blob: a0c650d6820b0ed859a5f7f9c62b45f717dd8f8d [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.
Jakob Stoklund Olesen4b0e1f12011-03-10 18:01:43 +000049void MemoryBuffer::init(const char *BufStart, const char *BufEnd) {
50 assert(BufEnd[0] == 0 && "Buffer is not null terminated!");
Chris Lattner333ffd42007-04-29 06:58:52 +000051 BufferStart = BufStart;
52 BufferEnd = BufEnd;
Chris Lattner333ffd42007-04-29 06:58:52 +000053}
54
55//===----------------------------------------------------------------------===//
56// MemoryBufferMem implementation.
57//===----------------------------------------------------------------------===//
58
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000059/// CopyStringRef - Copies contents of a StringRef into a block of memory and
60/// null-terminates it.
61static void CopyStringRef(char *Memory, StringRef Data) {
62 memcpy(Memory, Data.data(), Data.size());
63 Memory[Data.size()] = 0; // Null terminate string.
64}
65
66/// GetNamedBuffer - Allocates a new MemoryBuffer with Name copied after it.
67template <typename T>
Jakob Stoklund Olesen4b0e1f12011-03-10 18:01:43 +000068static T* GetNamedBuffer(StringRef Buffer, StringRef Name) {
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000069 char *Mem = static_cast<char*>(operator new(sizeof(T) + Name.size() + 1));
70 CopyStringRef(Mem + sizeof(T), Name);
Jakob Stoklund Olesen4b0e1f12011-03-10 18:01:43 +000071 return new (Mem) T(Buffer);
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000072}
73
Chris Lattner333ffd42007-04-29 06:58:52 +000074namespace {
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000075/// MemoryBufferMem - Named MemoryBuffer pointing to a block of memory.
Chris Lattner333ffd42007-04-29 06:58:52 +000076class MemoryBufferMem : public MemoryBuffer {
Chris Lattner333ffd42007-04-29 06:58:52 +000077public:
Jakob Stoklund Olesen4b0e1f12011-03-10 18:01:43 +000078 MemoryBufferMem(StringRef InputData) {
79 init(InputData.begin(), InputData.end());
Chris Lattner333ffd42007-04-29 06:58:52 +000080 }
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000081
Chris Lattner333ffd42007-04-29 06:58:52 +000082 virtual const char *getBufferIdentifier() const {
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000083 // The name is stored after the class itself.
84 return reinterpret_cast<const char*>(this + 1);
Chris Lattner333ffd42007-04-29 06:58:52 +000085 }
86};
87}
88
89/// getMemBuffer - Open the specified memory range as a MemoryBuffer. Note
90/// that EndPtr[0] must be a null byte and be accessible!
Chris Lattner4c842dd2010-04-05 22:42:30 +000091MemoryBuffer *MemoryBuffer::getMemBuffer(StringRef InputData,
Benjamin Kramerd4d1f852010-06-25 11:50:40 +000092 StringRef BufferName) {
Jakob Stoklund Olesen4b0e1f12011-03-10 18:01:43 +000093 return GetNamedBuffer<MemoryBufferMem>(InputData, BufferName);
Chris Lattner333ffd42007-04-29 06:58:52 +000094}
95
Chris Lattner3daae272007-10-09 21:46:38 +000096/// getMemBufferCopy - Open the specified memory range as a MemoryBuffer,
97/// copying the contents and taking ownership of it. This has no requirements
98/// on EndPtr[0].
Chris Lattner4c842dd2010-04-05 22:42:30 +000099MemoryBuffer *MemoryBuffer::getMemBufferCopy(StringRef InputData,
Benjamin Kramerd4d1f852010-06-25 11:50:40 +0000100 StringRef BufferName) {
101 MemoryBuffer *Buf = getNewUninitMemBuffer(InputData.size(), BufferName);
102 if (!Buf) return 0;
103 memcpy(const_cast<char*>(Buf->getBufferStart()), InputData.data(),
104 InputData.size());
105 return Buf;
Chris Lattner3daae272007-10-09 21:46:38 +0000106}
107
Chris Lattner333ffd42007-04-29 06:58:52 +0000108/// getNewUninitMemBuffer - Allocate a new MemoryBuffer of the specified size
Benjamin Kramerd4d1f852010-06-25 11:50:40 +0000109/// that is not initialized. Note that the caller should initialize the
110/// memory allocated by this method. The memory is owned by the MemoryBuffer
111/// object.
