Merge pull request #391 from vgough/pr/380

Return code rework
This commit is contained in:
Valient Gough 2017-08-25 23:31:28 -07:00 committed by GitHub
commit 03d0ae793e
25 changed files with 432 additions and 251 deletions

View File

@ -5,7 +5,7 @@ matrix:
- os: linux
compiler: gcc
dist: trusty
sudo: false
sudo: required
addons:
apt:
sources:
@ -17,12 +17,12 @@ matrix:
- gettext
- cmake3
env:
- INTEGRATION=false
- SUDO_TESTS=true
- os: linux
compiler: clang
dist: trusty
sudo: true
sudo: false
addons:
apt:
sources:
@ -37,14 +37,14 @@ matrix:
- clang-4.0
- clang-tidy-4.0
env:
- CC=clang-4.0 CXX=clang++-4.0 CHECK=true INTEGRATION=true
- CC=clang-4.0 CXX=clang++-4.0 CHECK=true INTEGRATION=false CMAKE=/tmp/bin/cmake
- os: osx
compiler: clang
osx_image: xcode8.3
sudo: true
sudo: required
env:
- INTEGRATION=true
- SUDO_TESTS=true
before_script:
- ./ci/setup.sh

View File

@ -168,7 +168,21 @@ if (${CMAKE_MAJOR_VERSION}.${CMAKE_MINOR_VERSION} GREATER 3.5) # Need 3.6 or abo
message(STATUS "clang-tidy not found.")
else()
message(STATUS "clang-tidy found: ${CLANG_TIDY_EXE}")
set(DO_CLANG_TIDY "${CLANG_TIDY_EXE}" "-checks=*,-modernize-loop-convert,-cppcoreguidelines-pro-*,-readability-inconsistent-declaration-parameter-name,-google-readability-casting,-cert-err58-cpp,-google-runtime-int,-readability-named-parameter,-google-build-using-namespace,-misc-unused-parameters,-google-runtime-references")
string(CONCAT TIDY_OPTS "-checks=*"
",-cert-err58-cpp"
",-cppcoreguidelines-pro-*"
",-google-build-using-namespace"
",-google-readability-casting"
",-google-readability-todo"
",-google-runtime-int"
",-google-runtime-references"
",-misc-misplaced-widening-cast"
",-misc-unused-parameters"
",-modernize-loop-convert"
",-readability-inconsistent-declaration-parameter-name"
",-readability-named-parameter"
)
set(DO_CLANG_TIDY "${CLANG_TIDY_EXE}" ${TIDY_OPTS})
#set(DO_CLANG_TIDY "${CLANG_TIDY_EXE}" "-fix" "-checks=-*,google-readability-redundant-smartptr-get")
endif()
else()

View File

@ -1,9 +1,10 @@
#!/bin/bash -eu
: ${CMAKE:=cmake}
: ${CHECK:=false}
: ${INTEGRATION:=true}
cmake --version
${CMAKE} --version
CFG=$*
if [[ "$CHECK" == "true" ]]; then
@ -20,7 +21,7 @@ then
fi
cd build
cmake .. ${CFG}
${CMAKE} .. ${CFG}
make -j2
make test
if [[ "$INTEGRATION" == "true" ]]; then

View File

@ -1,6 +1,6 @@
#!/bin/bash -eu
: ${INTEGRATION:=false}
: ${INTEGRATION:=true}
: ${CHECK:=false}
if [[ "$INTEGRATION" == "true" ]]; then
@ -15,8 +15,8 @@ if [[ "$CHECK" == "true" ]]; then
if uname -s | grep -q Linux; then
wget https://cmake.org/files/v3.9/cmake-3.9.1-Linux-x86_64.tar.gz -O /tmp/cmake.tar.gz
tar -C /tmp/ -xf /tmp/cmake.tar.gz
sudo rm -f $(which cmake)
sudo ln -s $(ls -1 /tmp/cmake*/bin/cmake) /bin/
mkdir /tmp/bin
ln -s $(ls -1 /tmp/cmake*/bin/cmake) /tmp/bin/
fi
fi

