffd42007-04-29 06:58:52 +000046
47/// init - Initialize this MemoryBuffer as a reference to externally allocated
48/// memory, memory that we know is already null terminated.
Rafael Espindolaf7fdad12011-03-10 18:33:29 +000049void MemoryBuffer::init(const char *BufStart, const char *BufEnd,
50 bool RequiresNullTerminator) {
51 assert((BufEnd[0] == 0 || !RequiresNullTerminator) &&
52 "Buffer is not null terminated!");
Chris Lattner333ffd42007-04-29 06:58:52 +000053 BufferStart = BufStart;
54 BufferEnd = BufEnd;
Chris Lattner333ffd42007-04-29 06:58:52 +000055}
56
57//===----------------------------------------------------------------------===//
58// MemoryBufferMem implementation.
59//===----------------------------------------------------------------------===//
60
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000061/// CopyStringRef - Copies contents of a StringRef into a block of memory and
62/// null-terminates it.
63static void CopyStringRef(char *Memory, StringRef Data) {
64 memcpy(Memory, Data.data(), Data.size());
65 Memory[Data.size()] = 0; // Null terminate string.
66}
67
68/// GetNamedBuffer - Allocates a new MemoryBuffer with Name copied after it.
69template <typename T>
Rafael Espindolaf7fdad12011-03-10 18:33:29 +000070static T* GetNamedBuffer(StringRef Buffer, StringRef Name,
71 bool RequiresNullTerminator) {
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000072 char *Mem = static_cast<char*>(operator new(sizeof(T) + Name.size() + 1));
73 CopyStringRef(Mem + sizeof(T), Name);
Rafael Espindolaf7fdad12011-03-10 18:33:29 +000074 return new (Mem) T(Buffer, RequiresNullTerminator);
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000075}
76
Chris Lattner333ffd42007-04-29 06:58:52 +000077namespace {
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000078/// MemoryBufferMem - Named MemoryBuffer pointing to a block of memory.
Chris Lattner333ffd42007-04-29 06:58:52 +000079class MemoryBufferMem : public MemoryBuffer {
Chris Lattner333ffd42007-04-29 06:58:52 +000080public:
Rafael Espindolaf7fdad12011-03-10 18:33:29 +000081 MemoryBufferMem(StringRef InputData, bool RequiresNullTerminator) {
82 init(InputData.begin(), InputData.end(), RequiresNullTerminator);
Chris Lattner333ffd42007-04-29 06:58:52 +000083 }
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000084
Chris Lattner333ffd42007-04-29 06:58:52 +000085 virtual const char *getBufferIdentifier() const {
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000086 // The name is stored after the class itself.
87 return reinterpret_cast<const char*>(this + 1);
Chris Lattner333ffd42007-04-29 06:58:52 +000088 }
89};
90}
91
92/// getMemBuffer - Open the specified memory range as a MemoryBuffer. Note
93/// that EndPtr[0] must be a null byte and be accessible!
Chris Lattner4c842dd2010-04-05 22:42:30 +000094MemoryBuffer *MemoryBuffer::getMemBuffer(StringRef InputData,
Rafael Espindola9916d2a2011-03-17 22:18:42 +000095 StringRef BufferName,
96 bool RequiresNullTerminator) {
97 return GetNamedBuffer<MemoryBufferMem>(InputData, BufferName,
98 RequiresNullTerminator);
Chris Lattner333ffd42007-04-29 06:58:52 +000099}
100
Chris Lattner3daae272007-10-09 21:46:38 +0000101/// getMemBufferCopy - Open the specified memory range as a MemoryBuffer,
102/// copying the contents and taking ownership of it. This has no requirements
103/// on EndPtr[0].
