Introduces `StreamExt` trait. This trait will be used to add utility functions
to make working with streams easier. This patch includes two functions:
* `next`: a future returning the item in the stream.
* `map`: transform each item in the stream.
`num_threads` is deprecated. Instead, `core_threads` and `max_threads` are
introduced. `core_threads` specifies the number of "always on" threads used
for the async task executor and `max_threads` specifies the maximum number
of threads that the runtime may spawn.
Adds a broadcast channel implementation. A broadcast channel is a
multi-producer, multi-consumer channel where each consumer receives a
clone of every value sent. This is useful for implementing pub / sub
style patterns.
Implemented as a ring buffer, a Vec of the specified capacity is
allocated on initialization of the channel. Values are pushed into
slots.
When the channel is full, a send overwrites the oldest value. Receivers
detect this and return an error on the next call to receive. This
prevents unbounded buffering and does not make the channel vulnerable to
the slowest consumer.
Closes: #1585
The blocking task queue was not explicitly drained as part of the
blocking pool shutdown logic. It was originally assumed that the
contents of the queue would be dropped when the blocking pool structure
is dropped. However, tasks must be explicitly shutdown, so we must drain
the queue can call `shutdown` on each task.
Fixes#1970, #1946
Calls to tasks should not be nested. Currently, while a task is being
executed and the runtime is shutting down, a call to wake() can result
in the wake target to be dropped. This, in turn, results in the drop
handler being called.
If the user holds a ref cell borrow, a mutex guard, or any such value,
dropping the task inline can result in a deadlock.
The fix is to permit tasks to be scheduled during the shutdown process
and dropping the tasks once they are popped from the queue.
Fixes#1929, #1886
Update the rotted thread_pool benchmarks. These benchmarks are not the
greatest, but as of now it is all we have for micro benchmarks.
Adds a little yielding in the parker as it helps a bit.
Currently, a `LocalSet` does not notify the `LocalFuture` again at the
end of a tick. This means that if we didn't poll every task in the run
queue during that tick (e.g. there are more than 61 tasks enqueued),
those tasks will not be polled.
This commit fixes this issue by changing `local::Scheduler::tick` to
return whether or not the local future needs to be notified again, and
waking the task if so.
Fixes#1899Fixes#1900
Signed-off-by: Eliza Weisman <[email protected]>
## Motivation
#1800 removed the lazy binding of `Delay`s to timers. With the removal of the
logic required for that, `HandlePriv` is no longer needed. This PR removes the
use of `HandlePriv`.
A `TODO` was also removed that would panic if when registering a new `Delay`
the current timer handle was full. That has been fixed to now immediately
transition that `Delay` to an error state that can be handled in a similar way
to other error states.
Signed-off-by: Kevin Leimkuhler <[email protected]>
## Motivation
There's currently an issue in `task::LocalSet` where dropping the local
set can result in an infinite loop if a task running in the local set is
notified from outside the local set (e.g. by a timer). This was reported
in issue #1885.
This issue exists because the `Drop` impl for `task::local::Scheduler`
does not drain the queue of tasks notified externally, the way the basic
scheduler does. Instead, only the local queue is drained, leaving some
tasks in place. Since these tasks are never removed, the loop that
continues trying to cancel tasks until the owned task list is totally
empty continues infinitely.
I think this issue was due to the `Drop` impl being written before a
remote queue was added to the local scheduler, and the need to close the
remote queue as well was overlooked.
## Solution
This branch solves the problem by clearing the local scheduler's remote
queue as well as the local one.
I've added a test that reproduces the behavior. The test fails on master
and passes after this change.
In addition, this branch factors out the common task queue logic in the
basic scheduler runtime and the `LocalSet` struct in `tokio::task`. This
is because as more work was done on the `LocalSet`, it has gotten closer
and closer to the basic scheduler in behavior, and factoring out the
shared code reduces the risk of errors caused by `LocalSet` not doing
something that the basic scheduler does. The queues are now encapsulated
by a `MpscQueues` struct in `tokio::task::queue` (crate-public). As a
follow-up, I'd also like to look into changing this type to use the same
remote queue type as the threadpool (a linked list).
In particular, I noticed the basic scheduler has a flag that indicates
the remote queue has been closed, which is set when dropping the
scheduler. This prevents tasks from being added after the scheduler has
started shutting down, stopping a potential task leak. Rather than
duplicating this code in `LocalSet`, I thought it was probably better to
factor it out into a shared type.
There are a few cases where there are small differences in behavior,
though, so there is still a need for separate types implemented _using_
the new `MpscQueues` struct. However, it should cover most of the
identical code.
Note that this diff is rather large, due to the refactoring. However, the
actual fix for the infinite loop is very simple. It can be reviewed on its own
by looking at commit 4f46ac6. The refactor is in a separate commit, with
the SHA 90b5b1f.
Fixes#1885
Signed-off-by: Eliza Weisman <[email protected]>
## Motivation
Currently, `tokio::task::LocalSet`'s `block_on` method requires the
future to live for the 'static lifetime. However, this bound is not
required — the future is wrapped in a `LocalFuture`, and then passed
into `Runtime::block_on`, which does _not_ require a `'static` future.
This came up while updating `tokio-compat` to work with version 0.2. To
mimic the behavior of `tokio` 0.1's `current_thread::Runtime::run`, we
want to be able to have a runtime block on the `recv` future from an
mpsc channel indicating when the runtime is idle. To support `!Send`
futures, as the old `current_thread::Runtime` did, we must do so inside
of a `LocalSet`. However, with the current bounds, we cannot await an
`mpsc::Receiver`'s `recv` future inside the `LocalSet::block_on` call.
## Solution
This branch removes the unnecessary `'static` bound.
Signed-off-by: Eliza Weisman <[email protected]>
Adds `read_buf` and `write_buf` which work with `T: BufMut` and `T: Buf`
respectively. This adds an easy API for using the buffer traits provided
by `bytes.
`oneshot::Receiver::try_recv` does not provide any information as to the
reason **why** receiving failed. The two cases are that the channel is
empty or that the channel closed.
`TryRecvError` is changed to be an enum of those two cases. This is
backwards compatible as `TryRecvError` was an opaque struct.
This also expands on `oneshot` API documentation, adding details and
examples.
Closes#1872
Provide convenience methods for encoding and decoding big-endian numbers
on top of asynchronous I/O streams. Only primitive types are provided
(24 and 48 bit numbers are omitted).
In general, using these methods won't be the fastest way to do
encoding/decoding with asynchronous byte streams, but they help to get
simple things working fast.
The "global executor" thread-local is to track where to spawn new tasks,
**not** which scheduler is active on the current thread. This fixes a
bug with scheduling tasks on the basic_scheduler by tracking the
currently active basic_scheduler with a dedicated thread-local variable.