View File

@ -69,7 +69,7 @@ ssize_t BlockFileIO::cacheReadOneBlock(const IORequest &req) const {
* the lower file may have changed behind our back. */
if ((!_noCache) && (req.offset == _cache.offset) && (_cache.dataLen != 0)) {
// satisfy request from cache
int len = req.dataLen;
size_t len = req.dataLen;
if (_cache.dataLen < len) {
len = _cache.dataLen; // Don't read past EOF
}
@ -97,17 +97,17 @@ ssize_t BlockFileIO::cacheReadOneBlock(const IORequest &req) const {
return result;
}
bool BlockFileIO::cacheWriteOneBlock(const IORequest &req) {
ssize_t BlockFileIO::cacheWriteOneBlock(const IORequest &req) {
// cache results of write (before pass-thru, because it may be modified
// in-place)
memcpy(_cache.data, req.data, req.dataLen);
_cache.offset = req.offset;
_cache.dataLen = req.dataLen;
bool ok = writeOneBlock(req);
if (!ok) {
ssize_t res = writeOneBlock(req);
if (res < 0) {
clearCache(_cache, _blockSize);
}
return ok;
return res;
}
/**
@ -116,11 +116,13 @@ bool BlockFileIO::cacheWriteOneBlock(const IORequest &req) {
* data from the front of the first block if the request is not aligned.
* Always requests aligned data of the size of one block or less from the
* lower layer.
* Returns the number of bytes read, or -errno in case of failure.
*/
ssize_t BlockFileIO::read(const IORequest &req) const {
CHECK(_blockSize != 0);
int partialOffset = req.offset % _blockSize;
int partialOffset =
req.offset % _blockSize; // can be int as _blockSize is int
off_t blockNum = req.offset / _blockSize;
ssize_t result = 0;
@ -154,11 +156,15 @@ ssize_t BlockFileIO::read(const IORequest &req) const {
}
ssize_t readSize = cacheReadOneBlock(blockReq);
if (readSize < 0) {
result = readSize;
break;
}
if (readSize <= partialOffset) {
break; // didn't get enough bytes
}
int cpySize = min((size_t)(readSize - partialOffset), size);
size_t cpySize = min((size_t)(readSize - partialOffset), size);
CHECK(cpySize <= readSize);
// if we read to a temporary buffer, then move the data
@ -184,21 +190,25 @@ ssize_t BlockFileIO::read(const IORequest &req) const {
return result;
}
bool BlockFileIO::write(const IORequest &req) {
/**
* Returns the number of bytes written, or -errno in case of failure.
*/
ssize_t BlockFileIO::write(const IORequest &req) {
CHECK(_blockSize != 0);
off_t fileSize = getSize();
if (fileSize < 0) {
return false;
return fileSize;
}
// where write request begins
off_t blockNum = req.offset / _blockSize;
int partialOffset = req.offset % _blockSize;
int partialOffset =
req.offset % _blockSize; // can be int as _blockSize is int
// last block of file (for testing write overlaps with file boundary)
off_t lastFileBlock = fileSize / _blockSize;
ssize_t lastBlockSize = fileSize % _blockSize;
size_t lastBlockSize = fileSize % _blockSize;
off_t lastNonEmptyBlock = lastFileBlock;
if (lastBlockSize == 0) {
@ -208,7 +218,10 @@ bool BlockFileIO::write(const IORequest &req) {
if (req.offset > fileSize) {
// extend file first to fill hole with 0's..
const bool forceWrite = false;
padFile(fileSize, req.offset, forceWrite);
int res = padFile(fileSize, req.offset, forceWrite);
if (res < 0) {
return res;
}
}
// check against edge cases where we can just let the base class handle the
@ -233,7 +246,7 @@ bool BlockFileIO::write(const IORequest &req) {
blockReq.data = nullptr;
blockReq.dataLen = _blockSize;
bool ok = true;
ssize_t res = 0;
size_t size = req.dataLen;
unsigned char *inPtr = req.data;
while (size != 0u) {
@ -258,11 +271,16 @@ bool BlockFileIO::write(const IORequest &req) {
if (blockNum > lastNonEmptyBlock) {
// just pad..
blockReq.dataLen = toCopy + partialOffset;
blockReq.dataLen = partialOffset + toCopy;
} else {
// have to merge with existing block data..
blockReq.dataLen = _blockSize;
blockReq.dataLen = cacheReadOneBlock(blockReq);
ssize_t readSize = cacheReadOneBlock(blockReq);
if (readSize < 0) {
res = readSize;
break;
}
blockReq.dataLen = readSize;
// extend data if necessary..
if (partialOffset + toCopy > blockReq.dataLen) {
@ -274,8 +292,8 @@ bool BlockFileIO::write(const IORequest &req) {
}
// Finally, write the damn thing!
if (!cacheWriteOneBlock(blockReq)) {
ok = false;
res = cacheWriteOneBlock(blockReq);
if (res < 0) {
break;
}
@ -290,15 +308,22 @@ bool BlockFileIO::write(const IORequest &req) {
MemoryPool::release(mb);
}
return ok;
if (res < 0) {
return res;
}
return req.dataLen;
}
int BlockFileIO::blockSize() const { return _blockSize; }
void BlockFileIO::padFile(off_t oldSize, off_t newSize, bool forceWrite) {
/**
* Returns 0 in case of success, or -errno in case of failure.
*/
int BlockFileIO::padFile(off_t oldSize, off_t newSize, bool forceWrite) {
off_t oldLastBlock = oldSize / _blockSize;
off_t newLastBlock = newSize / _blockSize;
int newBlockSize = newSize % _blockSize;
int newBlockSize = newSize % _blockSize; // can be int as _blockSize is int
ssize_t res = 0;
IORequest req;
MemBlock mb;
@ -317,9 +342,10 @@ void BlockFileIO::padFile(off_t oldSize, off_t newSize, bool forceWrite) {
if (outSize != 0) {
memset(mb.data, 0, outSize);
cacheReadOneBlock(req);
req.dataLen = outSize;
cacheWriteOneBlock(req);
if ((res = cacheReadOneBlock(req)) >= 0) {
req.dataLen = outSize;
res = cacheWriteOneBlock(req);
}
}
} else
VLOG(1) << "optimization: not padding last block";
@ -338,39 +364,48 @@ void BlockFileIO::padFile(off_t oldSize, off_t newSize, bool forceWrite) {
if (req.dataLen != 0) {
VLOG(1) << "padding block " << oldLastBlock;
memset(mb.data, 0, _blockSize);
cacheReadOneBlock(req);
req.dataLen = _blockSize; // expand to full block size
cacheWriteOneBlock(req);
if ((res = cacheReadOneBlock(req)) >= 0) {
req.dataLen = _blockSize; // expand to full block size
res = cacheWriteOneBlock(req);
}
++oldLastBlock;
}
// 2, pad zero blocks unless holes are allowed
if (!_allowHoles) {
for (; oldLastBlock != newLastBlock; ++oldLastBlock) {
for (; (res >= 0) && (oldLastBlock != newLastBlock); ++oldLastBlock) {
VLOG(1) << "padding block " << oldLastBlock;
req.offset = oldLastBlock * _blockSize;
req.dataLen = _blockSize;
memset(mb.data, 0, req.dataLen);
cacheWriteOneBlock(req);
res = cacheWriteOneBlock(req);
}
}
// 3. only necessary if write is forced and block is non 0 length
if (forceWrite && (newBlockSize != 0)) {
if ((res >= 0) && forceWrite && (newBlockSize != 0)) {
req.offset = newLastBlock * _blockSize;
req.dataLen = newBlockSize;
memset(mb.data, 0, req.dataLen);
cacheWriteOneBlock(req);
res = cacheWriteOneBlock(req);
}
}
if (mb.data != nullptr) {
MemoryPool::release(mb);
}
if (res < 0) {
return res;
}
return 0;
}
/**
* Returns 0 in case of success, or -errno in case of failure.
*/
int BlockFileIO::truncateBase(off_t size, FileIO *base) {
int partialBlock = size % _blockSize;
int partialBlock = size % _blockSize; // can be int as _blockSize is int
int res = 0;
off_t oldSize = getSize();
@ -381,11 +416,13 @@ int BlockFileIO::truncateBase(off_t size, FileIO *base) {
// do the truncate so that the underlying filesystem can allocate
// the space, and then we'll fill it in padFile..
if (base != nullptr) {
base->truncate(size);
res = base->truncate(size);
}
const bool forceWrite = true;
padFile(oldSize, size, forceWrite);
if (res == 0) {
res = padFile(oldSize, size, forceWrite);
}
} else if (size == oldSize) {
// the easiest case, but least likely....
} else if (partialBlock != 0) {
@ -400,21 +437,23 @@ int BlockFileIO::truncateBase(off_t size, FileIO *base) {
req.dataLen = _blockSize;
req.data = mb.data;
ssize_t rdSz = cacheReadOneBlock(req);
ssize_t readSize = cacheReadOneBlock(req);
if (readSize < 0) {
res = readSize;
}
// do the truncate
if (base != nullptr) {
else if (base != nullptr) {
// do the truncate
res = base->truncate(size);
}
// write back out partial block
req.dataLen = partialBlock;
bool wrRes = cacheWriteOneBlock(req);
if ((rdSz < 0) || (!wrRes)) {
// rwarning - unlikely to ever occur..
RLOG(WARNING) << "truncate failure: read " << rdSz
<< " bytes, partial block of " << partialBlock;
if (res == 0) {
ssize_t writeSize = cacheWriteOneBlock(req);
if (writeSize < 0) {
res = writeSize;
}
}
MemoryPool::release(mb);

