blob: b1a767c35a21822c194ded8be2204c7eab35cb6a [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 Kramer6a9cd412010-04-01 14:35:22 +000017#include "llvm/System/Errno.h"
Chris Lattner11da4cf2008-04-01 06:05:21 +000018#include "llvm/System/Path.h"
Chris Lattner333ffd42007-04-29 06:58:52 +000019#include "llvm/System/Process.h"
Jeff Cohen0fea8eb2007-05-07 15:21:46 +000020#include "llvm/System/Program.h"
Jeff Cohen9bc40602007-04-29 14:21:44 +000021#include <cassert>
Chris Lattner333ffd42007-04-29 06:58:52 +000022#include <cstdio>
23#include <cstring>
24#include <cerrno>
Chris Lattner11da4cf2008-04-01 06:05:21 +000025#include <sys/types.h>
26#include <sys/stat.h>
27#if !defined(_MSC_VER) && !defined(__MINGW32__)
28#include <unistd.h>
29#include <sys/uio.h>
Chris Lattner11da4cf2008-04-01 06:05:21 +000030#else
31#include <io.h>
32#endif
Gabor Greif88110802008-04-30 08:53:22 +000033#include <fcntl.h>
Chris Lattner333ffd42007-04-29 06:58:52 +000034using namespace llvm;
35
36//===----------------------------------------------------------------------===//
37// MemoryBuffer implementation itself.
38//===----------------------------------------------------------------------===//
39
40MemoryBuffer::~MemoryBuffer() {
41 if (MustDeleteBuffer)
Evan Chenge2b3fdd2009-02-13 08:24:55 +000042 free((void*)BufferStart);
Chris Lattner333ffd42007-04-29 06:58:52 +000043}
44
45/// initCopyOf - Initialize this source buffer with a copy of the specified
46/// memory range. We make the copy so that we can null terminate it
47/// successfully.
48void MemoryBuffer::initCopyOf(const char *BufStart, const char *BufEnd) {
49 size_t Size = BufEnd-BufStart;
Chris Lattner62dc8962009-12-23 23:03:24 +000050 BufferStart = (char *)malloc(Size+1);
Chris Lattner333ffd42007-04-29 06:58:52 +000051 BufferEnd = BufferStart+Size;
52 memcpy(const_cast<char*>(BufferStart), BufStart, Size);
53 *const_cast<char*>(BufferEnd) = 0; // Null terminate buffer.
Chris Lattnera24b04e2007-05-11 00:43:26 +000054 MustDeleteBuffer = true;
Chris Lattner333ffd42007-04-29 06:58:52 +000055}
56
57/// init - Initialize this MemoryBuffer as a reference to externally allocated
58/// memory, memory that we know is already null terminated.
59void MemoryBuffer::init(const char *BufStart, const char *BufEnd) {
60 assert(BufEnd[0] == 0 && "Buffer is not null terminated!");
61 BufferStart = BufStart;
62 BufferEnd = BufEnd;
63 MustDeleteBuffer = false;
64}
65
66//===----------------------------------------------------------------------===//
67// MemoryBufferMem implementation.
68//===----------------------------------------------------------------------===//
69
70namespace {
71class MemoryBufferMem : public MemoryBuffer {
72 std::string FileID;
73public:
Chris Lattner4c842dd2010-04-05 22:42:30 +000074 MemoryBufferMem(StringRef InputData, StringRef FID, bool Copy = false)
Chris Lattner333ffd42007-04-29 06:58:52 +000075 : FileID(FID) {
Chris Lattner3daae272007-10-09 21:46:38 +000076 if (!Copy)
Chris Lattner4c842dd2010-04-05 22:42:30 +000077 init(InputData.data(), InputData.data()+InputData.size());
Chris Lattner3daae272007-10-09 21:46:38 +000078 else
Chris Lattner4c842dd2010-04-05 22:42:30 +000079 initCopyOf(InputData.data(), InputData.data()+InputData.size());
Chris Lattner333ffd42007-04-29 06:58:52 +000080 }
81
82 virtual const char *getBufferIdentifier() const {
83 return FileID.c_str();
84 }
85};
86}
87
88/// getMemBuffer - Open the specified memory range as a MemoryBuffer. Note
89/// that EndPtr[0] must be a null byte and be accessible!