Chris Lattner4c842dd2010-04-05 22:42:30 +0000104MemoryBuffer *MemoryBuffer::getMemBufferCopy(StringRef InputData,
Benjamin Kramerd4d1f852010-06-25 11:50:40 +0000105 StringRef BufferName) {
106 MemoryBuffer *Buf = getNewUninitMemBuffer(InputData.size(), BufferName);
107 if (!Buf) return 0;
108 memcpy(const_cast<char*>(Buf->getBufferStart()), InputData.data(),
109 InputData.size());
110 return Buf;
Chris Lattner3daae272007-10-09 21:46:38 +0000111}
112
Chris Lattner333ffd42007-04-29 06:58:52 +0000113/// getNewUninitMemBuffer - Allocate a new MemoryBuffer of the specified size
Benjamin Kramerd4d1f852010-06-25 11:50:40 +0000114/// that is not initialized. Note that the caller should initialize the
115/// memory allocated by this method. The memory is owned by the MemoryBuffer
116/// object.
Evan Cheng34cd4a42008-05-05 18:30:58 +0000117MemoryBuffer *MemoryBuffer::getNewUninitMemBuffer(size_t Size,
Daniel Dunbard65267e2009-11-10 00:43:58 +0000118 StringRef BufferName) {
Benjamin Kramerd4d1f852010-06-25 11:50:40 +0000119 // Allocate space for the MemoryBuffer, the data and the name. It is important
120 // that MemoryBuffer and data are aligned so PointerIntPair works with them.
121 size_t AlignedStringLen =
122 RoundUpToAlignment(sizeof(MemoryBufferMem) + BufferName.size() + 1,
123 sizeof(void*)); // TODO: Is sizeof(void*) enough?
124 size_t RealLen = AlignedStringLen + Size + 1;
125 char *Mem = static_cast<char*>(operator new(RealLen, std::nothrow));
126 if (!Mem) return 0;
127
128 // The name is stored after the class itself.
129 CopyStringRef(Mem + sizeof(MemoryBufferMem), BufferName);
130
131 // The buffer begins after the name and must be aligned.
132 char *Buf = Mem + AlignedStringLen;
133 Buf[Size] = 0; // Null terminate buffer.
134
Rafael Espindolaf7fdad12011-03-10 18:33:29 +0000135 return new (Mem) MemoryBufferMem(StringRef(Buf, Size), true);
Chris Lattner333ffd42007-04-29 06:58:52 +0000136}
137
138/// getNewMemBuffer - Allocate a new MemoryBuffer of the specified size that
139/// is completely initialized to zeros. Note that the caller should
140/// initialize the memory allocated by this method. The memory is owned by
141/// the MemoryBuffer object.
Benjamin Kramerd4d1f852010-06-25 11:50:40 +0000142MemoryBuffer *MemoryBuffer::getNewMemBuffer(size_t Size, StringRef BufferName) {
Chris Lattner333ffd42007-04-29 06:58:52 +0000143 MemoryBuffer *SB = getNewUninitMemBuffer(Size, BufferName);
Evan Cheng726135a2009-02-13 07:54:34 +0000144 if (!SB) return 0;
Benjamin Kramerd4d1f852010-06-25 11:50:40 +0000145 memset(const_cast<char*>(SB->getBufferStart()), 0, Size);
Chris Lattner333ffd42007-04-29 06:58:52 +0000146 return SB;
147}
148
149
Chris Lattner2b1f1062007-11-18 18:52:28 +0000150/// getFileOrSTDIN - Open the specified file as a MemoryBuffer, or open stdin
151/// if the Filename is "-". If an error occurs, this returns null and fills
152/// in *ErrStr with a reason. If stdin is empty, this API (unlike getSTDIN)
153/// returns an empty buffer.
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000154error_code MemoryBuffer::getFileOrSTDIN(StringRef Filename,
155 OwningPtr<MemoryBuffer> &result,
156 int64_t FileSize) {
Daniel Dunbard65267e2009-11-10 00:43:58 +0000157 if (Filename == "-")
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000158 return getSTDIN(result);
159 return getFile(Filename, result, FileSize);
Chris Lattner2b1f1062007-11-18 18:52:28 +0000160}
161
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000162error_code MemoryBuffer::getFileOrSTDIN(const char *Filename,
163 OwningPtr<MemoryBuffer> &result,
164 int64_t FileSize) {
Dan Gohman60e6f3d2010-06-24 16:25:50 +0000165 if (strcmp(Filename, "-") == 0)
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000166 return getSTDIN(result);
167 return getFile(Filename, result, FileSize);
Dan Gohman60e6f3d2010-06-24 16:25:50 +0000168}
169
Chris Lattner333ffd42007-04-29 06:58:52 +0000170//===----------------------------------------------------------------------===//
Chris Lattner333ffd42007-04-29 06:58:52 +0000171// MemoryBuffer::getFile implementation.
