rust/src/libnative/io/mod.rs

339 lines
11 KiB
Rust
Raw Normal View History

// Copyright 2013 The Rust Project Developers. See the COPYRIGHT
// file at the top-level directory of this distribution and at
// http://rust-lang.org/COPYRIGHT.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.
//! Native thread-blocking I/O implementation
//!
//! This module contains the implementation of native thread-blocking
//! implementations of I/O on all platforms. This module is not intended to be
//! used directly, but rather the rust runtime will fall back to using it if
//! necessary.
//!
//! Rust code normally runs inside of green tasks with a local scheduler using
//! asynchronous I/O to cooperate among tasks. This model is not always
//! available, however, and that's where these native implementations come into
//! play. The only dependencies of these modules are the normal system libraries
//! that you would find on the respective platform.
use std::c_str::CString;
use std::io;
use std::io::IoError;
use std::io::net::ip::SocketAddr;
use std::io::process::ProcessConfig;
use std::io::signal::Signum;
use std::libc::c_int;
use std::libc;
use std::os;
use std::rt::rtio;
use std::rt::rtio::{RtioTcpStream, RtioTcpListener, RtioUdpSocket,
RtioUnixListener, RtioPipe, RtioFileStream, RtioProcess,
RtioSignal, RtioTTY, CloseBehavior, RtioTimer};
use ai = std::io::net::addrinfo;
// Local re-exports
pub use self::file::FileDesc;
pub use self::process::Process;
// Native I/O implementations
pub mod addrinfo;
2013-12-27 19:50:16 -06:00
pub mod net;
pub mod process;
#[cfg(unix)]
#[path = "file_unix.rs"]
pub mod file;
#[cfg(windows)]
#[path = "file_win32.rs"]
pub mod file;
Implement native timers Native timers are a much hairier thing to deal with than green timers due to the interface that we would like to expose (both a blocking sleep() and a channel-based interface). I ended up implementing timers in three different ways for the various platforms that we supports. In all three of the implementations, there is a worker thread which does send()s on channels for timers. This worker thread is initialized once and then communicated to in a platform-specific manner, but there's always a shared channel available for sending messages to the worker thread. * Windows - I decided to use windows kernel timer objects via CreateWaitableTimer and SetWaitableTimer in order to provide sleeping capabilities. The worker thread blocks via WaitForMultipleObjects where one of the objects is an event that is used to wake up the helper thread (which then drains the incoming message channel for requests). * Linux/(Android?) - These have the ideal interface for implementing timers, timerfd_create. Each timer corresponds to a timerfd, and the helper thread uses epoll to wait for all active timers and then send() for the next one that wakes up. The tricky part in this implementation is updating a timerfd, but see the implementation for the fun details * OSX/FreeBSD - These obviously don't have the windows APIs, and sadly don't have the timerfd api available to them, so I have thrown together a solution which uses select() plus a timeout in order to ad-hoc-ly implement a timer solution for threads. The implementation is backed by a sorted array of timers which need to fire. As I said, this is an ad-hoc solution which is certainly not accurate timing-wise. I have done this implementation due to the lack of other primitives to provide an implementation, and I've done it the best that I could, but I'm sure that there's room for improvement. I'm pretty happy with how these implementations turned out. In theory we could drop the timerfd implementation and have linux use the select() + timeout implementation, but it's so inaccurate that I would much rather continue to use timerfd rather than my ad-hoc select() implementation. The only change that I would make to the API in general is to have a generic sleep() method on an IoFactory which doesn't require allocating a Timer object. For everything but windows it's super-cheap to request a blocking sleep for a set amount of time, and it's probably worth it to provide a sleep() which doesn't do something like allocate a file descriptor on linux.