Chris Lattner4c842dd2010-04-05 22:42:30 +000090MemoryBuffer *MemoryBuffer::getMemBuffer(StringRef InputData,
Chris Lattner333ffd42007-04-29 06:58:52 +000091 const char *BufferName) {
Chris Lattner4c842dd2010-04-05 22:42:30 +000092 return new MemoryBufferMem(InputData, BufferName);
Chris Lattner333ffd42007-04-29 06:58:52 +000093}
94
Chris Lattner3daae272007-10-09 21:46:38 +000095/// getMemBufferCopy - Open the specified memory range as a MemoryBuffer,
96/// copying the contents and taking ownership of it. This has no requirements
97/// on EndPtr[0].
Chris Lattner4c842dd2010-04-05 22:42:30 +000098MemoryBuffer *MemoryBuffer::getMemBufferCopy(StringRef InputData,
Chris Lattner3daae272007-10-09 21:46:38 +000099 const char *BufferName) {
Chris Lattner4c842dd2010-04-05 22:42:30 +0000100 return new MemoryBufferMem(InputData, BufferName, true);
Chris Lattner3daae272007-10-09 21:46:38 +0000101}
102
Chris Lattner333ffd42007-04-29 06:58:52 +0000103/// getNewUninitMemBuffer - Allocate a new MemoryBuffer of the specified size
104/// that is completely initialized to zeros. Note that the caller should
105/// initialize the memory allocated by this method. The memory is owned by
106/// the MemoryBuffer object.
Evan Cheng34cd4a42008-05-05 18:30:58 +0000107MemoryBuffer *MemoryBuffer::getNewUninitMemBuffer(size_t Size,
Daniel Dunbard65267e2009-11-10 00:43:58 +0000108 StringRef BufferName) {
Chris Lattner62dc8962009-12-23 23:03:24 +0000109 char *Buf = (char *)malloc(Size+1);
Evan Cheng726135a2009-02-13 07:54:34 +0000110 if (!Buf) return 0;
Chris Lattner333ffd42007-04-29 06:58:52 +0000111 Buf[Size] = 0;
Chris Lattner4c842dd2010-04-05 22:42:30 +0000112 MemoryBufferMem *SB = new MemoryBufferMem(StringRef(Buf, Size), BufferName);
Chris Lattner333ffd42007-04-29 06:58:52 +0000113 // The memory for this buffer is owned by the MemoryBuffer.
114 SB->MustDeleteBuffer = true;
115 return SB;
116}
117
118/// getNewMemBuffer - Allocate a new MemoryBuffer of the specified size that
119/// is completely initialized to zeros. Note that the caller should
120/// initialize the memory allocated by this method. The memory is owned by
121/// the MemoryBuffer object.
Evan Cheng34cd4a42008-05-05 18:30:58 +0000122MemoryBuffer *MemoryBuffer::getNewMemBuffer(size_t Size,
Chris Lattner333ffd42007-04-29 06:58:52 +0000123 const char *BufferName) {
124 MemoryBuffer *SB = getNewUninitMemBuffer(Size, BufferName);
Evan Cheng726135a2009-02-13 07:54:34 +0000125 if (!SB) return 0;
Chris Lattner333ffd42007-04-29 06:58:52 +0000126 memset(const_cast<char*>(SB->getBufferStart()), 0, Size+1);
127 return SB;
128}
129
130
Chris Lattner2b1f1062007-11-18 18:52:28 +0000131/// getFileOrSTDIN - Open the specified file as a MemoryBuffer, or open stdin
132/// if the Filename is "-". If an error occurs, this returns null and fills
133/// in *ErrStr with a reason. If stdin is empty, this API (unlike getSTDIN)
134/// returns an empty buffer.