View File

@ -43,21 +43,21 @@ class BlockFileIO : public FileIO {
// implemented in terms of blocks.
virtual ssize_t read(const IORequest &req) const;
virtual bool write(const IORequest &req);
virtual ssize_t write(const IORequest &req);
virtual int blockSize() const;
protected:
int truncateBase(off_t size, FileIO *base);
void padFile(off_t oldSize, off_t newSize, bool forceWrite);
int padFile(off_t oldSize, off_t newSize, bool forceWrite);
// same as read(), except that the request.offset field is guarenteed to be
// block aligned, and the request size will not be larger then 1 block.
virtual ssize_t readOneBlock(const IORequest &req) const = 0;
virtual bool writeOneBlock(const IORequest &req) = 0;
virtual ssize_t writeOneBlock(const IORequest &req) = 0;
ssize_t cacheReadOneBlock(const IORequest &req) const;
bool cacheWriteOneBlock(const IORequest &req);
ssize_t cacheWriteOneBlock(const IORequest &req);
int _blockSize;
bool _allowHoles;

View File

@ -163,8 +163,12 @@ int BlockNameIO::encodeName(const char *plaintextName, int length, uint64_t *iv,
encodedName[0] = (mac >> 8) & 0xff;
encodedName[1] = (mac)&0xff;
_cipher->blockEncode((unsigned char *)encodedName + 2, length + padding,
(uint64_t)mac ^ tmpIV, _key);
bool ok;
ok = _cipher->blockEncode((unsigned char *)encodedName + 2, length + padding,
(uint64_t)mac ^ tmpIV, _key);
if (!ok) {
throw Error("block encode failed in filename encode");
}
// convert to base 64 ascii
int encodedStreamLen = length + 2 + padding;
@ -219,8 +223,12 @@ int BlockNameIO::decodeName(const char *encodedName, int length, uint64_t *iv,
tmpIV = *iv;
}
_cipher->blockDecode((unsigned char *)tmpBuf + 2, decodedStreamLen,
(uint64_t)mac ^ tmpIV, _key);
bool ok;
ok = _cipher->blockDecode((unsigned char *)tmpBuf + 2, decodedStreamLen,
(uint64_t)mac ^ tmpIV, _key);
if (!ok) {
throw Error("block decode failed in filename decode");
}
// find out true string length
int padding = (unsigned char)tmpBuf[2 + decodedStreamLen - 1];