Evan Cheng34cd4a42008-05-05 18:30:58 +0000112MemoryBuffer *MemoryBuffer::getNewUninitMemBuffer(size_t Size,
Daniel Dunbard65267e2009-11-10 00:43:58 +0000113 StringRef BufferName) {
Benjamin Kramerd4d1f852010-06-25 11:50:40 +0000114 // Allocate space for the MemoryBuffer, the data and the name. It is important
115 // that MemoryBuffer and data are aligned so PointerIntPair works with them.
116 size_t AlignedStringLen =
117 RoundUpToAlignment(sizeof(MemoryBufferMem) + BufferName.size() + 1,
118 sizeof(void*)); // TODO: Is sizeof(void*) enough?
119 size_t RealLen = AlignedStringLen + Size + 1;
120 char *Mem = static_cast<char*>(operator new(RealLen, std::nothrow));
121 if (!Mem) return 0;
122
123 // The name is stored after the class itself.
124 CopyStringRef(Mem + sizeof(MemoryBufferMem), BufferName);
125
126 // The buffer begins after the name and must be aligned.
127 char *Buf = Mem + AlignedStringLen;
128 Buf[Size] = 0; // Null terminate buffer.
129
Jakob Stoklund Olesen4b0e1f12011-03-10 18:01:43 +0000130 return new (Mem) MemoryBufferMem(StringRef(Buf, Size));
Chris Lattner333ffd42007-04-29 06:58:52 +0000131}
132
133/// getNewMemBuffer - Allocate a new MemoryBuffer of the specified size that
134/// is completely initialized to zeros. Note that the caller should
135/// initialize the memory allocated by this method. The memory is owned by
136/// the MemoryBuffer object.
Benjamin Kramerd4d1f852010-06-25 11:50:40 +0000137MemoryBuffer *MemoryBuffer::getNewMemBuffer(size_t Size, StringRef BufferName) {
Chris Lattner333ffd42007-04-29 06:58:52 +0000138 MemoryBuffer *SB = getNewUninitMemBuffer(Size, BufferName);
Evan Cheng726135a2009-02-13 07:54:34 +0000139 if (!SB) return 0;
Benjamin Kramerd4d1f852010-06-25 11:50:40 +0000140 memset(const_cast<char*>(SB->getBufferStart()), 0, Size);
Chris Lattner333ffd42007-04-29 06:58:52 +0000141 return SB;
142}
143
144
Chris Lattner2b1f1062007-11-18 18:52:28 +0000145/// getFileOrSTDIN - Open the specified file as a MemoryBuffer, or open stdin
146/// if the Filename is "-". If an error occurs, this returns null and fills
147/// in *ErrStr with a reason. If stdin is empty, this API (unlike getSTDIN)
148/// returns an empty buffer.
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000149error_code MemoryBuffer::getFileOrSTDIN(StringRef Filename,
150 OwningPtr<MemoryBuffer> &result,
151 int64_t FileSize) {
Daniel Dunbard65267e2009-11-10 00:43:58 +0000152 if (Filename == "-")
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000153 return getSTDIN(result);
154 return getFile(Filename, result, FileSize);
Chris Lattner2b1f1062007-11-18 18:52:28 +0000155}
156
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000157error_code MemoryBuffer::getFileOrSTDIN(const char *Filename,
158 OwningPtr<MemoryBuffer> &result,
159 int64_t FileSize) {
Dan Gohman60e6f3d2010-06-24 16:25:50 +0000160 if (strcmp(Filename, "-") == 0)
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000161 return getSTDIN(result);
162 return getFile(Filename, result, FileSize);
Dan Gohman60e6f3d2010-06-24 16:25:50 +0000163}
164
Chris Lattner333ffd42007-04-29 06:58:52 +0000165//===----------------------------------------------------------------------===//
Chris Lattner333ffd42007-04-29 06:58:52 +0000166// MemoryBuffer::getFile implementation.
167//===----------------------------------------------------------------------===//
168
Chris Lattner11da4cf2008-04-01 06:05:21 +0000169namespace {
170/// MemoryBufferMMapFile - This represents a file that was mapped in with the
171/// sys::Path::MapInFilePages method. When destroyed, it calls the
172/// sys::Path::UnMapFilePages method.