Daniel Dunbard65267e2009-11-10 00:43:58 +0000135MemoryBuffer *MemoryBuffer::getFileOrSTDIN(StringRef Filename,
Chris Lattner2b1f1062007-11-18 18:52:28 +0000136 std::string *ErrStr,
Douglas Gregor1bb30b62010-03-15 20:32:14 +0000137 int64_t FileSize,
138 struct stat *FileInfo) {
Daniel Dunbard65267e2009-11-10 00:43:58 +0000139 if (Filename == "-")
Dan Gohman30377e72010-05-27 17:31:51 +0000140 return getSTDIN(ErrStr);
Douglas Gregor1bb30b62010-03-15 20:32:14 +0000141 return getFile(Filename, ErrStr, FileSize, FileInfo);
Chris Lattner2b1f1062007-11-18 18:52:28 +0000142}
143
Dan Gohman60e6f3d2010-06-24 16:25:50 +0000144MemoryBuffer *MemoryBuffer::getFileOrSTDIN(const char *Filename,
145 std::string *ErrStr,
146 int64_t FileSize,
147 struct stat *FileInfo) {
148 if (strcmp(Filename, "-") == 0)
149 return getSTDIN(ErrStr);
150 return getFile(Filename, ErrStr, FileSize, FileInfo);
151}
152
Chris Lattner333ffd42007-04-29 06:58:52 +0000153//===----------------------------------------------------------------------===//
Chris Lattner333ffd42007-04-29 06:58:52 +0000154// MemoryBuffer::getFile implementation.
155//===----------------------------------------------------------------------===//
156
Chris Lattner11da4cf2008-04-01 06:05:21 +0000157namespace {
158/// MemoryBufferMMapFile - This represents a file that was mapped in with the
159/// sys::Path::MapInFilePages method. When destroyed, it calls the
160/// sys::Path::UnMapFilePages method.
161class MemoryBufferMMapFile : public MemoryBuffer {
162 std::string Filename;
163public:
Daniel Dunbard65267e2009-11-10 00:43:58 +0000164 MemoryBufferMMapFile(StringRef filename, const char *Pages, uint64_t Size)
Chris Lattner11da4cf2008-04-01 06:05:21 +0000165 : Filename(filename) {
166 init(Pages, Pages+Size);
167 }
168
169 virtual const char *getBufferIdentifier() const {
170 return Filename.c_str();
171 }
172
173 ~MemoryBufferMMapFile() {
174 sys::Path::UnMapFilePages(getBufferStart(), getBufferSize());
175 }
176};
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000177
178/// FileCloser - RAII object to make sure an FD gets closed properly.
179class FileCloser {
180 int FD;
181public:
Dan Gohman455164b2010-06-22 13:53:29 +0000182 explicit FileCloser(int FD) : FD(FD) {}
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000183 ~FileCloser() { ::close(FD); }
184};
Chris Lattner11da4cf2008-04-01 06:05:21 +0000185}
186
Daniel Dunbard65267e2009-11-10 00:43:58 +0000187MemoryBuffer *MemoryBuffer::getFile(StringRef Filename, std::string *ErrStr,
Douglas Gregor1bb30b62010-03-15 20:32:14 +0000188 int64_t FileSize, struct stat *FileInfo) {
Dan Gohman60e6f3d2010-06-24 16:25:50 +0000189 SmallString<256> PathBuf(Filename.begin(), Filename.end());
190 return MemoryBuffer::getFile(PathBuf.c_str(), ErrStr, FileSize, FileInfo);
191}
192
193MemoryBuffer *MemoryBuffer::getFile(const char *Filename, std::string *ErrStr,
194 int64_t FileSize, struct stat *FileInfo) {
195 int OpenFlags = O_RDONLY;
Chris Lattner11da4cf2008-04-01 06:05:21 +0000196#ifdef O_BINARY
Bill Wendlinga4420062008-04-01 22:09:20 +0000197 OpenFlags |= O_BINARY; // Open input file in binary mode on win32.