Fixes: #1851
In version 0.1 there was File::into_std method that destructured
tokio_fs::File into std::fs:File. That method was lacking in
version 0.2.
Fixes: #1852
It turns out that the `Scheduler::release` method on `LocalSet`'s
`Scheduler` *is* called, when the `Scheduler` is dropped with tasks
still running. Currently, that method is `unreachable!`, which means
that dropping a `LocalSet` with tasks running will panic.
This commit fixes the panic, by pushing released tasks to
`pending_drop`. This is the same as `BasicScheduler`.
Fixes#1842
This fixes the API docs for both `TcpListener::incoming` and
`UnixListener::incoming`. The function now takes `&mut self` instead of
`self`. Adds an example for both function.
## Motivation
In earlier versions of `tokio`, the `current_thread::Runtime` type could
be used to run `!Send` futures. However, PR #1716 merged the
current-thread and threadpool runtimes into a single type, which can no
longer run `!Send` futures. There is still a need in some cases to
support futures that don't implement `Send`, and the `tokio-compat`
crate requires this in order to provide APIs that existed in `tokio`
0.1.
## Solution
This branch implements the API described by @carllerche in
https://github.com/tokio-rs/tokio/pull/1716#issuecomment-549496309. It
adds a new `LocalSet` type and `spawn_local` function to `tokio::task`.
The `LocalSet` type is used to group together a set of tasks which must
run on the same thread and don't implement `Send`. These are available
when a new "rt-util" feature flag is enabled.
Currently, the local task set is run by passing it a reference to a
`Runtime` and a future to `block_on`. In the future, we may also want
to investigate allowing spawned futures to construct their own local
task sets, which would be executed on the worker that the future is
executing on.
In order to implement the new API, I've made some internal changes to
the `task` module and `Schedule` trait to support scheduling both `Send`
and `!Send` futures.
Signed-off-by: Eliza Weisman <[email protected]>
This provides the ability to get the raw OS handle for a `File`. The
`Into*` variant cannot be provided as `File` needs to maintain ownership
of the `File`. The actual handle may have been moved to a background
thread.
This updates the tokio `Command` and `Child` behavior to match that of
the stdlib: spawned processes will *not* be automatically killed when
the handle is dropped
Unlike the stdlib, any dropped (unix) processes may be reaped by tokio
behind-the-scenes after they exit and if new processes are awaited,
which mitigates the risks of piling up unreaped zombie unix processes
A `Command::kill_on_drop` method is added to allow the caller to
control whether the spawned child should be killed when the handle is
dropped. By default, this value is `false`.
The `Child::forget` method has been removed, as it is superseded by
`Command::kill_on_drop`
Changes the set of `default` feature flags to `[]`. By default, only
core traits are included without specifying feature flags. This makes it
easier for users to pick the components they need.
For convenience, a `full` feature flag is included that includes all
components.
Tests are configured to require the `full` feature. Testing individual
feature flags will need to be moved to a separate crate.
Closes#1791
This directory was deleted when `cargo hack` was introduced, however
there were some tests that were still useful (macro failure output).
Also, additional build tests will be added over time.
Annotates types in `tokio::io` module with their required feature flag.
This annotation is included in generated documentation.
Notes:
* The annotation must be on the type or function itself. Annotating just
the re-export is not sufficient.
* The annotation must be **inside** the `pin_project!` macro or it is
lost.
* runtime: cleanup and add config options
This patch finishes the cleanup as part of the transition to Tokio 0.2.
A number of changes were made to take advantage of having all Tokio
types in a single crate. Also, fixes using Tokio types from
`spawn_blocking`.
* Many threads, one resource driver
Previously, in the threaded scheduler, a resource driver (mio::Poll /
timer combo) was created per thread. This was more or less fine, except
it required balancing across the available drivers. When using a
resource driver from **outside** of the thread pool, balancing is
tricky. The change was original done to avoid having a dedicated driver
thread.
Now, instead of creating many resource drivers, a single resource driver
is used. Each scheduler thread will attempt to "lock" the resource
driver before parking on it. If the resource driver is already locked,
the thread uses a condition variable to park. Contention should remain
low as, under load, the scheduler avoids using the drivers.
* Add configuration options to enable I/O / time
New configuration options are added to `runtime::Builder` to allow
enabling I/O and time drivers on a runtime instance basis. This is
useful when wanting to create lightweight runtime instances to execute
compute only tasks.
* Bug fixes
The condition variable parker is updated to the same algorithm used in
`std`. This is motivated by some potential deadlock cases discovered by
`loom`.
The basic scheduler is fixed to fairly schedule tasks. `push_front` was
accidentally used instead of `push_back`.
I/O, time, and spawning now work from within `spawn_blocking` closures.
* Misc cleanup
The threaded scheduler is no longer generic over `P :Park`. Instead, it
is hard coded to a specific parker. Tests, including loom tests, are
updated to use `Runtime` directly. This provides greater coverage.
The `blocking` module is moved back into `runtime` as all usage is
within `runtime` itself.
## Motivation
Tokio's crate-level docs are currently pretty sparse, and in some cases
reference old names for APIs. Before 0.2 is released, they could use a
fresh coat of paint.
## Solution
This branch reworks and expands the `lib.rs` docs. In particular, I've
added a new "A Tour of Tokio" section, inspired by the [standard
library's similarly-named section][std]. This section lists all of
`tokio`'s public modules, and summarizes their major APIs. It also lists
the feature flags necessary to enable those APIs.
[std]: https://doc.rust-lang.org/std/index.html#a-tour-of-the-rust-standard-library
Signed-off-by: Eliza Weisman <[email protected]>
## Motivation
The `tokio::runtime` module's docs need to be updated to
track recent changes.
## Solution
This branch updates and expands the `runtime` docs.
Signed-off-by: Eliza Weisman <[email protected]>
## Motivation
The `tokio::io` module's docs are fairly sparse and not particularly up
to date. They ought to be improved before release.
## Solution
This branch adds new module-level docs to `tokio::io`. The new docs are
largely inspired by `std::io`'s documentation, and highlight the
similarities and differences between `tokio::io` and `std::io`.
Signed-off-by: Eliza Weisman <[email protected]>
## Motivation
The new `tokio::task` module is pretty lacking in API docs.
## Solution
This branch adds new API docs to the `task` module, including:
* Module-level docs with a summary of the differences between
tasks and threads
* Examples of how to use the `task` APIs in the module-level docs
* More docs for `yield_now`
* More docs and examples for `JoinHandle`, based on the
`std::thread::JoinHandle` API docs.
This branch contains commits cherry-picked from #1794
Signed-off-by: Eliza Weisman <[email protected]>
## Motivation
Similar to #1666, it is no longer necessary to lazily register delays with the
executions default timer. All delays are expected to be created from within a
runtime, and should panic if not done so.