View File

@ -98,20 +98,22 @@ bool CipherFileIO::setIV(uint64_t iv) {
} else if (haveHeader) {
// we have an old IV, and now a new IV, so we need to update the fileIV
// on disk.
// ensure the file is open for read/write..
int newFlags = lastFlags | O_RDWR;
int res = base->open(newFlags);
if (res < 0) {
if (res == -EISDIR) {
// duh -- there are no file headers for directories!
externalIV = iv;
return base->setIV(iv);
}
VLOG(1) << "setIV failed to re-open for write";
return false;
}
if (fileIV == 0) {
// ensure the file is open for read/write..
int newFlags = lastFlags | O_RDWR;
int res = base->open(newFlags);
if (res < 0) {
if (res == -EISDIR) {
// duh -- there are no file headers for directories!
externalIV = iv;
return base->setIV(iv);
}
VLOG(1) << "writeHeader failed to re-open for write";
if (initHeader() < 0) {
return false;
}
initHeader();
}
uint64_t oldIV = externalIV;
@ -172,7 +174,7 @@ off_t CipherFileIO::getSize() const {
return size;
}
void CipherFileIO::initHeader() {
int CipherFileIO::initHeader() {
// check if the file has a header, and read it if it does.. Otherwise,
// create one.
off_t rawSize = base->getSize();
@ -185,9 +187,14 @@ void CipherFileIO::initHeader() {
req.offset = 0;
req.data = buf;
req.dataLen = 8;
base->read(req);
ssize_t readSize = base->read(req);
if (readSize < 0) {
return readSize;
}
cipher->streamDecode(buf, sizeof(buf), externalIV, key);
if (!cipher->streamDecode(buf, sizeof(buf), externalIV, key)) {
return -EBADMSG;
}
fileIV = 0;
for (int i = 0; i < 8; ++i) {
@ -201,7 +208,8 @@ void CipherFileIO::initHeader() {
unsigned char buf[8] = {0};
do {
if (!cipher->randomize(buf, 8, false)) {
throw Error("Unable to generate a random file IV");
RLOG(ERROR) << "Unable to generate a random file IV";
return -EBADMSG;
}
fileIV = 0;
@ -215,31 +223,28 @@ void CipherFileIO::initHeader() {
} while (fileIV == 0); // don't accept 0 as an option..
if (base->isWritable()) {
cipher->streamEncode(buf, sizeof(buf), externalIV, key);
if (!cipher->streamEncode(buf, sizeof(buf), externalIV, key)) {
return -EBADMSG;
}
IORequest req;
req.offset = 0;
req.data = buf;
req.dataLen = 8;
base->write(req);
ssize_t writeSize = base->write(req);
if (writeSize < 0) {
return writeSize;
}
} else {
VLOG(1) << "base not writable, IV not written..";
}
}
VLOG(1) << "initHeader finished, fileIV = " << fileIV;
return 0;
}
bool CipherFileIO::writeHeader() {
if (!base->isWritable()) {
// open for write..
int newFlags = lastFlags | O_RDWR;
if (base->open(newFlags) < 0) {
VLOG(1) << "writeHeader failed to re-open for write";
return false;
}
}
if (fileIV == 0) {
RLOG(ERROR) << "Internal error: fileIV == 0 in writeHeader!!!";
}
@ -251,16 +256,16 @@ bool CipherFileIO::writeHeader() {
fileIV >>= 8;
}
cipher->streamEncode(buf, sizeof(buf), externalIV, key);
if (!cipher->streamEncode(buf, sizeof(buf), externalIV, key)) {
return false;
}
IORequest req;
req.offset = 0;
req.data = buf;
req.dataLen = 8;
base->write(req);
return true;
return (base->write(req) >= 0);
}
/**
@ -275,7 +280,7 @@ bool CipherFileIO::writeHeader() {
* the IV. This guarantees unpredictability and prevents watermarking
* attacks.
*/
void CipherFileIO::generateReverseHeader(unsigned char *headerBuf) {
int CipherFileIO::generateReverseHeader(unsigned char *headerBuf) {
struct stat stbuf;
int res = getAttr(&stbuf);
@ -309,7 +314,10 @@ void CipherFileIO::generateReverseHeader(unsigned char *headerBuf) {
VLOG(1) << "fileIV=" << fileIV;
// Encrypt externally-visible header
cipher->streamEncode(headerBuf, HEADER_SIZE, externalIV, key);
if (!cipher->streamEncode(headerBuf, HEADER_SIZE, externalIV, key)) {
return -EBADMSG;
}
return 0;
}
/**
@ -321,76 +329,90 @@ ssize_t CipherFileIO::readOneBlock(const IORequest &req) const {
int bs = blockSize();
off_t blockNum = req.offset / bs;
ssize_t readSize = 0;
IORequest tmpReq = req;
// adjust offset if we have a file header
if (haveHeader && !fsConfig->reverseEncryption) {
tmpReq.offset += HEADER_SIZE;
}
readSize = base->read(tmpReq);
ssize_t readSize = base->read(tmpReq);
bool ok;
if (readSize > 0) {
if (haveHeader && fileIV == 0) {
const_cast<CipherFileIO *>(this)->initHeader();
int res = const_cast<CipherFileIO *>(this)->initHeader();
if (res < 0) {
return res;
}
}
if (readSize != bs) {
VLOG(1) << "streamRead(data, " << readSize << ", IV)";
ok = streamRead(tmpReq.data, (int)readSize, blockNum ^ fileIV);
ok = streamRead(tmpReq.data, (int)readSize,
blockNum ^ fileIV); // cast works because we work on a
// block and blocksize fit an int
} else {
ok = blockRead(tmpReq.data, (int)readSize, blockNum ^ fileIV);
ok = blockRead(tmpReq.data, (int)readSize,
blockNum ^ fileIV); // cast works because we work on a
// block and blocksize fit an int
}
if (!ok) {
VLOG(1) << "decodeBlock failed for block " << blockNum << ", size "
<< readSize;
readSize = -1;
readSize = -EBADMSG;
}
} else {
} else if (readSize == 0) {
VLOG(1) << "readSize zero for offset " << req.offset;
}
return readSize;
}
bool CipherFileIO::writeOneBlock(const IORequest &req) {
ssize_t CipherFileIO::writeOneBlock(const IORequest &req) {
if (haveHeader && fsConfig->reverseEncryption) {
VLOG(1)
<< "writing to a reverse mount with per-file IVs is not implemented";
return false;
return -EPERM;
}
int bs = blockSize();
off_t blockNum = req.offset / bs;
if (haveHeader && fileIV == 0) {
initHeader();
int res = initHeader();
if (res < 0) {
return res;
}
}
bool ok;
if (req.dataLen != bs) {
ok = streamWrite(req.data, (int)req.dataLen, blockNum ^ fileIV);
ok = streamWrite(req.data, (int)req.dataLen,
blockNum ^ fileIV); // cast works because we work on a
// block and blocksize fit an int
} else {
ok = blockWrite(req.data, (int)req.dataLen, blockNum ^ fileIV);
ok = blockWrite(req.data, (int)req.dataLen,
blockNum ^ fileIV); // cast works because we work on a
// block and blocksize fit an int
}
ssize_t res = 0;
if (ok) {
if (haveHeader) {
IORequest tmpReq = req;
tmpReq.offset += HEADER_SIZE;
ok = base->write(tmpReq);
res = base->write(tmpReq);
} else {
ok = base->write(req);
res = base->write(req);
}
} else {
VLOG(1) << "encodeBlock failed for block " << blockNum << ", size "
<< req.dataLen;
ok = false;
res = -EBADMSG;
}
return ok;
return res;
}
bool CipherFileIO::blockWrite(unsigned char *buf, int size,
@ -439,27 +461,38 @@ bool CipherFileIO::streamRead(unsigned char *buf, int size,
int CipherFileIO::truncate(off_t size) {
int res = 0;
int reopen = 0;
// well, we will truncate, so we need a write access to the file
if (!base->isWritable()) {
int newFlags = lastFlags | O_RDWR;
int res = base->open(newFlags);
if (res < 0) {
VLOG(1) << "truncate failed to re-open for write";
base->open(lastFlags);
return res;
}
reopen = 1;
}
if (!haveHeader) {
res = BlockFileIO::truncateBase(size, base.get());
} else {
if (0 == fileIV) {
// empty file.. create the header..
if (!base->isWritable()) {
// open for write..
int newFlags = lastFlags | O_RDWR;
if (base->open(newFlags) < 0) {
VLOG(1) << "writeHeader failed to re-open for write";
}
}
initHeader();
res = initHeader();
}
// can't let BlockFileIO call base->truncate(), since it would be using
// the wrong size..
res = BlockFileIO::truncateBase(size, nullptr);
if (res == 0) {
base->truncate(size + HEADER_SIZE);
res = BlockFileIO::truncateBase(size, nullptr);
}
if (res == 0) {
res = base->truncate(size + HEADER_SIZE);
}
}
if (reopen == 1) {
reopen = base->open(lastFlags);
if (res < 0) {
res = reopen;
}
}
return res;
@ -484,7 +517,10 @@ ssize_t CipherFileIO::read(const IORequest &origReq) const {
// generate the file IV header
// this is needed in any case - without IV the file cannot be decoded
unsigned char headerBuf[HEADER_SIZE];
const_cast<CipherFileIO *>(this)->generateReverseHeader(headerBuf);
int res = const_cast<CipherFileIO *>(this)->generateReverseHeader(headerBuf);
if (res < 0) {
return res;
}
// Copy the request so we can modify it without affecting the caller
IORequest req = origReq;
@ -505,7 +541,8 @@ ssize_t CipherFileIO::read(const IORequest &origReq) const {
VLOG(1) << "Adding " << headerBytes << " header bytes";
// copy the header bytes into the data
int headerOffset = HEADER_SIZE - headerBytes;
int headerOffset =
HEADER_SIZE - headerBytes; // can be int as HEADER_SIZE is int
memcpy(req.data, &headerBuf[headerOffset], headerBytes);
// the read does not want data beyond the header
@ -528,7 +565,8 @@ ssize_t CipherFileIO::read(const IORequest &origReq) const {
if (readBytes < 0) {
return readBytes; // Return error code
}
ssize_t sum = headerBytes + readBytes;
ssize_t sum =
headerBytes + readBytes; // could be size_t, but as we return ssize_t...
VLOG(1) << "returning sum=" << sum;
return sum;
}

View File

@ -65,10 +65,10 @@ class CipherFileIO : public BlockFileIO {
private:
virtual ssize_t readOneBlock(const IORequest &req) const;
virtual bool writeOneBlock(const IORequest &req);
virtual void generateReverseHeader(unsigned char *data);
virtual ssize_t writeOneBlock(const IORequest &req);
virtual int generateReverseHeader(unsigned char *data);
void initHeader();
int initHeader();
bool writeHeader();
bool blockRead(unsigned char *buf, int size, uint64_t iv64) const;
bool streamRead(unsigned char *buf, int size, uint64_t iv64) const;