2013-12-29 01:33:56 -06:00
#[cfg(target_os = "macos")]
#[cfg(target_os = "freebsd")]
#[cfg(target_os = "android")]
Implement native timers Native timers are a much hairier thing to deal with than green timers due to the interface that we would like to expose (both a blocking sleep() and a channel-based interface). I ended up implementing timers in three different ways for the various platforms that we supports. In all three of the implementations, there is a worker thread which does send()s on channels for timers. This worker thread is initialized once and then communicated to in a platform-specific manner, but there's always a shared channel available for sending messages to the worker thread. * Windows - I decided to use windows kernel timer objects via CreateWaitableTimer and SetWaitableTimer in order to provide sleeping capabilities. The worker thread blocks via WaitForMultipleObjects where one of the objects is an event that is used to wake up the helper thread (which then drains the incoming message channel for requests). * Linux/(Android?) - These have the ideal interface for implementing timers, timerfd_create. Each timer corresponds to a timerfd, and the helper thread uses epoll to wait for all active timers and then send() for the next one that wakes up. The tricky part in this implementation is updating a timerfd, but see the implementation for the fun details * OSX/FreeBSD - These obviously don't have the windows APIs, and sadly don't have the timerfd api available to them, so I have thrown together a solution which uses select() plus a timeout in order to ad-hoc-ly implement a timer solution for threads. The implementation is backed by a sorted array of timers which need to fire. As I said, this is an ad-hoc solution which is certainly not accurate timing-wise. I have done this implementation due to the lack of other primitives to provide an implementation, and I've done it the best that I could, but I'm sure that there's room for improvement. I'm pretty happy with how these implementations turned out. In theory we could drop the timerfd implementation and have linux use the select() + timeout implementation, but it's so inaccurate that I would much rather continue to use timerfd rather than my ad-hoc select() implementation. The only change that I would make to the API in general is to have a generic sleep() method on an IoFactory which doesn't require allocating a Timer object. For everything but windows it's super-cheap to request a blocking sleep for a set amount of time, and it's probably worth it to provide a sleep() which doesn't do something like allocate a file descriptor on linux.
2013-12-29 01:33:56 -06:00
#[path = "timer_other.rs"]
pub mod timer;
#[cfg(target_os = "linux")]
#[path = "timer_timerfd.rs"]
pub mod timer;
#[cfg(target_os = "win32")]
#[path = "timer_win32.rs"]
pub mod timer;
#[cfg(unix)]
#[path = "pipe_unix.rs"]
pub mod pipe;
#[cfg(windows)]
#[path = "pipe_win32.rs"]
pub mod pipe;
Implement native timers Native timers are a much hairier thing to deal with than green timers due to the interface that we would like to expose (both a blocking sleep() and a channel-based interface). I ended up implementing timers in three different ways for the various platforms that we supports. In all three of the implementations, there is a worker thread which does send()s on channels for timers. This worker thread is initialized once and then communicated to in a platform-specific manner, but there's always a shared channel available for sending messages to the worker thread. * Windows - I decided to use windows kernel timer objects via CreateWaitableTimer and SetWaitableTimer in order to provide sleeping capabilities. The worker thread blocks via WaitForMultipleObjects where one of the objects is an event that is used to wake up the helper thread (which then drains the incoming message channel for requests). * Linux/(Android?) - These have the ideal interface for implementing timers, timerfd_create. Each timer corresponds to a timerfd, and the helper thread uses epoll to wait for all active timers and then send() for the next one that wakes up. The tricky part in this implementation is updating a timerfd, but see the implementation for the fun details * OSX/FreeBSD - These obviously don't have the windows APIs, and sadly don't have the timerfd api available to them, so I have thrown together a solution which uses select() plus a timeout in order to ad-hoc-ly implement a timer solution for threads. The implementation is backed by a sorted array of timers which need to fire. As I said, this is an ad-hoc solution which is certainly not accurate timing-wise. I have done this implementation due to the lack of other primitives to provide an implementation, and I've done it the best that I could, but I'm sure that there's room for improvement. I'm pretty happy with how these implementations turned out. In theory we could drop the timerfd implementation and have linux use the select() + timeout implementation, but it's so inaccurate that I would much rather continue to use timerfd rather than my ad-hoc select() implementation. The only change that I would make to the API in general is to have a generic sleep() method on an IoFactory which doesn't require allocating a Timer object. For everything but windows it's super-cheap to request a blocking sleep for a set amount of time, and it's probably worth it to provide a sleep() which doesn't do something like allocate a file descriptor on linux.