## Solution
`tokio::time` now assumes there to be a `CURRENT_TIMER` set when creating a
delay; this can be assumed if called within a tokio runtime. If there is no
current timer, the application will panic with a "no current timer" message.
## Follow-up
Similar to #1666, `HandlePriv` can probably be removed, but this mainly prepares
for 0.2 API changes. Because it is not in the public API, this can be done in a
following change.
Signed-off-by: Kevin Leimkuhler <[email protected]>
The I/O driver is made private and moved to `tokio::io::driver`. `Registration` is
moved to `tokio::io::Registration` and `PollEvented` is moved to `tokio::io::PollEvented`.
Additionally, the concurrent slab used by the I/O driver is cleaned up and extracted to
`tokio::util::slab`, allowing it to eventually be used by other types.
When checking if a worker needs to be unparked, the SeqCst load does not
provide the necessary synchronization to ensure the scheduled task is
visible to the searching worker. The `load` is switched to
`fetch_add(0)` which does establish the necessary synchronization.
Adding unit tests catching this bug will require a fix to loom and will
be done at a later time. The bug fix has been validated with manual
testing.
Fixes#1768
Removes dependencies between Tokio feature flags. For example, `process`
should not depend on `sync` simply because it uses the `mpsc` channel.
Instead, feature flags represent **public** APIs that become available
with the feature enabled. When the feature is not enabled, the
functionality is removed. If another Tokio component requires the
functionality, it is stays as `pub(crate)`.
The threaded scheduler is now exposed under `rt-threaded`. This feature
flag only enables the threaded scheduler and does not include I/O,
networking, or time. Those features must be explictly enabled.
A `full` feature flag is added that enables all features.
`stdin`, `stdout`, `stderr` are exposed under `io-std`.
Macros are used to scope code by feature flag.
There are limitations with `async/await` (no GAT) requiring the value to
be cloned on receive. The `poll` based API is not currently exposed.
This makes the `Clone` requirement explicit.
This exposes `std{in, out, err}` under io feature by moving
`fs::blocking` module into `io::blocking`.
As `fs` feature depends on `io-trait` feature, `fs` implementations can
always access `io` module.
It no longer supports executing !Send futures. The use case for
It is wanting a “light” runtime. There will be “local” task execution
using a different strategy coming later.
This patch also renames `thread_pool` -> `threaded_scheduler`, but
only in public APIs for now.
In an effort to reach API stability, the `tokio` crate is shedding its
_public_ dependencies on crates that are either a) do not provide a
stable (1.0+) release with longevity guarantees or b) match the `tokio`
release cadence. Of course, implementing `std` traits fits the
requirements.
The on exception, for now, is the `Stream` trait found in `futures_core`.
It is expected that this trait will not change much and be moved into `std.
Since Tokio is not yet going reaching 1.0, I feel that it is acceptable to maintain
a dependency on this trait given how foundational it is.
Since the `Stream` implementation is optional, types that are logically
streams provide `async fn next_*` functions to obtain the next value.
Avoiding the `next()` name prevents fn conflicts with `StreamExt::next()`.
Additionally, some misc cleanup is also done:
- `tokio::io::io` -> `tokio::io::util`.
- `delay` -> `delay_until`.
- `Timeout::new` -> `timeout(...)`.
- `signal::ctrl_c()` returns a future instead of a stream.
- `{tcp,unix}::Incoming` is removed (due to lack of `Stream` trait).
- `time::Throttle` is removed (due to lack of `Stream` trait).
- Fix: `mpsc::UnboundedSender::send(&self)` (no more conflict with `Sink` fns).
This patch started as an effort to make `time::Timer` private. However, in an
effort to get the build compiling again, more and more changes were made. This
probably should have been broken up, but here we are. I will attempt to
summarize the changes here.
* Feature flags are reorganized to make clearer. `net-driver` becomes
`io-driver`. `rt-current-thread` becomes `rt-core`.
* The `Runtime` can be created without any executor. This replaces `enter`. It
also allows creating I/O / time drivers that are standalone.
* `tokio::timer` is renamed to `tokio::time`. This brings it in line with `std`.
* `tokio::timer::Timer` is renamed to `Driver` and made private.
* The `clock` module is removed. Instead, an `Instant` type is provided. This
type defaults to calling `std::time::Instant`. A `test-util` feature flag can
be used to enable hooking into time.
* The `blocking` module is moved to the top level and is cleaned up.
* The `task` module is moved to the top level.
* The thread-pool's in-place blocking implementation is cleaned up.
* `runtime::Spawner` is renamed to `runtime::Handle` and can be used to "enter"
a runtime context.
It is a more common use case that is to simply cause a delay for an amount of time.
I think it is more appropriate to show off `delay_for` in the example rather than `delay` that is useful only for less common use cases.
Fixes a bug in the thread-pool executor related to shutdown
concurrent with a task that is self-notifying. A `loom` test is
added to validate the fix.
Additionally, in anticipation of the `thread_pool` module being
switched to private, tests are updated to use `Runtime` directly
instead of `thread_pool`. Those tests that cannot be updated
are switched to unit tests.
The `Executor` trait is sub-optimal as it forces a `Box<dyn Future>` to
spawn. Instead, `tokio::spawn` delegates to the specific runtime
implementation set for the current execution context.
`TypedExecutor`, while useful, has seen limited adoption. As such, it is
removed from `tokio` proper. Moving it to `tokio-util` is a possibility
that can be explored as follow up work.
Simplify Tokio's runtime construct by combining both Runtime variants
into a single type. The execution style can be controlled by a
configuration setting on `Builder`.
The implication of this change is that there is no longer any way to
spawn `!Send` futures. This, however, is a temporary limitation. A
different strategy will be employed for supporting `!Send` futures.
Included in this patch is a rework of `task::JoinHandle` to support
using this type from both the thread-pool and current-thread executors.
The compat crate is moved to https://github.com/tokio-rs/tokio-compat.
This allows pinning it to specific revisions of the Tokio git
repository. The master branch is intended to go through significant
churn and it will be easier to update the compat layer in batches.
## Motivation
The `futures` crate's [`compat` module][futures-compat] provides
interoperability between `futures` 0.1 and `std::future` _future types_
(e.g. implementing `std::future::Future` for a type that implements the
`futures` 0.1 `Future` trait). However, this on its own is insufficient
to run code written against `tokio` 0.1 on a `tokio` 0.2 runtime, if
that code also relies on `tokio`'s runtime services. If legacy tasks are
executed that rely on `tokio::timer`, perform IO using `tokio`'s
reactor, or call `tokio::spawn`, those API calls will fail unless there
is also a runtime compatibility layer.
