blob: 2b95089b5246f4620fb63fb354adf47fdfef6578 [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 == "-")
140 return getSTDIN();
Douglas Gregor1bb30b62010-03-15 20:32:14 +0000141 return getFile(Filename, ErrStr, FileSize, FileInfo);
Chris Lattner2b1f1062007-11-18 18:52:28 +0000142}
143
Chris Lattner333ffd42007-04-29 06:58:52 +0000144//===----------------------------------------------------------------------===//
Chris Lattner333ffd42007-04-29 06:58:52 +0000145// MemoryBuffer::getFile implementation.
146//===----------------------------------------------------------------------===//
147
Chris Lattner11da4cf2008-04-01 06:05:21 +0000148namespace {
149/// MemoryBufferMMapFile - This represents a file that was mapped in with the
150/// sys::Path::MapInFilePages method. When destroyed, it calls the
151/// sys::Path::UnMapFilePages method.
152class MemoryBufferMMapFile : public MemoryBuffer {
153 std::string Filename;
154public:
Daniel Dunbard65267e2009-11-10 00:43:58 +0000155 MemoryBufferMMapFile(StringRef filename, const char *Pages, uint64_t Size)
Chris Lattner11da4cf2008-04-01 06:05:21 +0000156 : Filename(filename) {
157 init(Pages, Pages+Size);
158 }
159
160 virtual const char *getBufferIdentifier() const {
161 return Filename.c_str();
162 }
163
164 ~MemoryBufferMMapFile() {
165 sys::Path::UnMapFilePages(getBufferStart(), getBufferSize());
166 }
167};
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000168
169/// FileCloser - RAII object to make sure an FD gets closed properly.
170class FileCloser {
171 int FD;
172public:
173 FileCloser(int FD) : FD(FD) {}
174 ~FileCloser() { ::close(FD); }
175};
Chris Lattner11da4cf2008-04-01 06:05:21 +0000176}
177
Daniel Dunbard65267e2009-11-10 00:43:58 +0000178MemoryBuffer *MemoryBuffer::getFile(StringRef Filename, std::string *ErrStr,
Douglas Gregor1bb30b62010-03-15 20:32:14 +0000179 int64_t FileSize, struct stat *FileInfo) {
Chris Lattner11da4cf2008-04-01 06:05:21 +0000180 int OpenFlags = 0;
181#ifdef O_BINARY
Bill Wendlinga4420062008-04-01 22:09:20 +0000182 OpenFlags |= O_BINARY; // Open input file in binary mode on win32.
Chris Lattner11da4cf2008-04-01 06:05:21 +0000183#endif
Benjamin Krameraabc26c2010-02-26 20:28:29 +0000184 SmallString<256> PathBuf(Filename.begin(), Filename.end());
185 int FD = ::open(PathBuf.c_str(), O_RDONLY|OpenFlags);
Chris Lattner333ffd42007-04-29 06:58:52 +0000186 if (FD == -1) {
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000187 if (ErrStr) *ErrStr = sys::StrError();
Chris Lattner333ffd42007-04-29 06:58:52 +0000188 return 0;
189 }
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000190 FileCloser FC(FD); // Close FD on return.
Chris Lattner333ffd42007-04-29 06:58:52 +0000191
Chris Lattner11da4cf2008-04-01 06:05:21 +0000192 // If we don't know the file size, use fstat to find out. fstat on an open
193 // file descriptor is cheaper than stat on a random path.
Douglas Gregor1bb30b62010-03-15 20:32:14 +0000194 if (FileSize == -1 || FileInfo) {
195 struct stat MyFileInfo;
196 struct stat *FileInfoPtr = FileInfo? FileInfo : &MyFileInfo;
197
Chris Lattner11da4cf2008-04-01 06:05:21 +0000198 // TODO: This should use fstat64 when available.
Douglas Gregor1bb30b62010-03-15 20:32:14 +0000199 if (fstat(FD, FileInfoPtr) == -1) {
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000200 if (ErrStr) *ErrStr = sys::StrError();
Chris Lattner11da4cf2008-04-01 06:05:21 +0000201 return 0;
202 }
Douglas Gregor1bb30b62010-03-15 20:32:14 +0000203 FileSize = FileInfoPtr->st_size;
Chris Lattner11da4cf2008-04-01 06:05:21 +0000204 }
205
206
207 // If the file is large, try to use mmap to read it in. We don't use mmap
208 // for small files, because this can severely fragment our address space. Also
209 // don't try to map files that are exactly a multiple of the system page size,
210 // as the file would not have the required null terminator.
Daniel Dunbard65267e2009-11-10 00:43:58 +0000211 //
212 // FIXME: Can we just mmap an extra page in the latter case?