2013-12-29 01:33:56 -06:00
mod timer_helper;
pub type IoResult<T> = Result<T, IoError>;
fn unimpl() -> IoError {
IoError {
kind: io::IoUnavailable,
desc: "unimplemented I/O interface",
detail: None,
}
}
fn translate_error(errno: i32, detail: bool) -> IoError {
#[cfg(windows)]
fn get_err(errno: i32) -> (io::IoErrorKind, &'static str) {
match errno {
libc::EOF => (io::EndOfFile, "end of file"),
libc::ERROR_NO_DATA => (io::BrokenPipe, "the pipe is being closed"),
libc::ERROR_FILE_NOT_FOUND => (io::FileNotFound, "file not found"),
libc::ERROR_INVALID_NAME => (io::InvalidInput, "invalid file name"),
2013-12-27 19:50:16 -06:00
libc::WSAECONNREFUSED => (io::ConnectionRefused, "connection refused"),
libc::WSAECONNRESET => (io::ConnectionReset, "connection reset"),
libc::WSAEACCES => (io::PermissionDenied, "permission denied"),
libc::WSAEWOULDBLOCK => {
(io::ResourceUnavailable, "resource temporarily unavailable")
}
2013-12-27 19:50:16 -06:00
libc::WSAENOTCONN => (io::NotConnected, "not connected"),
libc::WSAECONNABORTED => (io::ConnectionAborted, "connection aborted"),
libc::WSAEADDRNOTAVAIL => (io::ConnectionRefused, "address not available"),
libc::WSAEADDRINUSE => (io::ConnectionRefused, "address in use"),
libc::ERROR_BROKEN_PIPE => (io::EndOfFile, "the pipe has ended"),
2013-12-27 19:50:16 -06:00
// libuv maps this error code to EISDIR. we do too. if it is found
// to be incorrect, we can add in some more machinery to only
// return this message when ERROR_INVALID_FUNCTION after certain
// win32 calls.
libc::ERROR_INVALID_FUNCTION => (io::InvalidInput,
"illegal operation on a directory"),
log: Introduce liblog, the old std::logging This commit moves all logging out of the standard library into an external crate. This crate is the new crate which is responsible for all logging macros and logging implementation. A few reasons for this change are: * The crate map has always been a bit of a code smell among rust programs. It has difficulty being loaded on almost all platforms, and it's used almost exclusively for logging and only logging. Removing the crate map is one of the end goals of this movement. * The compiler has a fair bit of special support for logging. It has the __log_level() expression as well as generating a global word per module specifying the log level. This is unfairly favoring the built-in logging system, and is much better done purely in libraries instead of the compiler itself. * Initialization of logging is much easier to do if there is no reliance on a magical crate map being available to set module log levels. * If the logging library can be written outside of the standard library, there's no reason that it shouldn't be. It's likely that we're not going to build the highest quality logging library of all time, so third-party libraries should be able to provide just as high-quality logging systems as the default one provided in the rust distribution. With a migration such as this, the change does not come for free. There are some subtle changes in the behavior of liblog vs the previous logging macros: * The core change of this migration is that there is no longer a physical log-level per module. This concept is still emulated (it is quite useful), but there is now only a global log level, not a local one. This global log level is a reflection of the maximum of all log levels specified. The previously generated logging code looked like: if specified_level <= __module_log_level() { println!(...) } The newly generated code looks like: if specified_level <= ::log::LOG_LEVEL { if ::log::module_enabled(module_path!()) { println!(...) } } Notably, the first layer of checking is still intended to be "super fast" in that it's just a load of a global word and a compare. The second layer of checking is executed to determine if the current module does indeed have logging turned on. This means that if any module has a debug log level turned on, all modules with debug log levels get a little bit slower (they all do more expensive dynamic checks to determine if they're turned on or not). Semantically, this migration brings no change in this respect, but runtime-wise, this will have a perf impact on some code. * A `RUST_LOG=::help` directive will no longer print out a list of all modules that can be logged. This is because the crate map will no longer specify the log levels of all modules, so the list of modules is not known. Additionally, warnings can no longer be provided if a malformed logging directive was supplied. The new "hello world" for logging looks like: #[phase(syntax, link)] extern crate log; fn main() { debug!("Hello, world!"); }
2014-03-09 00:11:44 -06:00
_ => (io::OtherIoError, "unknown error")
}
}
#[cfg(not(windows))]
fn get_err(errno: i32) -> (io::IoErrorKind, &'static str) {
// FIXME: this should probably be a bit more descriptive...