## Solution
As proposed in #1549, this branch introduces a new `tokio-compat` crate,
with implementations of the thread pool and current-thread runtimes that
are capable of running both tokio 0.1 and tokio 0.2 tasks. The compat
runtime creates a background thread that runs a `tokio` 0.1 timer and
reactor, and sets itself as the `tokio` 0.1 executor as well as the
default 0.2 executor. This allows 0.1 futures that use 0.1 timer,
reactor, and executor APIs may run alongside `std::future` tasks on the
0.2 runtime.
### Examples
Spawning both `tokio` 0.1 and `tokio` 0.2 futures:
```rust
use futures_01::future::lazy;
tokio_compat::run(lazy(|| {
// spawn a `futures` 0.1 future using the `spawn` function from the
// `tokio` 0.1 crate:
tokio_01::spawn(lazy(|| {
println!("hello from tokio 0.1!");
Ok(())
}));
// spawn an `async` block future on the same runtime using `tokio`
// 0.2's `spawn`:
tokio_02::spawn(async {
println!("hello from tokio 0.2!");
});
Ok(())
}))
```
Futures on the compat runtime can use `timer` APIs from both 0.1 and 0.2
versions of `tokio`:
```rust
use std::time::{Duration, Instant};
use futures_01::future::lazy;
use tokio_compat::prelude::*;
tokio_compat::run_03(async {
// Wait for a `tokio` 0.1 `Delay`...
let when = Instant::now() + Duration::from_millis(10);
tokio_01::timer::Delay::new(when)
// convert the delay future into a `std::future` that we can `await`.
.compat()
.await
.expect("tokio 0.1 timer should work!");
println!("10 ms have elapsed");
// Wait for a `tokio` 0.2 `Delay`...
let when = Instant::now() + Duration::from_millis(20);
tokio_02::timer::delay(when).await;
println!("20 ms have elapsed");
});
```
## Future Work
This is just an initial implementation of a `tokio-compat` crate; there
are more compatibility layers we'll want to provide before that crate is
complete. For example, we should also provide compatibility between
`tokio` 0.2's `AsyncRead` and `AsyncWrite` traits and the `futures` 0.1
and `futures` 0.3 versions of those traits. In #1549, @carllerche also
suggests that the `compat` crate provide reimplementations of APIs that
were removed from `tokio` 0.2 proper, such as the `tcp::Incoming`
future.
Additionally, there is likely extra work required to get the
`tokio-threadpool` 0.1 `blocking` APIs to work on the compat runtime.
This will be addressed in a follow-up PR.
Fixes: #1605Fixes: #1552
Refs: #1549
[futures-compat]: https://rust-lang-nursery.github.io/futures-api-docs/0.3.0-alpha.19/futures/compat/index.html
It is hard to maintain features list manually, so use cargo-hack's
`--each-feature` flag. And cargo-hack provides a workaround for an issue
that dev-dependencies leaking into normal build (`--no-dev-deps` flag),
so removed own ci tool.
Also, compared to running tests on all features, there is not much
advantage in running tests on each feature, so only the default features
and all features are tested.
If the behavior changes depending on the feature, we need to test it as
another job in CI.
I need this to implement SMTP pipelining checks. I mostly need to
flush my send buffer when the read buffer is empty before waiting for
the next command.
A step towards collapsing Tokio sub crates into a single `tokio`
crate (#1318).
The sync implementation is now provided by the main `tokio` crate.
Functionality can be opted out of by using the various net related
feature flags.
A step towards collapsing Tokio sub crates into a single `tokio`
crate (#1318).
The executor implementation is now provided by the main `tokio` crate.
Functionality can be opted out of by using the various net related
feature flags.
## Motivation
The `tokio_net::driver` module currently stores the state associated
with scheduled IO resources in a `Slab` implementation from the `slab`
crate. Because inserting items into and removing items from `slab::Slab`
requires mutable access, the slab must be placed within a `RwLock`. This
has the potential to be a performance bottleneck especially in the context of
the work-stealing scheduler where tasks and the reactor are often located on
the same thread.
`tokio-net` currently reimplements the `ShardedRwLock` type from
`crossbeam` on top of `parking_lot`'s `RwLock` in an attempt to squeeze
as much performance as possible out of the read-write lock around the
slab. This introduces several dependencies that are not used elsewhere.
## Solution
This branch replaces the `RwLock<Slab>` with a lock-free sharded slab
implementation.
The sharded slab is based on the concept of _free list sharding_
described by Leijen, Zorn, and de Moura in [_Mimalloc: Free List
Sharding in Action_][mimalloc], which describes the implementation of a
concurrent memory allocator. In this approach, the slab is sharded so
that each thread has its own thread-local list of slab _pages_. Objects
are always inserted into the local slab of the thread where the
insertion is performed. Therefore, the insert operation needs not be
synchronized.
However, since objects can be _removed_ from the slab by threads other
than the one on which they were inserted, removal operations can still
occur concurrently. Therefore, Leijen et al. introduce a concept of
_local_ and _global_ free lists. When an object is removed on the same
thread it was originally inserted on, it is placed on the local free
list; if it is removed on another thread, it goes on the global free
list for the heap of the thread from which it originated. To find a free
slot to insert into, the local free list is used first; if it is empty,
the entire global free list is popped onto the local free list. Since
the local free list is only ever accessed by the thread it belongs to,
it does not require synchronization at all, and because the global free
list is popped from infrequently, the cost of synchronization has a
reduced impact. A majority of insertions can occur without any
synchronization at all; and removals only require synchronization when
an object has left its parent thread.
The sharded slab was initially implemented in a separate crate (soon to
be released), vendored in-tree to decrease `tokio-net`'s dependencies.
Some code from the original implementation was removed or simplified,
since it is only necessary to support `tokio-net`'s use case, rather
than to provide a fully generic implementation.
[mimalloc]: https://www.microsoft.com/en-us/research/uploads/prod/2019/06/mimalloc-tr-v1.pdf
## Performance
These graphs were produced by out-of-tree `criterion` benchmarks of the
sharded slab implementation.
The first shows the results of a benchmark where an increasing number of
items are inserted and then removed into a slab concurrently by five
threads. It compares the performance of the sharded slab implementation
with a `RwLock<slab::Slab>`:
<img width="1124" alt="Screen Shot 2019-10-01 at 5 09 49 PM" src="https://user-images.githubusercontent.com/2796466/66078398-cd6c9f80-e516-11e9-9923-0ed6292e8498.png">
The second graph shows the results of a benchmark where an increasing
number of items are inserted and then removed by a _single_ thread. It
compares the performance of the sharded slab implementation with an
`RwLock<slab::Slab>` and a `mut slab::Slab`.
