/* A handle to something
(C) 2015-2019 Niall Douglas (11 commits)
File Created: Dec 2015
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License in the accompanying file
Licence.txt or at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
Distributed under the Boost Software License, Version 1.0.
(See accompanying file Licence.txt or copy at
http://www.boost.org/LICENSE_1_0.txt)
*/
#include "../../../io_handle.hpp"
#include "import.hpp"
#include // for IOV_MAX
#include
#include
#include // for preadv etc
#include
#include "quickcpplib/signal_guard.hpp"
LLFIO_V2_NAMESPACE_BEGIN
constexpr inline void _check_iovec_match()
{
static_assert(sizeof(io_handle::buffer_type) == sizeof(iovec), "buffer_type and struct iovec do not match in size");
static_assert(offsetof(io_handle::buffer_type, _data) == offsetof(iovec, iov_base), "buffer_type and struct iovec do not have same offset of data member");
static_assert(offsetof(io_handle::buffer_type, _len) == offsetof(iovec, iov_len), "buffer_type and struct iovec do not have same offset of len member");
}
size_t io_handle::_do_max_buffers() const noexcept
{
static size_t v;
if(v == 0u)
{
#ifdef __APPLE__
v = 1;
#else
long r = sysconf(_SC_IOV_MAX);
if(r == -1)
{
#ifdef IOV_MAX
r = IOV_MAX;
#else
r = 1;
#endif
}
v = r;
#endif
}
return v;
}
io_handle::io_result io_handle::_do_read(io_handle::io_request reqs, deadline d) noexcept
{
LLFIO_LOG_FUNCTION_CALL(this);
if(d && !_v.is_nonblocking())
{
return errc::not_supported;
}
if(reqs.buffers.size() > IOV_MAX)
{
return errc::argument_list_too_long;
}
LLFIO_POSIX_DEADLINE_TO_SLEEP_INIT(d);
#if 0
struct iovec *iov = (struct iovec *) alloca(reqs.buffers.size() * sizeof(struct iovec));
for(size_t n = 0; n < reqs.buffers.size(); n++)
{
iov[n].iov_base = reqs.buffers[n].data;
iov[n].iov_len = reqs.buffers[n].len;
}
#else
auto *iov = reinterpret_cast(reqs.buffers.data());
#endif
#ifndef NDEBUG
if(_v.requires_aligned_io())
{
assert((reqs.offset & 511) == 0);
for(size_t n = 0; n < reqs.buffers.size(); n++)
{
assert((reinterpret_cast(iov[n].iov_base) & 511) == 0);
assert((iov[n].iov_len & 511) == 0);
}
}
#endif
ssize_t bytesread = 0;
if(is_seekable())
{
#if LLFIO_MISSING_PIOV
off_t offset = reqs.offset;
for(size_t n = 0; n < reqs.buffers.size(); n++)
{
bytesread += ::pread(_v.fd, iov[n].iov_base, iov[n].iov_len, offset);
offset += iov[n].iov_len;
}
#else
bytesread = ::preadv(_v.fd, iov, reqs.buffers.size(), reqs.offset);
#endif
if(bytesread < 0)
{
return posix_error();
}
}
else
{
do
{
bytesread = ::readv(_v.fd, iov, reqs.buffers.size());
if(bytesread <= 0)
{
if(bytesread < 0 && EWOULDBLOCK != errno && EAGAIN != errno)
{
return posix_error();
}
if(!d || !d.steady || d.nsecs != 0)
{
LLFIO_POSIX_DEADLINE_TO_SLEEP_LOOP(d);
int mstimeout = (timeout == nullptr) ? -1 : (timeout->tv_sec * 1000 + timeout->tv_nsec / 1000000LL);
pollfd p;
memset(&p, 0, sizeof(p));
p.fd = _v.fd;
p.events = POLLIN | POLLERR;
if(-1 == ::poll(&p, 1, mstimeout))
{
return posix_error();
}
}
LLFIO_POSIX_DEADLINE_TO_TIMEOUT_LOOP(d);
}
} while(bytesread <= 0);
}
for(size_t i = 0; i < reqs.buffers.size(); i++)
{
auto &buffer = reqs.buffers[i];
if(buffer.size() <= static_cast(bytesread))
{
bytesread -= buffer.size();
}
else
{
buffer = {buffer.data(), (size_type) bytesread};
reqs.buffers = {reqs.buffers.data(), i + 1};
break;
}
}
return {reqs.buffers};
}
io_handle::io_result io_handle::_do_write(io_handle::io_request reqs, deadline d) noexcept
{
LLFIO_LOG_FUNCTION_CALL(this);
if(d && !_v.is_nonblocking())
{
return errc::not_supported;
}
if(reqs.buffers.size() > IOV_MAX)
{
return errc::argument_list_too_long;
}
LLFIO_POSIX_DEADLINE_TO_SLEEP_INIT(d);