match errno {
libc::EOF => (io::EndOfFile, "end of file"),
2013-12-27 19:50:16 -06:00
libc::ECONNREFUSED => (io::ConnectionRefused, "connection refused"),
libc::ECONNRESET => (io::ConnectionReset, "connection reset"),
libc::EPERM | libc::EACCES =>
(io::PermissionDenied, "permission denied"),
libc::EPIPE => (io::BrokenPipe, "broken pipe"),
libc::ENOTCONN => (io::NotConnected, "not connected"),
libc::ECONNABORTED => (io::ConnectionAborted, "connection aborted"),
libc::EADDRNOTAVAIL => (io::ConnectionRefused, "address not available"),
libc::EADDRINUSE => (io::ConnectionRefused, "address in use"),
libc::ENOENT => (io::FileNotFound, "no such file or directory"),
libc::EISDIR => (io::InvalidInput, "illegal operation on a directory"),
// These two constants can have the same value on some systems, but
// different values on others, so we can't use a match clause
x if x == libc::EAGAIN || x == libc::EWOULDBLOCK =>
(io::ResourceUnavailable, "resource temporarily unavailable"),
log: Introduce liblog, the old std::logging This commit moves all logging out of the standard library into an external crate. This crate is the new crate which is responsible for all logging macros and logging implementation. A few reasons for this change are: * The crate map has always been a bit of a code smell among rust programs. It has difficulty being loaded on almost all platforms, and it's used almost exclusively for logging and only logging. Removing the crate map is one of the end goals of this movement. * The compiler has a fair bit of special support for logging. It has the __log_level() expression as well as generating a global word per module specifying the log level. This is unfairly favoring the built-in logging system, and is much better done purely in libraries instead of the compiler itself. * Initialization of logging is much easier to do if there is no reliance on a magical crate map being available to set module log levels. * If the logging library can be written outside of the standard library, there's no reason that it shouldn't be. It's likely that we're not going to build the highest quality logging library of all time, so third-party libraries should be able to provide just as high-quality logging systems as the default one provided in the rust distribution. With a migration such as this, the change does not come for free. There are some subtle changes in the behavior of liblog vs the previous logging macros: * The core change of this migration is that there is no longer a physical log-level per module. This concept is still emulated (it is quite useful), but there is now only a global log level, not a local one. This global log level is a reflection of the maximum of all log levels specified. The previously generated logging code looked like: if specified_level <= __module_log_level() { println!(...) } The newly generated code looks like: if specified_level <= ::log::LOG_LEVEL { if ::log::module_enabled(module_path!()) { println!(...) } } Notably, the first layer of checking is still intended to be "super fast" in that it's just a load of a global word and a compare. The second layer of checking is executed to determine if the current module does indeed have logging turned on. This means that if any module has a debug log level turned on, all modules with debug log levels get a little bit slower (they all do more expensive dynamic checks to determine if they're turned on or not). Semantically, this migration brings no change in this respect, but runtime-wise, this will have a perf impact on some code. * A `RUST_LOG=::help` directive will no longer print out a list of all modules that can be logged. This is because the crate map will no longer specify the log levels of all modules, so the list of modules is not known. Additionally, warnings can no longer be provided if a malformed logging directive was supplied. The new "hello world" for logging looks like: #[phase(syntax, link)] extern crate log; fn main() { debug!("Hello, world!"); }
2014-03-09 00:11:44 -06:00
_ => (io::OtherIoError, "unknown error")
}
}
let (kind, desc) = get_err(errno);
IoError {
kind: kind,
desc: desc,
detail: if detail {Some(os::last_os_error())} else {None},
}
}
fn last_error() -> IoError { translate_error(os::errno() as i32, true) }
// unix has nonzero values as errors
fn mkerr_libc(ret: libc::c_int) -> IoResult<()> {
if ret != 0 {
Err(last_error())
} else {
Ok(())
}
}
// windows has zero values as errors
2013-12-08 01:55:28 -06:00
#[cfg(windows)]
fn mkerr_winbool(ret: libc::c_int) -> IoResult<()> {
if ret == 0 {
Err(last_error())
} else {
Ok(())
}
}
#[cfg(windows)]
#[inline]
fn retry(f: || -> libc::c_int) -> libc::c_int {
loop {
match f() {
-1 if os::errno() as int == libc::WSAEINTR as int => {}
n => return n,
}
}
}
#[cfg(unix)]
#[inline]
fn retry(f: || -> libc::c_int) -> libc::c_int {
loop {
match f() {
-1 if os::errno() as int == libc::EINTR as int => {}
n => return n,
}
}
}
fn keep_going(data: &[u8], f: |*u8, uint| -> i64) -> i64 {
let origamt = data.len();
let mut data = data.as_ptr();
let mut amt = origamt;
while amt > 0 {
let ret = retry(|| f(data, amt) as libc::c_int);
if ret == 0 {
break
} else if ret != -1 {
amt -= ret as uint;
data = unsafe { data.offset(ret as int) };
} else {
return ret as i64;
}
}
return (origamt - amt) as i64;
}
/// Implementation of rt::rtio's IoFactory trait to generate handles to the
/// native I/O functionality.