Benjamin Kramerd4d1f852010-06-25 11:50:40 +0000173class MemoryBufferMMapFile : public MemoryBufferMem {
Chris Lattner11da4cf2008-04-01 06:05:21 +0000174public:
Jakob Stoklund Olesen4b0e1f12011-03-10 18:01:43 +0000175 MemoryBufferMMapFile(StringRef Buffer)
176 : MemoryBufferMem(Buffer) { }
Benjamin Kramerd4d1f852010-06-25 11:50:40 +0000177
Chris Lattner11da4cf2008-04-01 06:05:21 +0000178 ~MemoryBufferMMapFile() {
Jakob Stoklund Olesen4b0e1f12011-03-10 18:01:43 +0000179 sys::Path::UnMapFilePages(getBufferStart(), getBufferSize());
Chris Lattner11da4cf2008-04-01 06:05:21 +0000180 }
181};
182}
183
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000184error_code MemoryBuffer::getFile(StringRef Filename,
185 OwningPtr<MemoryBuffer> &result,
186 int64_t FileSize) {
Chris Lattner7cf70542010-11-23 22:20:27 +0000187 // Ensure the path is null terminated.
Dan Gohman60e6f3d2010-06-24 16:25:50 +0000188 SmallString<256> PathBuf(Filename.begin(), Filename.end());
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000189 return MemoryBuffer::getFile(PathBuf.c_str(), result, FileSize);
Dan Gohman60e6f3d2010-06-24 16:25:50 +0000190}
191
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000192error_code MemoryBuffer::getFile(const char *Filename,
193 OwningPtr<MemoryBuffer> &result,
194 int64_t FileSize) {
Dan Gohman60e6f3d2010-06-24 16:25:50 +0000195 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) {
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000201 return error_code(errno, posix_category());
Chris Lattner333ffd42007-04-29 06:58:52 +0000202 }
Rafael Espindolab4cc0312011-02-08 22:40:47 +0000203 error_code ret = getOpenFile(FD, Filename, result, FileSize);
204 close(FD);
205 return ret;
Chris Lattner7cf70542010-11-23 22:20:27 +0000206}
Michael J. Spencer1f6efa32010-11-29 18:16:10 +0000207
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000208error_code MemoryBuffer::getOpenFile(int FD, const char *Filename,
209 OwningPtr<MemoryBuffer> &result,
Jakob Stoklund Olesen4b0e1f12011-03-10 18:01:43 +0000210 int64_t FileSize) {
Chris Lattner11da4cf2008-04-01 06:05:21 +0000211 // If we don't know the file size, use fstat to find out. fstat on an open
212 // file descriptor is cheaper than stat on a random path.
Jakob Stoklund Olesen4b0e1f12011-03-10 18:01:43 +0000213 if (FileSize == -1) {
Chris Lattner9f9bd8e2010-11-23 06:09:51 +0000214 struct stat FileInfo;
Chris Lattner11da4cf2008-04-01 06:05:21 +0000215 // TODO: This should use fstat64 when available.
Chris Lattner9f9bd8e2010-11-23 06:09:51 +0000216 if (fstat(FD, &FileInfo) == -1) {
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000217 return error_code(errno, posix_category());
Chris Lattner11da4cf2008-04-01 06:05:21 +0000218 }
Chris Lattner9f9bd8e2010-11-23 06:09:51 +0000219 FileSize = FileInfo.st_size;
Chris Lattner11da4cf2008-04-01 06:05:21 +0000220 }
Michael J. Spencer1f6efa32010-11-29 18:16:10 +0000221
222
Jakob Stoklund Olesen4b0e1f12011-03-10 18:01:43 +0000223 // If the file is large, try to use mmap to read it in. We don't use mmap
224 // for small files, because this can severely fragment our address space. Also
225 // don't try to map files that are exactly a multiple of the system page size,
226 // as the file would not have the required null terminator.
227 //
228 // FIXME: Can we just mmap an extra page in the latter case?