<img width="925" alt="Screen Shot 2019-10-01 at 5 13 45 PM" src="https://user-images.githubusercontent.com/2796466/66078469-f0974f00-e516-11e9-95b5-f65f0aa7e494.png">
Note that while the `mut slab::Slab` (i.e. no read-write lock) is
(unsurprisingly) faster than the sharded slab in the single-threaded
benchmark, the sharded slab outperforms the un-contended
`RwLock<slab::Slab>`. This case, where the lock is uncontended and only
accessed from a single thread, represents the best case for the current
use of `slab` in `tokio-net`, since the lock cannot be conditionally
removed in the single-threaded case.
These benchmarks demonstrate that, while the sharded approach introduces
a small constant-factor overhead, it offers significantly better
performance across concurrent accesses.
## Notes
This branch removes the following dependencies `tokio-net`:
- `parking_lot`
- `num_cpus`
- `crossbeam_util`
- `slab`
This branch adds the following dev-dependencies:
- `proptest`
- `loom`
Note that these dev dependencies were used to implement tests for the
sharded-slab crate out-of-tree, and were necessary in order to vendor
the existing tests. Alternatively, since the implementation is tested
externally, we _could_ remove these tests in order to avoid picking up
dev-dependencies. However, this means that we should try to ensure that
`tokio-net`'s vendored implementation doesn't diverge significantly from
upstream's, since it would be missing a majority of its tests.
Signed-off-by: Eliza Weisman <[email protected]>
This adds an extra spawned task during the thread-pool shutdown loom
test. This results in additional cases being tested, primarily tasks
being stolen.
A step towards collapsing Tokio sub crates into a single `tokio`
crate (#1318).
The `io` implementation is now provided by the main `tokio` crate.
Functionality can be opted out of by using the various net related
feature flags.
A step towards collapsing Tokio sub crates into a single `tokio`
crate (#1318).
The `net` implementation is now provided by the main `tokio` crate.
Functionality can be opted out of by using the various net related
feature flags.
Previously, support for `blocking` was done through a static `POOL` that
would spawn threads on demand. While this made the pool accessible at
all times, it made it hard to configure, and it was impossible to keep
multiple blocking pools.
This patch changes `blocking` to instead use a "default" global like the
ones used for timers, executors, and the like. There is now
`blocking::with_pool`, which is used by both thread-pool workers and the
current-thread runtime to ensure that a pool is available to tasks.
This patch also changes `ThreadPool` to spawn its worker threads on the
blocking pool rather than as free-standing threads. This is in
preparation for the coming in-place blocking work.
One downside of this change is that thread names are no longer
"semantic". All threads are named by the pool name, and individual
threads are not (currently) given names with numerical suffixes like
before.
Historically, logging has been added haphazardly. Here, we entirely
remove logging as none of it is particularly useful. In the future, we
will add tracing back in order to expose useful data to the user of
Tokio.
Related to #1318, Tokio APIs that are "less stable" are moved into a new
`tokio-util` crate. This crate will mirror `tokio` and provide
additional APIs that may require a greater rate of breaking changes.
As examples require `tokio-util`, they are moved into a separate
crate (`examples`). This has the added advantage of being able to avoid
example only dependencies in the `tokio` crate.
A step towards collapsing Tokio sub crates into a single `tokio`
crate (#1318).
The `timer` implementation is now provided by the main `tokio` crate.
The `timer` functionality may still be excluded from the build by
skipping the `timer` feature flag.
## Motivation
The `tokio_net` resources can be created outside of a runtime due to how tokio
has been used with futures to date. For example, this allows a `TcpStream` to be
created, and later passed into a runtime:
```
let stream = TcpStream::connect(...).and_then(|socket| {
// do something
});
tokio::run(stream);
```
In order to support this functionality, the reactor was lazily bound to the
resource on the first call to `poll_read_ready`/`poll_write_ready`. This
required a lot of additional complexity in the binding logic to support.
With the tokio 0.2 common case, this is no longer necessary and can be removed.
All resources are expected to be created from within a runtime, and should panic
if not done so.
Closes#1168
## Solution
The `tokio_net` crate now assumes there to be a `CURRENT_REACTOR` set on the
worker thread creating a resource; this can be assumed if called within a tokio
runtime. If there is no current reactor, the application will panic with a "no
current reactor" message.
With this assumption, all the unsafe and atomics have been removed from
`tokio_net::driver::Registration` as it is no longer needed.
There is no longer any reason to pass in handles to the family of `from_std` methods on `net` resources. `Handle::current` has therefore a more restricted private use where it is only used in `driver::Registration::new`.
Signed-off-by: Kevin Leimkuhler <[email protected]>
A step towards collapsing Tokio sub crates into a single `tokio`
crate (#1318).
The `fs` implementation is now provided by the main `tokio` crate. The
`fs` functionality may still be excluded from the build by skipping the
`fs` feature flag.
This patch is a ground up rewrite of the existing work-stealing thread
pool. The goal is to reduce overhead while simplifying code when
possible.
At a high level, the following architectural changes were made:
- The local run queues were switched for bounded circle buffer queues.
- Reduce cross-thread synchronization.
- Refactor task constructs to use a single allocation and always include
a join handle (#887).
- Simplify logic around putting workers to sleep and waking them up.
**Local run queues**
Move away from crossbeam's implementation of the Chase-Lev deque. This
implementation included unnecessary overhead as it supported
capabilities that are not needed for the work-stealing thread pool.
Instead, a fixed size circle buffer is used for the local queue. When
the local queue is full, half of the tasks contained in it are moved to
the global run queue.
**Reduce cross-thread synchronization**
This is done via many small improvements. Primarily, an upper bound is
placed on the number of concurrent stealers. Limiting the number of
stealers results in lower contention. Secondly, the rate at which
workers are notified and woken up is throttled. This also reduces
contention by preventing many threads from racing to steal work.
**Refactor task structure**
Now that Tokio is able to target a rust version that supports
`std::alloc` as well as `std::task`, the pool is able to optimize how
the task structure is laid out. Now, a single allocation per task is
required and a join handle is always provided enabling the spawner to
retrieve the result of the task (#887).
**Simplifying logic**
When possible, complexity is reduced in the implementation. This is done
by using locks and other simpler constructs in cold paths. The set of
sleeping workers is now represented as a `Mutex<VecDeque<usize>>`.
Instead of optimizing access to this structure, we reduce the amount the
pool must access this structure.
Secondly, we have (temporarily) removed `threadpool::blocking`. This
capability will come back later, but the original implementation was way
more complicated than necessary.