2013-12-27 19:50:16 -06:00
pub struct IoFactory {
priv cannot_construct_outside_of_this_module: ()
}
impl IoFactory {
pub fn new() -> IoFactory {
net::init();
IoFactory { cannot_construct_outside_of_this_module: () }
}
}
impl rtio::IoFactory for IoFactory {
// networking
2013-12-27 19:50:16 -06:00
fn tcp_connect(&mut self, addr: SocketAddr) -> IoResult<~RtioTcpStream> {
net::TcpStream::connect(addr).map(|s| ~s as ~RtioTcpStream)
}
2013-12-27 19:50:16 -06:00
fn tcp_bind(&mut self, addr: SocketAddr) -> IoResult<~RtioTcpListener> {
net::TcpListener::bind(addr).map(|s| ~s as ~RtioTcpListener)
}
2013-12-28 18:40:15 -06:00
fn udp_bind(&mut self, addr: SocketAddr) -> IoResult<~RtioUdpSocket> {
net::UdpSocket::bind(addr).map(|u| ~u as ~RtioUdpSocket)
}
fn unix_bind(&mut self, path: &CString) -> IoResult<~RtioUnixListener> {
pipe::UnixListener::bind(path).map(|s| ~s as ~RtioUnixListener)
}
fn unix_connect(&mut self, path: &CString) -> IoResult<~RtioPipe> {
pipe::UnixStream::connect(path).map(|s| ~s as ~RtioPipe)
}
fn get_host_addresses(&mut self, host: Option<&str>, servname: Option<&str>,
hint: Option<ai::Hint>) -> IoResult<~[ai::Info]> {
addrinfo::GetAddrInfoRequest::run(host, servname, hint)
}
// filesystem operations
fn fs_from_raw_fd(&mut self, fd: c_int,
close: CloseBehavior) -> ~RtioFileStream {
let close = match close {
rtio::CloseSynchronously | rtio::CloseAsynchronously => true,
rtio::DontClose => false
};
~file::FileDesc::new(fd, close) as ~RtioFileStream
}
fn fs_open(&mut self, path: &CString, fm: io::FileMode, fa: io::FileAccess)
-> IoResult<~RtioFileStream> {
file::open(path, fm, fa).map(|fd| ~fd as ~RtioFileStream)
}
fn fs_unlink(&mut self, path: &CString) -> IoResult<()> {
file::unlink(path)
}
fn fs_stat(&mut self, path: &CString) -> IoResult<io::FileStat> {
file::stat(path)
}
fn fs_mkdir(&mut self, path: &CString,
mode: io::FilePermission) -> IoResult<()> {
file::mkdir(path, mode)
}
fn fs_chmod(&mut self, path: &CString,
mode: io::FilePermission) -> IoResult<()> {
file::chmod(path, mode)
}
fn fs_rmdir(&mut self, path: &CString) -> IoResult<()> {
file::rmdir(path)
}
fn fs_rename(&mut self, path: &CString, to: &CString) -> IoResult<()> {
file::rename(path, to)
}
fn fs_readdir(&mut self, path: &CString, _flags: c_int) -> IoResult<~[Path]> {
file::readdir(path)
}
fn fs_lstat(&mut self, path: &CString) -> IoResult<io::FileStat> {
file::lstat(path)
}
fn fs_chown(&mut self, path: &CString, uid: int, gid: int) -> IoResult<()> {
file::chown(path, uid, gid)
}
fn fs_readlink(&mut self, path: &CString) -> IoResult<Path> {
file::readlink(path)
}
fn fs_symlink(&mut self, src: &CString, dst: &CString) -> IoResult<()> {
file::symlink(src, dst)
}
fn fs_link(&mut self, src: &CString, dst: &CString) -> IoResult<()> {
file::link(src, dst)
}
fn fs_utime(&mut self, src: &CString, atime: u64,
mtime: u64) -> IoResult<()> {
file::utime(src, atime, mtime)
}
// misc
fn timer_init(&mut self) -> IoResult<~RtioTimer> {