#if 0
struct iovec *iov = (struct iovec *) alloca(reqs.buffers.size() * sizeof(struct iovec));
for(size_t n = 0; n < reqs.buffers.size(); n++)
{
iov[n].iov_base = const_cast(reqs.buffers[n].data);
iov[n].iov_len = reqs.buffers[n].len;
}
#else
auto *iov = reinterpret_cast(reqs.buffers.data());
#endif
#ifndef NDEBUG
if(_v.requires_aligned_io())
{
assert((reqs.offset & 511) == 0);
for(size_t n = 0; n < reqs.buffers.size(); n++)
{
assert((reinterpret_cast(iov[n].iov_base) & 511) == 0);
assert((iov[n].iov_len & 511) == 0);
}
}
#endif
ssize_t byteswritten = 0;
if(is_seekable())
{
#if LLFIO_MISSING_PIOV
off_t offset = reqs.offset;
for(size_t n = 0; n < reqs.buffers.size(); n++)
{
byteswritten += ::pwrite(_v.fd, iov[n].iov_base, iov[n].iov_len, offset);
offset += iov[n].iov_len;
}
#else
byteswritten = ::pwritev(_v.fd, iov, reqs.buffers.size(), reqs.offset);
#endif
if(byteswritten < 0)
{
return posix_error();
}
}
else
{
do
{
// Can't guarantee that user code hasn't enabled SIGPIPE
byteswritten = QUICKCPPLIB_NAMESPACE::signal_guard::signal_guard(
QUICKCPPLIB_NAMESPACE::signal_guard::signalc_set::broken_pipe, [&] { return ::writev(_v.fd, iov, reqs.buffers.size()); },
[&](const QUICKCPPLIB_NAMESPACE::signal_guard::raised_signal_info * /*unused*/) {
errno = EPIPE;
return -1;
});
if(byteswritten <= 0)
{
if(byteswritten < 0 && EWOULDBLOCK != errno && EAGAIN != errno)
{
return posix_error();
}
if(!d || !d.steady || d.nsecs != 0)
{
LLFIO_POSIX_DEADLINE_TO_SLEEP_LOOP(d);
int mstimeout = (timeout == nullptr) ? -1 : (timeout->tv_sec * 1000 + timeout->tv_nsec / 1000000LL);
pollfd p;
memset(&p, 0, sizeof(p));
p.fd = _v.fd;
p.events = POLLOUT | POLLERR;
if(-1 == ::poll(&p, 1, mstimeout))
{
return posix_error();
}
}
LLFIO_POSIX_DEADLINE_TO_TIMEOUT_LOOP(d);
}
} while(byteswritten <= 0);
}
for(size_t i = 0; i < reqs.buffers.size(); i++)
{
auto &buffer = reqs.buffers[i];
if(buffer.size() <= static_cast(byteswritten))
{
byteswritten -= buffer.size();
}
else
{
buffer = {buffer.data(), (size_type) byteswritten};
reqs.buffers = {reqs.buffers.data(), i + 1};
break;
}
}
return {reqs.buffers};
}
io_handle::io_result io_handle::_do_barrier(io_handle::io_request reqs, barrier_kind kind, deadline d) noexcept
{
(void) kind;
LLFIO_LOG_FUNCTION_CALL(this);
if(is_pipe() || is_socket())
{
return success(); // nothing was flushed
}
if(d)
{
return errc::not_supported;
}
#ifdef __linux__
if(kind <= barrier_kind::wait_data_only)
{
// Linux has a lovely dedicated syscall giving us exactly what we need here
extent_type offset = reqs.offset, bytes = 0;
// empty buffers means bytes = 0 which means sync entire file
for(const auto &req : reqs.buffers)
{
bytes += req.size();
}
unsigned flags = SYNC_FILE_RANGE_WRITE; // start writing all dirty pages in range now
if(kind == barrier_kind::wait_data_only)
{
flags |= SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WAIT_AFTER; // block until they're on storage
}
if(-1 != ::sync_file_range(_v.fd, offset, bytes, flags))
{
return {reqs.buffers};
}
}
#endif
#if !defined(__FreeBSD__) && !defined(__APPLE__) // neither of these have fdatasync()
if(kind <= barrier_kind::wait_data_only)
{
if(-1 == ::fdatasync(_v.fd))
{
return posix_error();
}
return {reqs.buffers};
}
#endif
#ifdef __APPLE__
if(((uint8_t) kind & 1) == 0)
{
// OS X fsync doesn't wait for the device to flush its buffers
if(-1 == ::fsync(_v.fd))
return posix_error();
return {std::move(reqs.buffers)};
}
// This is the fsync as on every other OS
if(-1 == ::fcntl(_v.fd, F_FULLFSYNC))
return posix_error();
#else
if(-1 == ::fsync(_v.fd))
{
return posix_error();
}
#endif
return {reqs.buffers};
}
LLFIO_V2_NAMESPACE_END