**Results**
The thread pool benchmarks have improved significantly:
Old thread pool:
```
test chained_spawn ... bench: 2,019,796 ns/iter (+/- 302,168)
test ping_pong ... bench: 1,279,948 ns/iter (+/- 154,365)
test spawn_many ... bench: 10,283,608 ns/iter (+/- 1,284,275)
test yield_many ... bench: 21,450,748 ns/iter (+/- 1,201,337)
```
New thread pool:
```
test chained_spawn ... bench: 147,943 ns/iter (+/- 6,673)
test ping_pong ... bench: 537,744 ns/iter (+/- 20,928)
test spawn_many ... bench: 7,454,898 ns/iter (+/- 283,449)
test yield_many ... bench: 16,771,113 ns/iter (+/- 733,424)
```
Real-world benchmarks improve significantly as well. This is testing the hyper hello
world server using: `wrk -t1 -c50 -d10`:
Old scheduler:
```
Running 10s test @ http://127.0.0.1:3000
1 threads and 50 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 371.53us 99.05us 1.97ms 60.53%
Req/Sec 114.61k 8.45k 133.85k 67.00%
1139307 requests in 10.00s, 95.61MB read
Requests/sec: 113923.19
Transfer/sec: 9.56MB
```
New scheduler:
```
Running 10s test @ http://127.0.0.1:3000
1 threads and 50 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 275.05us 69.81us 1.09ms 73.57%
Req/Sec 153.17k 10.68k 171.51k 71.00%
1522671 requests in 10.00s, 127.79MB read
Requests/sec: 152258.70
Transfer/sec: 12.78MB
```
As discussed in #1620, the attribute names for `#[tokio::main]` and
`#[tokio::test]` aren't great. Specifically, they both use
`single_thread` and `multi_thread`, as opposed to names that match the
runtime names: `current_thread` and `threadpool`. This PR changes the
former to the latter.
Fixes#1627.
`is_terminated` must return `true` until the future has been polled at least once to make sure that the associated block in select is called even after the delay has elapsed.
You use `Delay` in a `select!` by [fusing it](https://docs.rs/futures-preview/0.3.0-alpha.19/futures/future/trait.FutureExt.html#method.fuse):
```rust
let delay = tokio::timer::delay(/* ... */);
let delay = delay.fuse();
select! {
_ = delay => {
/* work here */
}
}
```
When polling the task, the current waker is saved to the oneshot state.
When the handle is migrated to a new task and polled again, the waker
must be swaped from the old waker to the new waker. In some cases, there
is a potential for the old waker to leak.
This bug was caught by loom with the recently added memory leak
detection.
Use a counter to count notifications. This protects against spurious
wakeups by pthreads and other libraries. The state transitions now
track num_idle precisely.
The standard library's `io` module has small utilities such as `repeat`,
`empty`, and `sink`, which return `Read` and `Write` implementations.
These can come in handy in some circiumstances. `tokio::io` has no
equivalents that implement `AsyncRead`/`AsyncWrite`.
This commit adds `repeat`, `empty`, and `sink` helpers to `tokio::io`.
In the past, it was not possible to choose to use the multi-threaded
tokio `Runtime` in tests, which meant that any test that transitively
used `executor::threadpool::blocking` would fail with
```
'blocking' annotation used from outside the context of a thread pool
```
This patch adds a runtime annotation attribute to `#[tokio::test]` just
like `#[tokio::main]` has, which lets users opt in to the threadpool
runtime over `current_thread` (the default).
The algorithm backing `AtomicWaker` effectively uses a spin lock backed
by notifying & yielding the current task. This adds a `spin_lock_hint`
annotation to cover this case.
While, in practice, the omission of `spin_lock_hint` would not cause
problems, there are platforms that do not handle spin locks very well
and could enter a deadlock in pathological cases.
- Adds a minimum `rt-current-thread` optional feature that exports
`tokio::runtime::current_thread`.
- Adds a `macros` optional feature to enable the `#[tokio::main]` and
`#[tokio::test]` attributes.
- Adjusts `#[tokio::main]` macro to select a runtime "automatically" if
a specific strategy isn't specified. Allows using the macro with only
the rt-current-thread feature.
* Removes most pin-projection related unsafe code.
* Removes manual Unpin implementations.
As references always implement Unpin, there is no need to implement
Unpin manually.
* Adds tests to check that Unpin requirement does not change accidentally
because changing Unpin requirements will be breaking changes.
`BufWriter` and `BufReader` did not previously forward the "opposite" trait (`AsyncRead` for `BufWriter` and `AsyncWrite` for `BufReader`). This meant that there was no way to have both directions buffered at once. This patch fixes that, and introduces a convenience type + constructor for this double-wrapped construct.
This adds `Barrier` to `tokio-sync`, which is an asynchronous alternative to [`std::sync::Barrier`](https://doc.rust-lang.org/std/sync/struct.Barrier.html). It is a synchronization primitive that allows multiple futures to "rendezvous" at certain points in their execution.
Currently, when threads in the blocking pool shutdown due to being idle
the counter tracking threads is not decremented. This prevents new threads
from being spawned to replace the shutdown threads.
This renames `Lock` to `Mutex`, and brings the API more in line with `std::sync::Mutex`.
In partcular, locking now only takes `&self`, with the expectation that you place the `Mutex` in an `Arc` (or something similar) to share it between threads.
Fixes#1544.
Part of #1210.
Bring back `split` utility as a free fn instead of a method on
`AsyncRead`. This utility wraps the `stream` in an `Arc` and uses mutual
exclusion to ensure correct access.
Additionally, the specialized `split_mut` fn on TcpStream and UdsStream
is promoted to `split`.
A sealed `net::ToSocketAddrs` trait is added. This trait is not intended
to be used by users. Instead, it is an argument to `connect` and `bind`
functions.
The operating system's DNS lookup functionality is used. Blocking
operations are performed on a thread pool in order to avoid blocking the
runtime.
* net: switch from `log` to `tracing`.
Motivation:
The `tracing` crate implements scoped, structured, context-aware
diagnostics, which can add significant debugging value over unstructured
log messages. `tracing` is part of the Tokio project. As part of the
`tokio` 0.2 changes, I thought it would be good to move over from `log`
to `tracing` in the tokio runtime.
Solution:
This branch replaces the use of `log` in `tokio-net` with
`tracing`. I've tried to leave all the instrumentation points more or
less the same, but modified to use structured fields instead of string
interpolation.
Notes:
I removed the timing in `Reactor::poll` in favor of simply adding a
`#[tracing::instrument]` attribute. Since the generated `tracing` span
will have enter and exit events, a `tracing::Subscriber`
implemementation can use those to record timestamps, and process that
timing data in a much more sophisticated manner than including it in a
log line.
We can add the timestamps back if they're desired.
Signed-off-by: Eliza Weisman <[email protected]>
This code directly relies on `future-preview`'s `RemoteHandle`, and
exposes it via a `spawn_with_handle` method that is identical to
`future-preview`'s implementation.
Related: #1180
Provides a thread pool dedicated to running blocking operations (#588)
and update `tokio-fs` to use this pool.