229 if (FileSize >= 4096*4 &&
230 (FileSize & (sys::Process::GetPageSize()-1)) != 0) {
231 if (const char *Pages = sys::Path::MapInFilePages(FD, FileSize)) {
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000232 result.reset(GetNamedBuffer<MemoryBufferMMapFile>(
Jakob Stoklund Olesen4b0e1f12011-03-10 18:01:43 +0000233 StringRef(Pages, FileSize), Filename));
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000234 return success;
Chris Lattner11da4cf2008-04-01 06:05:21 +0000235 }
236 }
Evan Cheng726135a2009-02-13 07:54:34 +0000237
Jakob Stoklund Olesen4b0e1f12011-03-10 18:01:43 +0000238 MemoryBuffer *Buf = MemoryBuffer::getNewUninitMemBuffer(FileSize, Filename);
Evan Cheng726135a2009-02-13 07:54:34 +0000239 if (!Buf) {
Michael J. Spencer333fb042010-12-09 17:36:48 +0000240 // Failed to create a buffer. The only way it can fail is if
241 // new(std::nothrow) returns 0.
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000242 return make_error_code(errc::not_enough_memory);
Evan Cheng726135a2009-02-13 07:54:34 +0000243 }
244
245 OwningPtr<MemoryBuffer> SB(Buf);
Chris Lattner11da4cf2008-04-01 06:05:21 +0000246 char *BufPtr = const_cast<char*>(SB->getBufferStart());
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000247
Jakob Stoklund Olesen4b0e1f12011-03-10 18:01:43 +0000248 size_t BytesLeft = FileSize;
Chris Lattner333ffd42007-04-29 06:58:52 +0000249 while (BytesLeft) {
250 ssize_t NumRead = ::read(FD, BufPtr, BytesLeft);
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000251 if (NumRead == -1) {
252 if (errno == EINTR)
253 continue;
254 // Error while reading.
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000255 return error_code(errno, posix_category());
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000256 } else if (NumRead == 0) {
Benjamin Kramerd48c1f92010-04-01 14:39:55 +0000257 // We hit EOF early, truncate and terminate buffer.
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000258 Buf->BufferEnd = BufPtr;
Benjamin Kramerd48c1f92010-04-01 14:39:55 +0000259 *BufPtr = 0;
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000260 result.swap(SB);
261 return success;
Chris Lattner333ffd42007-04-29 06:58:52 +0000262 }
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000263 BytesLeft -= NumRead;
264 BufPtr += NumRead;
Chris Lattner333ffd42007-04-29 06:58:52 +0000265 }
Benjamin Kramer6a9cd412010-04-01 14:35:22 +0000266
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000267 result.swap(SB);
268 return success;
Chris Lattner333ffd42007-04-29 06:58:52 +0000269}
270
Chris Lattner333ffd42007-04-29 06:58:52 +0000271//===----------------------------------------------------------------------===//
272// MemoryBuffer::getSTDIN implementation.
273//===----------------------------------------------------------------------===//
274
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000275error_code MemoryBuffer::getSTDIN(OwningPtr<MemoryBuffer> &result) {
Chris Lattner333ffd42007-04-29 06:58:52 +0000276 // Read in all of the data from stdin, we cannot mmap stdin.
Daniel Dunbard65267e2009-11-10 00:43:58 +0000277 //
278 // FIXME: That isn't necessarily true, we should try to mmap stdin and
279 // fallback if it fails.
Jeff Cohen0fea8eb2007-05-07 15:21:46 +0000280 sys::Program::ChangeStdinToBinary();
Benjamin Kramer831338b2010-06-25 16:07:18 +0000281
282 const ssize_t ChunkSize = 4096*4;
283 SmallString<ChunkSize> Buffer;
284 ssize_t ReadBytes;
285 // Read into Buffer until we hit EOF.
Eli Friedman96cd7af2009-05-18 08:44:04 +0000286 do {
Benjamin Kramer831338b2010-06-25 16:07:18 +0000287 Buffer.reserve(Buffer.size() + ChunkSize);
288 ReadBytes = read(0, Buffer.end(), ChunkSize);
289 if (ReadBytes == -1) {
290 if (errno == EINTR) continue;
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000291 return error_code(errno, posix_category());
Benjamin Kramer831338b2010-06-25 16:07:18 +0000292 }
293 Buffer.set_size(Buffer.size() + ReadBytes);
294 } while (ReadBytes != 0);
Reid Spencer2372ccc2007-08-08 20:01:58 +0000295
Michael J. Spencer3ff95632010-12-16 03:29:14 +0000296 result.reset(getMemBufferCopy(Buffer, "<stdin>"));
297 return success;
Chris Lattner333ffd42007-04-29 06:58:52 +0000298}