172//===----------------------------------------------------------------------===//
173
Chris Lattner11da4cf2008-04-01 06:05:21 +0000174namespace {
175/// MemoryBufferMMapFile - This represents a file that was mapped in with the
176/// sys::Path::MapInFilePages method. When destroyed, it calls the
177/// sys::Path::UnMapFilePages method.
Benjamin Kramerd4d1f852010-06-25 11:50:40 +0000178class MemoryBufferMMapFile : public MemoryBufferMem {
Chris Lattner11da4cf2008-04-01 06:05:21 +0000179public:
Rafael Espindolaf7fdad12011-03-10 18:33:29 +0000180 MemoryBufferMMapFile(StringRef Buffer, bool RequiresNullTerminator)
181 : MemoryBufferMem(Buffer, RequiresNullTerminator) { }
Benjamin Kramerd4d1f852010-06-25 11:50:40 +0000182
Chris Lattner11da4cf2008-04-01 06:05:21 +0000183 ~MemoryBufferMMapFile() {
Rafael Espindolaf7fdad12011-03-10 18:33:29 +0000184 static int PageSize = sys::Process::GetPageSize();
185
186 uintptr_t Start = reinterpret_cast<uintptr_t>(getBufferStart());
187 size_t Size = getBufferSize();
188 uintptr_t RealStart = Start & ~(PageSize - 1);
189 size_t RealSize = Size + (Start - RealStart);
190
191 sys::Path::UnMapFilePages(reinterpret_cast<const char*>(RealStart),
192 RealSize);
Chris Lattner11da4cf2008-04-01 06:05:21 +0000193 }
194};
195}
196
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000197error_code MemoryBuffer::getFile(StringRef Filename,
198 OwningPtr<MemoryBuffer> &result,
199 int64_t FileSize) {
Chris Lattner7cf70542010-11-23 22:20:27 +0000200 // Ensure the path is null terminated.
Dan Gohman60e6f3d2010-06-24 16:25:50 +0000201 SmallString<256> PathBuf(Filename.begin(), Filename.end());
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000202 return MemoryBuffer::getFile(PathBuf.c_str(), result, FileSize);
Dan Gohman60e6f3d2010-06-24 16:25:50 +0000203}
204
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000205error_code MemoryBuffer::getFile(const char *Filename,
206 OwningPtr<MemoryBuffer> &result,
207 int64_t FileSize) {
Dan Gohman60e6f3d2010-06-24 16:25:50 +0000208 int OpenFlags = O_RDONLY;
Chris Lattner11da4cf2008-04-01 06:05:21 +0000209#ifdef O_BINARY
Bill Wendlinga4420062008-04-01 22:09:20 +0000210 OpenFlags |= O_BINARY; // Open input file in binary mode on win32.
Chris Lattner11da4cf2008-04-01 06:05:21 +0000211#endif
Dan Gohman60e6f3d2010-06-24 16:25:50 +0000212 int FD = ::open(Filename, OpenFlags);
Chris Lattner333ffd42007-04-29 06:58:52 +0000213 if (FD == -1) {
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000214 return error_code(errno, posix_category());
Chris Lattner333ffd42007-04-29 06:58:52 +0000215 }
Rafael Espindolab4cc0312011-02-08 22:40:47 +0000216 error_code ret = getOpenFile(FD, Filename, result, FileSize);
217 close(FD);
218 return ret;
Chris Lattner7cf70542010-11-23 22:20:27 +0000219}
Michael J. Spencer1f6efa32010-11-29 18:16:10 +0000220
Rafael Espindola9d2234d2011-03-10 20:54:07 +0000221static bool shouldUseMmap(int FD,
222 size_t FileSize,
Rafael Espindolaf7fdad12011-03-10 18:33:29 +0000223 size_t MapSize,
224 off_t Offset,
225 bool RequiresNullTerminator,
226 int PageSize) {
227 // We don't use mmap for small files because this can severely fragment our
228 // address space.