Chris Lattner11da4cf2008-04-01 06:05:21 +0000198#endif
Dan Gohman60e6f3d2010-06-24 16:25:50 +0000199 int FD = ::open(Filename, OpenFlags);
Chris Lattner333ffd42007-04-29 06:58:52 +0000200 if (FD == -1) {
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000201 if (ErrStr) *ErrStr = sys::StrError();
Chris Lattner333ffd42007-04-29 06:58:52 +0000202 return 0;
203 }
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000204 FileCloser FC(FD); // Close FD on return.
Chris Lattner333ffd42007-04-29 06:58:52 +0000205
Chris Lattner11da4cf2008-04-01 06:05:21 +0000206 // If we don't know the file size, use fstat to find out. fstat on an open
207 // file descriptor is cheaper than stat on a random path.
Douglas Gregor1bb30b62010-03-15 20:32:14 +0000208 if (FileSize == -1 || FileInfo) {
209 struct stat MyFileInfo;
210 struct stat *FileInfoPtr = FileInfo? FileInfo : &MyFileInfo;
211
Chris Lattner11da4cf2008-04-01 06:05:21 +0000212 // TODO: This should use fstat64 when available.
Douglas Gregor1bb30b62010-03-15 20:32:14 +0000213 if (fstat(FD, FileInfoPtr) == -1) {
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000214 if (ErrStr) *ErrStr = sys::StrError();
Chris Lattner11da4cf2008-04-01 06:05:21 +0000215 return 0;
216 }
Douglas Gregor1bb30b62010-03-15 20:32:14 +0000217 FileSize = FileInfoPtr->st_size;
Chris Lattner11da4cf2008-04-01 06:05:21 +0000218 }
219
220
221 // If the file is large, try to use mmap to read it in. We don't use mmap
222 // for small files, because this can severely fragment our address space. Also
223 // don't try to map files that are exactly a multiple of the system page size,
224 // as the file would not have the required null terminator.
Daniel Dunbard65267e2009-11-10 00:43:58 +0000225 //
226 // FIXME: Can we just mmap an extra page in the latter case?
Chris Lattner11da4cf2008-04-01 06:05:21 +0000227 if (FileSize >= 4096*4 &&
228 (FileSize & (sys::Process::GetPageSize()-1)) != 0) {
229 if (const char *Pages = sys::Path::MapInFilePages(FD, FileSize)) {
230 // Close the file descriptor, now that the whole file is in memory.
Chris Lattner038112a2008-04-01 18:04:03 +0000231 return new MemoryBufferMMapFile(Filename, Pages, FileSize);
Chris Lattner11da4cf2008-04-01 06:05:21 +0000232 }
233 }
Evan Cheng726135a2009-02-13 07:54:34 +0000234
235 MemoryBuffer *Buf = MemoryBuffer::getNewUninitMemBuffer(FileSize, Filename);
236 if (!Buf) {
237 // Failed to create a buffer.
238 if (ErrStr) *ErrStr = "could not allocate buffer";
Evan Cheng726135a2009-02-13 07:54:34 +0000239 return 0;
240 }
241
242 OwningPtr<MemoryBuffer> SB(Buf);
Chris Lattner11da4cf2008-04-01 06:05:21 +0000243 char *BufPtr = const_cast<char*>(SB->getBufferStart());
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000244
Evan Cheng34cd4a42008-05-05 18:30:58 +0000245 size_t BytesLeft = FileSize;
Chris Lattner333ffd42007-04-29 06:58:52 +0000246 while (BytesLeft) {
247 ssize_t NumRead = ::read(FD, BufPtr, BytesLeft);
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000248 if (NumRead == -1) {
249 if (errno == EINTR)
250 continue;
251 // Error while reading.