View File

@ -111,7 +111,8 @@ void EncFS_Context::renameNode(const char *from, const char *to) {
// putNode stores "node" under key "path" in the "openFiles" map. It
// increments the reference count if the key already exists.
void EncFS_Context::putNode(const char *path, std::shared_ptr<FileNode> node) {
void EncFS_Context::putNode(const char *path,
const std::shared_ptr<FileNode> &node) {
Lock lock(contextMutex);
auto &list = openFiles[std::string(path)];
// The length of "list" serves as the reference count.
@ -122,7 +123,7 @@ void EncFS_Context::putNode(const char *path, std::shared_ptr<FileNode> node) {
// eraseNode is called by encfs_release in response to the RELEASE
// FUSE-command we get from the kernel.
void EncFS_Context::eraseNode(const char *path,
std::shared_ptr<FileNode> fnode) {
const std::shared_ptr<FileNode> &fnode) {
Lock lock(contextMutex);
auto it = openFiles.find(std::string(path));
@ -151,7 +152,7 @@ void EncFS_Context::eraseNode(const char *path,
// nextFuseFh returns the next unused uint64 to serve as the FUSE file
// handle for the kernel.
uint64_t EncFS_Context::nextFuseFh(void) {
uint64_t EncFS_Context::nextFuseFh() {
// This is thread-safe because currentFuseFh is declared as std::atomic
return currentFuseFh++;
}

View File

@ -48,9 +48,9 @@ class EncFS_Context {
void getAndResetUsageCounter(int *usage, int *openCount);
void putNode(const char *path, std::shared_ptr<FileNode> node);
void putNode(const char *path, const std::shared_ptr<FileNode> &node);
void eraseNode(const char *path, std::shared_ptr<FileNode> fnode);
void eraseNode(const char *path, const std::shared_ptr<FileNode> &fnode);
void renameNode(const char *oldName, const char *newName);

View File

@ -149,7 +149,7 @@ class RenameOp {
~RenameOp();
operator bool() const { return renameList != nullptr; }
explicit operator bool() const { return renameList != nullptr; }
bool apply();
void undo();
@ -181,8 +181,9 @@ bool RenameOp::apply() {
// rename on disk..
if (::rename(last->oldCName.c_str(), last->newCName.c_str()) == -1) {
int eno = errno;
RLOG(WARNING) << "Error renaming " << last->oldCName << ": "
<< strerror(errno);
<< strerror(eno);
dn->renameNode(last->newPName.c_str(), last->oldPName.c_str(), false);
return false;
}
@ -351,7 +352,8 @@ DirTraverse DirNode::openDir(const char *plaintextPath) {
DIR *dir = ::opendir(cyName.c_str());
if (dir == nullptr) {
VLOG(1) << "opendir error " << strerror(errno);
int eno = errno;
VLOG(1) << "opendir error " << strerror(eno);
return DirTraverse(shared_ptr<DIR>(), 0, std::shared_ptr<NameIO>());
}
std::shared_ptr<DIR> dp(dir, DirDeleter());
@ -584,8 +586,7 @@ int DirNode::rename(const char *fromPlaintext, const char *toPlaintext) {
}
if (res != 0) {
VLOG(1) << "rename failed: " << strerror(errno);
res = -errno;
VLOG(1) << "rename failed: " << strerror(-res);
}
return res;
@ -726,7 +727,7 @@ int DirNode::unlink(const char *plaintextName) {
res = ::unlink(fullName.c_str());
if (res == -1) {
res = -errno;
VLOG(1) << "unlink error: " << strerror(errno);
VLOG(1) << "unlink error: " << strerror(-res);
}
}

View File

@ -34,7 +34,7 @@ struct IORequest {
off_t offset;
// amount of bytes to read/write.
int dataLen;
size_t dataLen;
unsigned char *data;
IORequest();
@ -68,7 +68,7 @@ class FileIO {
virtual off_t getSize() const = 0;
virtual ssize_t read(const IORequest &req) const = 0;
virtual bool write(const IORequest &req) = 0;
virtual ssize_t write(const IORequest &req) = 0;
virtual int truncate(off_t size) = 0;

View File

@ -157,14 +157,16 @@ int FileNode::mknod(mode_t mode, dev_t rdev, uid_t uid, gid_t gid) {
if (uid != 0) {
olduid = setfsuid(uid);
if (olduid == -1) {
RLOG(DEBUG) << "setfsuid error: " << strerror(errno);
int eno = errno;
RLOG(DEBUG) << "setfsuid error: " << strerror(eno);
return -EPERM;
}
}
if (gid != 0) {
oldgid = setfsgid(gid);
if (oldgid == -1) {
RLOG(DEBUG) << "setfsgid error: " << strerror(errno);
int eno = errno;
RLOG(DEBUG) << "setfsgid error: " << strerror(eno);
return -EPERM;
}
}
@ -185,6 +187,12 @@ int FileNode::mknod(mode_t mode, dev_t rdev, uid_t uid, gid_t gid) {
res = ::mknod(_cname.c_str(), mode, rdev);
}
if (res == -1) {
int eno = errno;
VLOG(1) << "mknod error: " << strerror(eno);
res = -eno;
}
if (olduid >= 0) {
setfsuid(olduid);
}
@ -192,12 +200,6 @@ int FileNode::mknod(mode_t mode, dev_t rdev, uid_t uid, gid_t gid) {
setfsgid(oldgid);
}
if (res == -1) {
int eno = errno;
VLOG(1) << "mknod error: " << strerror(eno);
res = -eno;
}
return res;
}
@ -222,7 +224,7 @@ off_t FileNode::getSize() const {
return res;
}
ssize_t FileNode::read(off_t offset, unsigned char *data, ssize_t size) const {
ssize_t FileNode::read(off_t offset, unsigned char *data, size_t size) const {
IORequest req;
req.offset = offset;
req.dataLen = size;
@ -233,7 +235,7 @@ ssize_t FileNode::read(off_t offset, unsigned char *data, ssize_t size) const {
return io->read(req);
}
bool FileNode::write(off_t offset, unsigned char *data, ssize_t size) {
ssize_t FileNode::write(off_t offset, unsigned char *data, size_t size) {
VLOG(1) << "FileNode::write offset " << offset << ", data size " << size;
IORequest req;
@ -243,7 +245,12 @@ bool FileNode::write(off_t offset, unsigned char *data, ssize_t size) {
Lock _lock(mutex);
return io->write(req);
ssize_t res = io->write(req);
// Of course due to encryption we genrally write more than requested
if (res < 0) {
return res;
}
return size;
}
int FileNode::truncate(off_t size) {
@ -266,8 +273,6 @@ int FileNode::sync(bool datasync) {
}
#else
(void)datasync;
// no fdatasync support
// TODO: use autoconfig to check for it..
res = fsync(fh);
#endif