In an effort to make incremental progress, this is an initial step
towards a final solution. First, it provides a very basic pool
implementation with the intend that the pool will be
replaced before the final release. Second, it updates `tokio-fs` to
always use this blocking pool instead of conditionally using
`threadpool::blocking`. Issue #588 contains additional discussion around
potential improvements to the "blocking for all" strategy.
The implementation provided here builds on work started in #954 and
continued in #1045. The general idea is th same as #1045, but the PR
improves on some of the details:
* The number of explicit operations tracked by `File` is reduced only to
the ones that could interact. All other ops are spawned on the
blocking pool without being tracked by the `File` instance.
* The `seek` implementation is not backed by a trait and `poll_seek`
function. This avoids the question of how to model non-blocking seeks
on top of a blocking file. In this patch, `seek` is represented as an
`async fn`. If the associated future is dropped before the caller
observes the return value, we make no effort to define the state in
which the file ends up.
## Motivation
The `tracing` crate implements scoped, structured, context-aware
diagnostics, which can add significant debugging value over unstructured
log messages. `tracing` is part of the Tokio project. As part of the
`tokio` 0.2 changes, I thought it would be good to move over from `log`
to `tracing` in the tokio runtime. Updating the executor crate is an obvious
starting point.
## Solution
This branch replaces the use of `log` in `tokio-executor` with
`tracing`. I've tried to leave all the instrumentation points more or
less the same, but modified to use structured fields instead of string
interpolation. I've also added a few `tracing` spans, primarily in
places where a variable is added to all the log messages in a scope.
## Notes
For users who are using the legacy `log` output, there is a feature flag
to enable `log` support in `tracing`. I thought about making this on by
default, but that would also enable the `tracing` dependency by default,
and it is only pulled in when the `threadpool` feature flag is enabled.
The `tokio` crate could enable the log feature in its default features
instead, since the threadpool feature is on by default in `tokio`. If
this isn't the right approach, I can change how `log` back-compatibility
is enabled.
We might want to consider adding more `tracing` spans in the threadpool
later. This could be useful for profiling, and for helping users debug
the way their applications interact with the executor. This branch is
just intended as a starting point so that we can begin emitting
`tracing` data from the executor; we should revisit what instrumentation
should be exposed, as well.
Signed-off-by: Eliza Weisman <[email protected]>
Since `tokio_sync::oneshot` makes a `CausalCell::with_mut()` mutable
access in the `Inner::drop()`, we must use the instrumented
`loom::sync::Arc`.
Uncovered by carllerche/loom#42
This avoids having consumers import libc for common signals, and it
improves discoverability since users need not be aware that libc
contains all supported constants.
Windows guarantees handler routines are always invoked in a new thread
(https://docs.microsoft.com/en-us/windows/console/handlerroutine), so we
don't need to use the handler-wake-another-driver technique used in the
Unix implementation
By broadcasting the event notifications from the handler, we no longer
need the Driver task to be spawned, which fixes the starvation issue if
the executor which runs the Driver task goes away
Also changed the behavior so that the default event handler runs if
all listeners for CTRL_{C, BREAK} events go away.
How the buffer is managed is often critical for performance. Not
taking care of it will be catastrophic for performance beyond the
initial buffer size with the current implementation (a loop of
`reserve(1)`).
* Denied all warnings in tests, and denied rust_2018_idioms violations
* Bumped the crate version and set publish = false
* Pruned dependencies:
- Only pull in tokio-sync on windows where it is used
- Removed unused dev-dependencies
* Switch to Async{Read, Write} traits from tokio-io rather than
futures-io
* Use #[tokio::test] where possible
* Removed deprecated items
* Fix all doc examples
* TryFrom<net::TcpListener> for TcpListener
* TryFrom<net::TcpStream> for TcpStream
* TryFrom<net::UdpSocket> for UdpSocket
* TryFrom<net::UnixDatagram> for UnixDatagram
* TryFrom<net::UnixListener> for UnixListener
* TryFrom<net::UnixStream> for UnixStream
* TryFrom<UnixDatagram> for mio_uds::UnixDatagram
* TryFrom<File> for io::File
* From<io::File> for File
* This simplifies the API surface by returning () instead of the signal
number that was used during registration. This also more closely mirrors
the cross-platform `CtrlC` event stream API
* This is a **breaking change**
* Add a new `windows::CtrlBreak` struct which wil represent a stream of
CTRL_BREAK_EVENT signals on Windows systems
* The `windows::Event` type is no longer publicly accessible and is
replaced by using `CtrlC` or `windows::CtrlBreak`.
[breaking-change]
Removes the `Send` requirement to futures passed to `Runtime::block_on`.
Previously, `block_on` was implemented by sending the future to a
runtime thread. In order to do this, the future must be Send.
The reason why the future is sent to the pool is because we cannot
guarantee, while off the pool, that a reactor / timer thread is running.
This is due to a limitation in the current version of tokio-threadpool.
There is a plan to fix this (#1177), but the proper fix is non trivial.
In order to unblock APIs that require this, this patch updates the
runtime to spawn an always running thread containing a reactor and
timer. All calls to `block_on` will use that reactor and timer.
We want executors to enforce that there are never multiple active at the
same time. This is ensured through `Enter`, which will panic if you
attempt to create more than one. However, by requiring you to pass an
`&mut Enter` to `executor::with_default`, we were *also* disallowing
temporarily overriding the current executor.
This patch removes that requirement.
* Add a new `CtrlC` struct which will represent a stream of SIGINT
signals on Unix or the CTRL_C event on Windows
* `CtrlC` implements `Stream<Output = ()>` rather than `IoSteam` as
previously
Migrate to std::futures and the futures 0.3 preview and use async/await
where possible
**Breaking change:** the IoFuture and IoStream definitions used to refer
to Box<dyn Future> and Box<dyn Stream>, but now they are defined as
Pin<...> versions which are technically breaking.
No other breaking or functional changes have been made
## Motivation
Currently, the process for releasing a new version of a Tokio crate is
somewhat complex, and is not well-documented. To make it easier for
contributors to release minor versions more frequently, there should be
documentation describing this process.
## Solution
This branch adds a section to `CONTRIBUTING.md` describing how to
release a new version of a Tokio crate. The steps are based on those
described by @carllerche in an offline conversation.
I've also added a quick shell script to actually publish new crate
versions. This should make it harder to make mistakes when
publishing.
Signed-off-by: Eliza Weisman <[email protected]>
* timer: restructure feature flags
* update timer tests
* Add `async-traits` to CI
This also disables a buggy `threadpool` test. This test should be fixed in the future.
Refs #1225
## Motivation
The `tokio-trace` and `tokio-trace-core` crates have been renamed to
`tracing` and `tracing-core`, and moved to their own repository
(`tokio-rs/tracing`).