Implement native timers Native timers are a much hairier thing to deal with than green timers due to the interface that we would like to expose (both a blocking sleep() and a channel-based interface). I ended up implementing timers in three different ways for the various platforms that we supports. In all three of the implementations, there is a worker thread which does send()s on channels for timers. This worker thread is initialized once and then communicated to in a platform-specific manner, but there's always a shared channel available for sending messages to the worker thread. * Windows - I decided to use windows kernel timer objects via CreateWaitableTimer and SetWaitableTimer in order to provide sleeping capabilities. The worker thread blocks via WaitForMultipleObjects where one of the objects is an event that is used to wake up the helper thread (which then drains the incoming message channel for requests). * Linux/(Android?) - These have the ideal interface for implementing timers, timerfd_create. Each timer corresponds to a timerfd, and the helper thread uses epoll to wait for all active timers and then send() for the next one that wakes up. The tricky part in this implementation is updating a timerfd, but see the implementation for the fun details * OSX/FreeBSD - These obviously don't have the windows APIs, and sadly don't have the timerfd api available to them, so I have thrown together a solution which uses select() plus a timeout in order to ad-hoc-ly implement a timer solution for threads. The implementation is backed by a sorted array of timers which need to fire. As I said, this is an ad-hoc solution which is certainly not accurate timing-wise. I have done this implementation due to the lack of other primitives to provide an implementation, and I've done it the best that I could, but I'm sure that there's room for improvement. I'm pretty happy with how these implementations turned out. In theory we could drop the timerfd implementation and have linux use the select() + timeout implementation, but it's so inaccurate that I would much rather continue to use timerfd rather than my ad-hoc select() implementation. The only change that I would make to the API in general is to have a generic sleep() method on an IoFactory which doesn't require allocating a Timer object. For everything but windows it's super-cheap to request a blocking sleep for a set amount of time, and it's probably worth it to provide a sleep() which doesn't do something like allocate a file descriptor on linux.
2013-12-29 01:33:56 -06:00
timer::Timer::new().map(|t| ~t as ~RtioTimer)
}
fn spawn(&mut self, config: ProcessConfig)
-> IoResult<(~RtioProcess, ~[Option<~RtioPipe>])> {
process::Process::spawn(config).map(|(p, io)| {
(~p as ~RtioProcess,
io.move_iter().map(|p| p.map(|p| ~p as ~RtioPipe)).collect())
})
}
fn kill(&mut self, pid: libc::pid_t, signum: int) -> IoResult<()> {
process::Process::kill(pid, signum)
}
fn pipe_open(&mut self, fd: c_int) -> IoResult<~RtioPipe> {
Ok(~file::FileDesc::new(fd, true) as ~RtioPipe)
}
fn tty_open(&mut self, fd: c_int, _readable: bool) -> IoResult<~RtioTTY> {
if unsafe { libc::isatty(fd) } != 0 {
2013-12-27 19:50:16 -06:00
Ok(~file::FileDesc::new(fd, true) as ~RtioTTY)
} else {
Err(IoError {
kind: io::MismatchedFileTypeForOperation,
desc: "file descriptor is not a TTY",
detail: None,
})
}
}
fn signal(&mut self, _signal: Signum, _channel: Sender<Signum>)
-> IoResult<~RtioSignal> {
Err(unimpl())
}
}