Chris Lattner11da4cf2008-04-01 06:05:21 +0000213 if (FileSize >= 4096*4 &&
214 (FileSize & (sys::Process::GetPageSize()-1)) != 0) {
215 if (const char *Pages = sys::Path::MapInFilePages(FD, FileSize)) {
216 // Close the file descriptor, now that the whole file is in memory.
Chris Lattner038112a2008-04-01 18:04:03 +0000217 return new MemoryBufferMMapFile(Filename, Pages, FileSize);
Chris Lattner11da4cf2008-04-01 06:05:21 +0000218 }
219 }
Evan Cheng726135a2009-02-13 07:54:34 +0000220
221 MemoryBuffer *Buf = MemoryBuffer::getNewUninitMemBuffer(FileSize, Filename);
222 if (!Buf) {
223 // Failed to create a buffer.
224 if (ErrStr) *ErrStr = "could not allocate buffer";
Evan Cheng726135a2009-02-13 07:54:34 +0000225 return 0;
226 }
227
228 OwningPtr<MemoryBuffer> SB(Buf);
Chris Lattner11da4cf2008-04-01 06:05:21 +0000229 char *BufPtr = const_cast<char*>(SB->getBufferStart());
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000230
Evan Cheng34cd4a42008-05-05 18:30:58 +0000231 size_t BytesLeft = FileSize;
Chris Lattner333ffd42007-04-29 06:58:52 +0000232 while (BytesLeft) {
233 ssize_t NumRead = ::read(FD, BufPtr, BytesLeft);
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000234 if (NumRead == -1) {
235 if (errno == EINTR)
236 continue;
237 // Error while reading.
238 if (ErrStr) *ErrStr = sys::StrError();
Chris Lattner333ffd42007-04-29 06:58:52 +0000239 return 0;
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000240 } else if (NumRead == 0) {
Benjamin Kramerd48c1f92010-04-01 14:39:55 +0000241 // We hit EOF early, truncate and terminate buffer.
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000242 Buf->BufferEnd = BufPtr;
Benjamin Kramerd48c1f92010-04-01 14:39:55 +0000243 *BufPtr = 0;
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000244 return SB.take();
Chris Lattner333ffd42007-04-29 06:58:52 +0000245 }
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000246 BytesLeft -= NumRead;
247 BufPtr += NumRead;
Chris Lattner333ffd42007-04-29 06:58:52 +0000248 }
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000249
Chris Lattner11da4cf2008-04-01 06:05:21 +0000250 return SB.take();
Chris Lattner333ffd42007-04-29 06:58:52 +0000251}
252
Chris Lattner333ffd42007-04-29 06:58:52 +0000253//===----------------------------------------------------------------------===//
254// MemoryBuffer::getSTDIN implementation.
255//===----------------------------------------------------------------------===//
256
257namespace {
258class STDINBufferFile : public MemoryBuffer {
259public:
260 virtual const char *getBufferIdentifier() const {
261 return "<stdin>";
262 }
263};
264}
265
266MemoryBuffer *MemoryBuffer::getSTDIN() {
267 char Buffer[4096*4];
Eli Friedman96cd7af2009-05-18 08:44:04 +0000268
Chris Lattner333ffd42007-04-29 06:58:52 +0000269 std::vector<char> FileData;
Eli Friedman96cd7af2009-05-18 08:44:04 +0000270
Chris Lattner333ffd42007-04-29 06:58:52 +0000271 // Read in all of the data from stdin, we cannot mmap stdin.
Daniel Dunbard65267e2009-11-10 00:43:58 +0000272 //
273 // FIXME: That isn't necessarily true, we should try to mmap stdin and
274 // fallback if it fails.
Jeff Cohen0fea8eb2007-05-07 15:21:46 +0000275 sys::Program::ChangeStdinToBinary();
Eli Friedman96cd7af2009-05-18 08:44:04 +0000276 size_t ReadBytes;
277 do {
278 ReadBytes = fread(Buffer, sizeof(char), sizeof(Buffer), stdin);
Chris Lattner333ffd42007-04-29 06:58:52 +0000279 FileData.insert(FileData.end(), Buffer, Buffer+ReadBytes);
Eli Friedman96cd7af2009-05-18 08:44:04 +0000280 } while (ReadBytes == sizeof(Buffer));
Reid Spencer2372ccc2007-08-08 20:01:58 +0000281
Nick Lewyckyea332942007-07-01 03:06:30 +0000282 FileData.push_back(0); // &FileData[Size] is invalid. So is &*FileData.end().
Chris Lattner333ffd42007-04-29 06:58:52 +0000283 size_t Size = FileData.size();
284 MemoryBuffer *B = new STDINBufferFile();
Nick Lewyckyea332942007-07-01 03:06:30 +0000285 B->initCopyOf(&FileData[0], &FileData[Size-1]);
Chris Lattner333ffd42007-04-29 06:58:52 +0000286 return B;
287}