229 if (MapSize < 4096*4)
230 return false;
231
232 if (!RequiresNullTerminator)
233 return true;
234
Rafael Espindola9d2234d2011-03-10 20:54:07 +0000235
236 // If we don't know the file size, use fstat to find out. fstat on an open
237 // file descriptor is cheaper than stat on a random path.
238 // FIXME: this chunk of code is duplicated, but it avoids a fstat when
239 // RequiresNullTerminator = false and MapSize != -1.
240 if (FileSize == size_t(-1)) {
241 struct stat FileInfo;
242 // TODO: This should use fstat64 when available.
243 if (fstat(FD, &FileInfo) == -1) {
244 return error_code(errno, posix_category());
245 }
246 FileSize = FileInfo.st_size;
247 }
248
Rafael Espindolaf7fdad12011-03-10 18:33:29 +0000249 // If we need a null terminator and the end of the map is inside the file,
250 // we cannot use mmap.
251 size_t End = Offset + MapSize;
252 assert(End <= FileSize);
253 if (End != FileSize)
254 return false;
255
256 // Don't try to map files that are exactly a multiple of the system page size
257 // if we need a null terminator.
258 if ((FileSize & (PageSize -1)) == 0)
259 return false;
260
261 return true;
262}
263
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000264error_code MemoryBuffer::getOpenFile(int FD, const char *Filename,
265 OwningPtr<MemoryBuffer> &result,
Rafael Espindolaf7fdad12011-03-10 18:33:29 +0000266 size_t FileSize, size_t MapSize,
267 off_t Offset,
268 bool RequiresNullTerminator) {
269 static int PageSize = sys::Process::GetPageSize();
270
Rafael Espindola9d2234d2011-03-10 20:54:07 +0000271 // Default is to map the full file.
272 if (MapSize == size_t(-1)) {
273 // If we don't know the file size, use fstat to find out. fstat on an open
274 // file descriptor is cheaper than stat on a random path.
275 if (FileSize == size_t(-1)) {
276 struct stat FileInfo;
277 // TODO: This should use fstat64 when available.
278 if (fstat(FD, &FileInfo) == -1) {
279 return error_code(errno, posix_category());
280 }
281 FileSize = FileInfo.st_size;
Chris Lattner11da4cf2008-04-01 06:05:21 +0000282 }
Rafael Espindola9d2234d2011-03-10 20:54:07 +0000283 MapSize = FileSize;
Chris Lattner11da4cf2008-04-01 06:05:21 +0000284 }
Michael J. Spencer1f6efa32010-11-29 18:16:10 +0000285
Rafael Espindola9d2234d2011-03-10 20:54:07 +0000286 if (shouldUseMmap(FD, FileSize, MapSize, Offset, RequiresNullTerminator,
Rafael Espindolaf7fdad12011-03-10 18:33:29 +0000287 PageSize)) {
288 off_t RealMapOffset = Offset & ~(PageSize - 1);
289 off_t Delta = Offset - RealMapOffset;
290 size_t RealMapSize = MapSize + Delta;
291
292 if (const char *Pages = sys::Path::MapInFilePages(FD,
293 RealMapSize,
294 RealMapOffset)) {
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000295 result.reset(GetNamedBuffer<MemoryBufferMMapFile>(
Rafael Espindolaf7fdad12011-03-10 18:33:29 +0000296 StringRef(Pages + Delta, MapSize), Filename, RequiresNullTerminator));
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000297 return success;
Chris Lattner11da4cf2008-04-01 06:05:21 +0000298 }
299 }
Evan Cheng726135a2009-02-13 07:54:34 +0000300
Rafael Espindolaf7fdad12011-03-10 18:33:29 +0000301 MemoryBuffer *Buf = MemoryBuffer::getNewUninitMemBuffer(MapSize, Filename);
Evan Cheng726135a2009-02-13 07:54:34 +0000302 if (!Buf) {
Michael J. Spencer333fb042010-12-09 17:36:48 +0000303 // Failed to create a buffer. The only way it can fail is if
304 // new(std::nothrow) returns 0.