252 if (ErrStr) *ErrStr = sys::StrError();
Chris Lattner333ffd42007-04-29 06:58:52 +0000253 return 0;
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000254 } else if (NumRead == 0) {
Benjamin Kramerd48c1f92010-04-01 14:39:55 +0000255 // We hit EOF early, truncate and terminate buffer.
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000256 Buf->BufferEnd = BufPtr;
Benjamin Kramerd48c1f92010-04-01 14:39:55 +0000257 *BufPtr = 0;
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000258 return SB.take();
Chris Lattner333ffd42007-04-29 06:58:52 +0000259 }
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000260 BytesLeft -= NumRead;
261 BufPtr += NumRead;
Chris Lattner333ffd42007-04-29 06:58:52 +0000262 }
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000263
Chris Lattner11da4cf2008-04-01 06:05:21 +0000264 return SB.take();
Chris Lattner333ffd42007-04-29 06:58:52 +0000265}
266
Chris Lattner333ffd42007-04-29 06:58:52 +0000267//===----------------------------------------------------------------------===//
268// MemoryBuffer::getSTDIN implementation.
269//===----------------------------------------------------------------------===//
270
271namespace {
272class STDINBufferFile : public MemoryBuffer {
273public:
274 virtual const char *getBufferIdentifier() const {
275 return "<stdin>";
276 }
277};
278}
279
Dan Gohman30377e72010-05-27 17:31:51 +0000280MemoryBuffer *MemoryBuffer::getSTDIN(std::string *ErrStr) {
Chris Lattner333ffd42007-04-29 06:58:52 +0000281 char Buffer[4096*4];
Eli Friedman96cd7af2009-05-18 08:44:04 +0000282
Chris Lattner333ffd42007-04-29 06:58:52 +0000283 std::vector<char> FileData;
Eli Friedman96cd7af2009-05-18 08:44:04 +0000284
Chris Lattner333ffd42007-04-29 06:58:52 +0000285 // Read in all of the data from stdin, we cannot mmap stdin.
Daniel Dunbard65267e2009-11-10 00:43:58 +0000286 //
287 // FIXME: That isn't necessarily true, we should try to mmap stdin and
288 // fallback if it fails.
Jeff Cohen0fea8eb2007-05-07 15:21:46 +0000289 sys::Program::ChangeStdinToBinary();
Eli Friedman96cd7af2009-05-18 08:44:04 +0000290 size_t ReadBytes;
291 do {
292 ReadBytes = fread(Buffer, sizeof(char), sizeof(Buffer), stdin);
Chris Lattner333ffd42007-04-29 06:58:52 +0000293 FileData.insert(FileData.end(), Buffer, Buffer+ReadBytes);
Eli Friedman96cd7af2009-05-18 08:44:04 +0000294 } while (ReadBytes == sizeof(Buffer));
Reid Spencer2372ccc2007-08-08 20:01:58 +0000295
Dan Gohman30377e72010-05-27 17:31:51 +0000296 if (!feof(stdin)) {
297 if (ErrStr) *ErrStr = "error reading from stdin";
298 return 0;
299 }
300
Nick Lewyckyea332942007-07-01 03:06:30 +0000301 FileData.push_back(0); // &FileData[Size] is invalid. So is &*FileData.end().
Chris Lattner333ffd42007-04-29 06:58:52 +0000302 size_t Size = FileData.size();
303 MemoryBuffer *B = new STDINBufferFile();
Nick Lewyckyea332942007-07-01 03:06:30 +0000304 B->initCopyOf(&FileData[0], &FileData[Size-1]);
Chris Lattner333ffd42007-04-29 06:58:52 +0000305 return B;
306}