View File

@ -79,8 +79,8 @@ class FileNode {
int getAttr(struct stat *stbuf) const;
off_t getSize() const;
ssize_t read(off_t offset, unsigned char *data, ssize_t size) const;
bool write(off_t offset, unsigned char *data, ssize_t size);
ssize_t read(off_t offset, unsigned char *data, size_t size) const;
ssize_t write(off_t offset, unsigned char *data, size_t size);
// truncate the file to a particular size
int truncate(off_t size);

View File

@ -147,7 +147,7 @@ off_t MACFileIO::getSize() const {
ssize_t MACFileIO::readOneBlock(const IORequest &req) const {
int headerSize = macBytes + randBytes;
int bs = blockSize() + headerSize;
int bs = blockSize() + headerSize; // ok, should clearly fit into an int
MemBlock mb = MemoryPool::allocate(bs);
@ -194,7 +194,7 @@ ssize_t MACFileIO::readOneBlock(const IORequest &req) const {
RLOG(WARNING) << "MAC comparison failure in block " << blockNum;
if (!warnOnly) {
MemoryPool::release(mb);
throw Error(_("MAC comparison failure, refusing to read"));
return -EBADMSG;
}
}
}
@ -214,10 +214,10 @@ ssize_t MACFileIO::readOneBlock(const IORequest &req) const {
return readSize;
}
bool MACFileIO::writeOneBlock(const IORequest &req) {
ssize_t MACFileIO::writeOneBlock(const IORequest &req) {
int headerSize = macBytes + randBytes;
int bs = blockSize() + headerSize;
int bs = blockSize() + headerSize; // ok, should clearly fit into an int
// we have the unencrypted data, so we need to attach a header to it.
MemBlock mb = MemoryPool::allocate(bs);
@ -231,7 +231,7 @@ bool MACFileIO::writeOneBlock(const IORequest &req) {
memcpy(newReq.data + headerSize, req.data, req.dataLen);
if (randBytes > 0) {
if (!cipher->randomize(newReq.data + macBytes, randBytes, false)) {
return false;
return -EBADMSG;
}
}
@ -247,21 +247,21 @@ bool MACFileIO::writeOneBlock(const IORequest &req) {
}
// now, we can let the next level have it..
bool ok = base->write(newReq);
ssize_t writeSize = base->write(newReq);
MemoryPool::release(mb);
return ok;
return writeSize;
}
int MACFileIO::truncate(off_t size) {
int headerSize = macBytes + randBytes;
int bs = blockSize() + headerSize;
int bs = blockSize() + headerSize; // ok, should clearly fit into an int
int res = BlockFileIO::truncateBase(size, nullptr);
if (res == 0) {
base->truncate(locWithHeader(size, bs, headerSize));
res = base->truncate(locWithHeader(size, bs, headerSize));
}
return res;

View File

@ -64,7 +64,7 @@ class MACFileIO : public BlockFileIO {
private:
virtual ssize_t readOneBlock(const IORequest &req) const;
virtual bool writeOneBlock(const IORequest &req);
virtual ssize_t writeOneBlock(const IORequest &req);
std::shared_ptr<FileIO> base;
std::shared_ptr<Cipher> cipher;

View File

@ -90,6 +90,8 @@ Interface RawFileIO::interface() const { return RawFileIO_iface; }
This works around the problem described in
https://github.com/vgough/encfs/issues/181
Without this, "umask 0777 ; echo foo > bar" fails.
Sets errno when -1 is returned.
*/
static int open_readonly_workaround(const char *path, int flags) {
int fd = -1;
@ -97,13 +99,11 @@ static int open_readonly_workaround(const char *path, int flags) {
memset(&stbuf, 0, sizeof(struct stat));
if (lstat(path, &stbuf) != -1) {
// make sure user has read/write permission..
chmod(path, stbuf.st_mode | 0600);
fd = ::open(path, flags);
chmod(path, stbuf.st_mode);
} else {
RLOG(INFO) << "can't stat file " << path;
if (chmod(path, stbuf.st_mode | 0600) != -1) {
fd = ::open(path, flags);
chmod(path, stbuf.st_mode);
}
}
return fd;
}
@ -120,50 +120,53 @@ int RawFileIO::open(int flags) {
bool requestWrite = (((flags & O_RDWR) != 0) || ((flags & O_WRONLY) != 0));
VLOG(1) << "open call, requestWrite = " << requestWrite;
int result = 0;
// if we have a descriptor and it is writable, or we don't need writable..
if ((fd >= 0) && (canWrite || !requestWrite)) {
VLOG(1) << "using existing file descriptor";
result = fd; // success
} else {
int finalFlags = requestWrite ? O_RDWR : O_RDONLY;
return fd; // success
}
int finalFlags = requestWrite ? O_RDWR : O_RDONLY;
#if defined(O_LARGEFILE)
if ((flags & O_LARGEFILE) != 0) {
finalFlags |= O_LARGEFILE;
}
if ((flags & O_LARGEFILE) != 0) {
finalFlags |= O_LARGEFILE;
}
#else
#warning O_LARGEFILE not supported
#endif
int newFd = ::open(name.c_str(), finalFlags);
VLOG(1) << "open file with flags " << finalFlags << ", result = " << newFd;
if ((newFd == -1) && (errno == EACCES)) {
VLOG(1) << "using readonly workaround for open";
newFd = open_readonly_workaround(name.c_str(), finalFlags);
}
if (newFd >= 0) {
if (oldfd >= 0) {
RLOG(ERROR) << "leaking FD?: oldfd = " << oldfd << ", fd = " << fd
<< ", newfd = " << newFd;
}
// the old fd might still be in use, so just keep it around for
// now.
canWrite = requestWrite;
oldfd = fd;
result = fd = newFd;
} else {
result = -errno;
RLOG(DEBUG) << "::open error: " << strerror(errno);
}
int eno;
int newFd = ::open(name.c_str(), finalFlags);
if (newFd < 0) {
eno = errno;
}
return result;
VLOG(1) << "open file with flags " << finalFlags << ", result = " << newFd;
if ((newFd == -1) && (eno == EACCES)) {
VLOG(1) << "using readonly workaround for open";
newFd = open_readonly_workaround(name.c_str(), finalFlags);
eno = errno;
}
if (newFd < 0) {
RLOG(DEBUG) << "::open error: " << strerror(eno);
return -eno;
}
if (oldfd >= 0) {
RLOG(ERROR) << "leaking FD?: oldfd = " << oldfd << ", fd = " << fd
<< ", newfd = " << newFd;
}
// the old fd might still be in use, so just keep it around for
// now.
canWrite = requestWrite;
oldfd = fd;
fd = newFd;
return fd;
}
int RawFileIO::getAttr(struct stat *stbuf) const {
@ -192,8 +195,9 @@ off_t RawFileIO::getSize() const {
const_cast<RawFileIO *>(this)->knownSize = true;
return fileSize;
}
RLOG(ERROR) << "getSize on " << name << " failed: " << strerror(errno);
return -1;
int eno = errno;
RLOG(ERROR) << "getSize on " << name << " failed: " << strerror(eno);
return -eno;
}
return fileSize;
}
@ -204,44 +208,59 @@ ssize_t RawFileIO::read(const IORequest &req) const {
ssize_t readSize = pread(fd, req.data, req.dataLen, req.offset);
if (readSize < 0) {
int eno = errno;
RLOG(WARNING) << "read failed at offset " << req.offset << " for "
<< req.dataLen << " bytes: " << strerror(errno);
<< req.dataLen << " bytes: " << strerror(eno);
return -eno;
}
return readSize;
}
bool RawFileIO::write(const IORequest &req) {
ssize_t RawFileIO::write(const IORequest &req) {
rAssert(fd >= 0);
rAssert(canWrite);
int retrys = 10;
// int retrys = 10;
void *buf = req.data;
ssize_t bytes = req.dataLen;
off_t offset = req.offset;
while ((bytes != 0) && retrys > 0) {
/*
* Let's write while pwrite() writes, to avoid writing only a part of the
* request,
* whereas it could have been fully written. This to avoid inconsistencies /
* corruption.
*/
// while ((bytes != 0) && retrys > 0) {
while (bytes != 0) {
ssize_t writeSize = ::pwrite(fd, buf, bytes, offset);
if (writeSize < 0) {
int eno = errno;
knownSize = false;
RLOG(WARNING) << "write failed at offset " << offset << " for " << bytes
<< " bytes: " << strerror(errno);
return false;
<< " bytes: " << strerror(eno);
// pwrite is not expected to return 0, so eno should always be set, but we
// never know...
return -eno;
}
if (writeSize == 0) {
return -EIO;
}
bytes -= writeSize;
offset += writeSize;
buf = (void *)((char *)buf + writeSize);
--retrys;
}
if (bytes != 0) {
RLOG(ERROR) << "Write error: wrote " << req.dataLen - bytes << " bytes of "
<< req.dataLen << ", max retries reached";
knownSize = false;
return false;
}
// if (bytes != 0) {
// RLOG(ERROR) << "Write error: wrote " << req.dataLen - bytes << " bytes of
// "
// << req.dataLen << ", max retries reached";
// knownSize = false;
// return (eno) ? -eno : -EIO;
// }
if (knownSize) {
off_t last = req.offset + req.dataLen;
if (last > fileSize) {
@ -249,7 +268,7 @@ bool RawFileIO::write(const IORequest &req) {
}
}
return true;
return req.dataLen;
}
int RawFileIO::truncate(off_t size) {