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000305 return make_error_code(errc::not_enough_memory);
Evan Cheng726135a2009-02-13 07:54:34 +0000306 }
307
308 OwningPtr<MemoryBuffer> SB(Buf);
Chris Lattner11da4cf2008-04-01 06:05:21 +0000309 char *BufPtr = const_cast<char*>(SB->getBufferStart());
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000310
Rafael Espindolaf7fdad12011-03-10 18:33:29 +0000311 size_t BytesLeft = MapSize;
312 if (lseek(FD, Offset, SEEK_SET) == -1)
313 return error_code(errno, posix_category());
314
Chris Lattner333ffd42007-04-29 06:58:52 +0000315 while (BytesLeft) {
316 ssize_t NumRead = ::read(FD, BufPtr, BytesLeft);
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000317 if (NumRead == -1) {
318 if (errno == EINTR)
319 continue;
320 // Error while reading.
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000321 return error_code(errno, posix_category());
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000322 } else if (NumRead == 0) {
Benjamin Kramerd48c1f92010-04-01 14:39:55 +0000323 // We hit EOF early, truncate and terminate buffer.
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000324 Buf->BufferEnd = BufPtr;
Benjamin Kramerd48c1f92010-04-01 14:39:55 +0000325 *BufPtr = 0;
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000326 result.swap(SB);
327 return success;
Chris Lattner333ffd42007-04-29 06:58:52 +0000328 }
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000329 BytesLeft -= NumRead;
330 BufPtr += NumRead;
Chris Lattner333ffd42007-04-29 06:58:52 +0000331 }
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000332
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000333 result.swap(SB);
334 return success;
Chris Lattner333ffd42007-04-29 06:58:52 +0000335}
336
Chris Lattner333ffd42007-04-29 06:58:52 +0000337//===----------------------------------------------------------------------===//
338// MemoryBuffer::getSTDIN implementation.
339//===----------------------------------------------------------------------===//
340
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000341error_code MemoryBuffer::getSTDIN(OwningPtr<MemoryBuffer> &result) {
Chris Lattner333ffd42007-04-29 06:58:52 +0000342 // Read in all of the data from stdin, we cannot mmap stdin.
Daniel Dunbard65267e2009-11-10 00:43:58 +0000343 //
344 // FIXME: That isn't necessarily true, we should try to mmap stdin and
345 // fallback if it fails.
Jeff Cohen0fea8eb2007-05-07 15:21:46 +0000346 sys::Program::ChangeStdinToBinary();
Benjamin Kramer831338b2010-06-25 16:07:18 +0000347
348 const ssize_t ChunkSize = 4096*4;
349 SmallString<ChunkSize> Buffer;
350 ssize_t ReadBytes;
351 // Read into Buffer until we hit EOF.
Eli Friedman96cd7af2009-05-18 08:44:04 +0000352 do {
Benjamin Kramer831338b2010-06-25 16:07:18 +0000353 Buffer.reserve(Buffer.size() + ChunkSize);
354 ReadBytes = read(0, Buffer.end(), ChunkSize);
355 if (ReadBytes == -1) {
356 if (errno == EINTR) continue;
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000357 return error_code(errno, posix_category());
Benjamin Kramer831338b2010-06-25 16:07:18 +0000358 }
359 Buffer.set_size(Buffer.size() + ReadBytes);
360 } while (ReadBytes != 0);
Reid Spencer2372ccc2007-08-08 20:01:58 +0000361
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000362 result.reset(getMemBufferCopy(Buffer, "<stdin>"));
363 return success;
Chris Lattner333ffd42007-04-29 06:58:52 +0000364}