View File

@ -46,7 +46,7 @@ class RawFileIO : public FileIO {
virtual off_t getSize() const;
virtual ssize_t read(const IORequest &req) const;
virtual bool write(const IORequest &req);
virtual ssize_t write(const IORequest &req);
virtual int truncate(off_t size);

View File

@ -106,7 +106,7 @@ int BytesToKey(int keyLen, int ivLen, const EVP_MD *md,
memcpy(iv, mdBuf + offset, toCopy);
iv += toCopy;
niv -= toCopy;
offset += toCopy;
// offset += toCopy;
}
if ((nkey == 0) && (niv == 0)) {
break;
@ -170,12 +170,14 @@ static Range CAMELLIABlockRange(64, 4096, 16);
static std::shared_ptr<Cipher> NewCAMELLIACipher(const Interface &iface,
int keyLen) {
if (keyLen <= 0) keyLen = 192;
if (keyLen <= 0) {
keyLen = 192;
}
keyLen = CAMELLIAKeyRange.closest(keyLen);
const EVP_CIPHER *blockCipher = 0;
const EVP_CIPHER *streamCipher = 0;
const EVP_CIPHER *blockCipher = nullptr;
const EVP_CIPHER *streamCipher = nullptr;
switch (keyLen) {
case 128:
@ -503,7 +505,7 @@ CipherKey SSL_Cipher::newRandomKey() {
compute a 64-bit check value for the data using HMAC.
*/
static uint64_t _checksum_64(SSLKey *key, const unsigned char *data,
int dataLen, uint64_t *chainedIV) {
int dataLen, uint64_t *const chainedIV) {
rAssert(dataLen > 0);
Lock lock(key->mutex);
@ -810,6 +812,7 @@ bool SSL_Cipher::streamEncode(unsigned char *buf, int size, uint64_t iv64,
if (dstLen != size) {
RLOG(ERROR) << "encoding " << size << " bytes, got back " << dstLen << " ("
<< tmpLen << " in final_ex)";
return false;
}
return true;
@ -846,6 +849,7 @@ bool SSL_Cipher::streamDecode(unsigned char *buf, int size, uint64_t iv64,
if (dstLen != size) {
RLOG(ERROR) << "decoding " << size << " bytes, got back " << dstLen << " ("
<< tmpLen << " in final_ex)";
return false;
}
return true;
@ -861,7 +865,8 @@ bool SSL_Cipher::blockEncode(unsigned char *buf, int size, uint64_t iv64,
// data must be integer number of blocks
const int blockMod = size % EVP_CIPHER_CTX_block_size(key->block_enc);
if (blockMod != 0) {
throw Error("Invalid data size, not multiple of block size");
RLOG(ERROR) << "Invalid data size, not multiple of block size";
return false;
}
Lock lock(key->mutex);
@ -879,6 +884,7 @@ bool SSL_Cipher::blockEncode(unsigned char *buf, int size, uint64_t iv64,
if (dstLen != size) {
RLOG(ERROR) << "encoding " << size << " bytes, got back " << dstLen << " ("
<< tmpLen << " in final_ex)";
return false;
}
return true;
@ -894,7 +900,8 @@ bool SSL_Cipher::blockDecode(unsigned char *buf, int size, uint64_t iv64,
// data must be integer number of blocks
const int blockMod = size % EVP_CIPHER_CTX_block_size(key->block_dec);
if (blockMod != 0) {
throw Error("Invalid data size, not multiple of block size");
RLOG(ERROR) << "Invalid data size, not multiple of block size";
return false;
}
Lock lock(key->mutex);
@ -912,6 +919,7 @@ bool SSL_Cipher::blockDecode(unsigned char *buf, int size, uint64_t iv64,
if (dstLen != size) {
RLOG(ERROR) << "decoding " << size << " bytes, got back " << dstLen << " ("
<< tmpLen << " in final_ex)";
return false;
}
return true;

View File

@ -103,7 +103,7 @@ bool XmlValue::readB64(const char *path, unsigned char *data,
std::string s = value->text();
s.erase(std::remove_if(s.begin(), s.end(), ::isspace), s.end());
s.erase(s.find_last_not_of("=") + 1);
s.erase(s.find_last_not_of('=') + 1);
int decodedSize = B64ToB256Bytes(s.size());
if (decodedSize != length) {

View File

@ -25,6 +25,7 @@
#include <cstring>
#include <ctime>
#include <fcntl.h>
#include <limits>
#include <memory>
#include <sys/stat.h>
#include <sys/statvfs.h>
@ -64,7 +65,7 @@ using namespace std::placeholders;
namespace encfs {
#define GET_FN(ctx, finfo) ctx->getNode((void *)(uintptr_t)finfo->fh)
#define GET_FN(ctx, finfo) (ctx)->getNode((void *)(uintptr_t)(finfo)->fh)
static EncFS_Context *context() {
return (EncFS_Context *)fuse_get_context()->private_data;
@ -78,9 +79,10 @@ static EncFS_Context *context() {
static bool isReadOnly(EncFS_Context *ctx) { return ctx->opts->readOnly; }
// helper function -- apply a functor to a cipher path, given the plain path
static int withCipherPath(const char *opName, const char *path,
function<int(EncFS_Context *, const string &)> op,
bool passReturnCode = false) {
static int withCipherPath(
const char *opName, const char *path,
const function<int(EncFS_Context *, const string &)> &op,
bool passReturnCode = false) {
EncFS_Context *ctx = context();
int res = -EIO;
@ -109,7 +111,7 @@ static int withCipherPath(const char *opName, const char *path,
return res;
}
static void checkCanary(std::shared_ptr<FileNode> fnode) {
static void checkCanary(const std::shared_ptr<FileNode> &fnode) {
if (fnode->canary == CANARY_OK) {
return;
}
@ -408,11 +410,7 @@ int _do_readlink(EncFS_Context *ctx, const string &cyName, char *buf,
buf[res] = '\0'; // ensure null termination
string decodedName;
try {
decodedName = FSRoot->plainPath(buf);
} catch (...) {
VLOG(1) << "caught error decoding path";
}
decodedName = FSRoot->plainPath(buf);
if (!decodedName.empty()) {
strncpy(buf, decodedName.c_str(), size - 1);
@ -690,12 +688,18 @@ int encfs_release(const char *path, struct fuse_file_info *finfo) {
}
}
int _do_read(FileNode *fnode, unsigned char *ptr, size_t size, off_t off) {
ssize_t _do_read(FileNode *fnode, unsigned char *ptr, size_t size, off_t off) {
return fnode->read(off, ptr, size);
}
int encfs_read(const char *path, char *buf, size_t size, off_t offset,
struct fuse_file_info *file) {
// Unfortunately we have to convert from ssize_t (pread) to int (fuse), so
// let's check this will be OK
if (size > std::numeric_limits<int>::max()) {
RLOG(ERROR) << "tried to read too much data: " << size;
return -EIO;
}
return withFileNode("read", path, file,
bind(_do_read, _1, (unsigned char *)buf, size, offset));
}
@ -712,15 +716,19 @@ int encfs_fsync(const char *path, int dataSync, struct fuse_file_info *file) {
return withFileNode("fsync", path, file, bind(_do_fsync, _1, dataSync));
}
int _do_write(FileNode *fnode, unsigned char *ptr, size_t size, off_t offset) {
if (fnode->write(offset, ptr, size)) {
return size;
}
return -EIO;
ssize_t _do_write(FileNode *fnode, unsigned char *ptr, size_t size,
off_t offset) {
return fnode->write(offset, ptr, size);
}
int encfs_write(const char *path, const char *buf, size_t size, off_t offset,
struct fuse_file_info *file) {
// Unfortunately we have to convert from ssize_t (pwrite) to int (fuse), so
// let's check this will be OK
if (size > std::numeric_limits<int>::max()) {
RLOG(ERROR) << "tried to write too much data: " << size;
return -EIO;
}
EncFS_Context *ctx = context();
if (isReadOnly(ctx)) {
return -EROFS;

View File

@ -41,7 +41,8 @@ static __inline int setfsuid(uid_t uid) {
uid_t olduid = geteuid();
if (seteuid(uid) != 0) {
VLOG(1) << "seteuid error: " << errno;
int eno = errno;
VLOG(1) << "seteuid error: " << strerror(eno);
}
return olduid;
@ -51,7 +52,8 @@ static __inline int setfsgid(gid_t gid) {
gid_t oldgid = getegid();
if (setegid(gid) != 0) {
VLOG(1) << "setfsgid error: " << errno;
int eno = errno;
VLOG(1) << "setfsgid error: " << strerror(eno);
}
return oldgid;

View File

@ -534,7 +534,7 @@ void *encfs_init(fuse_conn_info *conn) {
if (res != 0) {
RLOG(ERROR) << "error starting idle monitor thread, "
"res = "
<< res << ", errno = " << errno;
<< res << ", " << strerror(res);
}
}

View File

@ -2,7 +2,7 @@
# Test EncFS normal and paranoid mode
use Test::More tests => 122;
use Test::More tests => 132;
use File::Path;
use File::Copy;
use File::Temp;
@ -78,6 +78,8 @@ sub runTests
&grow;
&umask0777;
&create_unmount_remount;
&checkReadError;
&checkWriteError;
&configFromPipe;
&cleanup;
@ -109,7 +111,7 @@ sub corruption
ok( open(IN, "< $crypt/corrupt"), "open corrupted file");
my $content;
$result = read(IN, $content, 20);
ok(! defined $result, "corrupted file with MAC returns read error: $!");
ok($!{EBADMSG} && (! defined $result), "corrupted file with MAC returns read error: $!");
}
# Test internal modification
@ -430,3 +432,38 @@ sub create_unmount_remount
portable_unmount($mnt);
}
# Test that read errors are correctly thrown up to us
sub checkReadError
{
# Not sure how to implement this, so feel free !
ok(1, "read error");
}
# Test that write errors are correctly thrown up to us
sub checkWriteError
{
# No OSX impl, and requires sudo which is inconvenient outside of CI.
if($^O eq "darwin" || !defined($ENV{'SUDO_TESTS'})) {
ok(1, "write error");
ok(1, "write error");
ok(1, "write error");
ok(1, "write error");
}
else {
my $crypt = "$workingDir/checkWriteError.crypt";
my $mnt = "$workingDir/checkWriteError.mnt";
mkdir($crypt) || BAIL_OUT($!);
mkdir($mnt) || BAIL_OUT($!);
system("sudo mount -t tmpfs -o size=1m tmpfs $crypt");
ok( $? == 0, "mount command returns 0") || return;
system("./build/encfs --standard --extpass=\"echo test\" $crypt $mnt 2>&1");
ok( $? == 0, "encfs command returns 0") || return;
ok(open(OUT , "> $mnt/file"), "write content");
while(print OUT "0123456789") {}
ok ($!{ENOSPC}, "write returned $! instead of ENOSPC");
close OUT;
portable_unmount($mnt);
system("sudo umount $crypt");
}
}