## Solution
This branch removes `tokio-trace` and `tokio-trace-core` from the
`tokio` repository. In addition, I've added a "Related Projects" section
to the root README, which lists `tracing` (as well as `mio`, and
`bytes`) as other libraries maintained by the Tokio project. I thought
that this would help folks looking for `tokio-trace` here find it in its
new home.
In addition, it changes `tokio` to depend on `tracing-core` rather than
`tokio-trace-core`.
Closes#1159
Signed-off-by: Eliza Weisman <[email protected]>
An initial pass at updating `tokio-threadpool` to `std::future`. The
codebase and tests both now run using `std::future` but the wake
mechanism is not ideal. Follow up work will be required to improve on
this.
Refs: #1200
Strategy was to
- copy the old codec code that was temporarily being stashed in `tokio-io`
- modify all the type signatures to use Pin, as literal a translation as possible
- fix up the tests likewise
This is intended just to get things compiling and passing tests. Beyond that there is surely
lots of refactoring that can be done to make things more idiomatic. The docs are unchanged.
Closes#1189
* `TryFrom<TcpListener> for mio::net::TcpListener`
* `TryFrom<TcpStream> for mio::net::TcpStream`
* `TryFrom<UdpSocket> for mio::net::UdpSocket`
* `TryFrom<UnixListener> for mio_uds::UnixListener`
* `TryFrom<UnixStream> for mio_uds::UnixStream`
## Motivation
As mentioned in tokio-rs/tracing#1100 it makes sense to be able to set
the parents of events explicitly.
## Solution
For that to happen the Parent type is extracted from span.rs and a
`parent` field is added to Event. Additionally the appropriate macros
arms are added with corresponding tests as described in
tokio-rs/tracing#1100Closestokio-rs/tracing#1100
Signed-off-by: Zahari Dichev <[email protected]>
Today the Unix and Windows implementations have similar yet differing
implementations of hooking into OS events and propagating them to any
listening futures. Rather than re-implement the same behavior two
different ways, we should factor out any commonality into a shared
module and keep the Unix/Windows modules focused solely on OS
integrations.
Reusing the same implementation across OS versions also allows for more
consistent behavior between platforms, which also makes squashing bugs
much easier.
This change introduces the `registry` module which handles creating and
initializing a global map of signals/events and their registered
listeners. Each OS specific module is expected to implement the OS hooks
which delegate to invoking the registry module's methods for
distributing the event notifications.
# Use registry module for Windows implementation
Note this still uses the same architecture as previously: a driver task
is spawned by the first registered event, and that task is responsible
for delivering any events to registered futures. (If that first event
loop goes away, all events will deadlock). A solution to this issue will
be explored at a later time.
A first pass at updating Tokio to use `std::future`.
Implementations of `Future` from the futures crate are updated to implement
`Future` from std. Implementations of `Stream` are moved to a feature flag.
This commits disables a number of crates that have not yet been updated.
PR #1103 accidentally changed the log level for the debug and
debug_span macros to use the INFO level instead of the DEBUG
level. This PR corrects this regression back to the intended
behavior.
## Motivation
I was just trying to use tokio-trace for a greenfield project, but I was frustrated to discover that I couldn't really use it easily.
I was using the [`runtime`](https://docs.rs/runtime/0.3.0-alpha.4/runtime/) crate, which transparently spawns a thread pool executor for futures. In that thread pool, there's no way to set a tokio-trace subscriber for the duration of each thread, since you don't control the thread initialization. You *might* be able to wrap every future you spawn with a subscriber call, but that's a lot of work.
I was also confused because the documentation said that setting a subscriber in the main thread would use that subscriber for the rest of the program. That isn't the case, though -- the subscriber will be used only on the main thread, and not on worker threads, etc.
## Solution
I added a function `set_global_default`, which works similarly to the `log` crate:
```rust
tokio_trace::subscriber::set_global_default(FooSubscriber::new());
```
The global subscriber (actually a global `Dispatch`) is a `static mut` protected by an atomic; implementation is copied from the `log` crate. It is used as a fallback if a thread has no `Dispatch` currently set. This is extremely simple to use, and doesn't break any existing functionality.
Performance-wise, thread-local `Dispatch` lookup goes from ~4.5ns to ~5ns, according to the benchmarks. So, barely any runtime overhead. (Presumably there's a little compile-time overhead but idk how to measure that.) Since the atomic guard is only ever written once, it will be shared among a CPU's cores and read very cheaply.
I added some docs to partially address #1151. I also switched the tokio-trace benchmarks to criterion because the nightly benchmarks weren't compiling (missing `dyn` flags?)
Currently, when the `trace_span!`, `debug_span!`, `info_span!`,
`warn_span!`, and `error_span!` macros are invoked with an explicit
parent, a name, and zero or more fields (no target), the macros don't
pass along the explicitly provided parent when expanding to the `span!`
macro. This is likely due to an oversight on my part.
This branch fixes these macros by adding the parent into the `span!`
macro expansion. I've also added a test to catch regressions
Shoutout to @jonhoo for catching this one!
Signed-off-by: Eliza Weisman <[email protected]>
While we're making breaking changes to `tokio-trace`, it would be good
to get rid of the `AsId` trait. The goal of span functions that are
generic over `Span`/`Id` can be achieved without the unnecessary
complexity of defining a new trait. This would also make the API added
to `tokio_trace_core::Event` in #1109 more consistent with the
`tokio-trace::Span` API.
This branch removes `AsId` from `tokio-trace` and replaces its uses with
`impl Into<Option<Id>>` and `impl Into<Option<&'a Id>>`. While `AsRef`
might be more semantically correct for the borrowed-`Id` conversion, its
signature doesn't permit conversion into an `Option`. Implementations of
`Into<Option<Id>>` and `Into<Option<&'a Id>>` have been added for
`tokio_trace::Span`.
This is _technically_ a breaking API change, as it changes function
signatures. However, the existing macro syntax still works as-is, and
the tests which pass `&Id`, `&Span`, and `&Option<Id>` to the span
macros all still compile after this change.
Closes#1143
Signed-off-by: Eliza Weisman <[email protected]>
The runtime is inherently multi-threaded, so it's going to have to deal
with synchronization when submitting new tasks anyway. This allows a
runtime to be shared by multiple threads more easily when e.g. building
a blocking facade over a tokio-based API.
## Motivation
A common pattern in `tokio-trace` is to use the value of a local
variable as a field on a span or event. Currently, this requires code
like:
```rust
info!(foo = foo);
```
which is not particularly ergonomic given how commonly this occurs.
Struct initializers support a shorthand syntax for fields where the name
of the field is the same as a local variable, and `tokio-trace` should
as well.
## Solution
This branch adds support for syntax like
```rust
let foo = ...;
info!(foo);
```
and
```rust
let foo = Foo {
bar: ...,
...
};
info!(foo.bar)
```
to the `tokio-trace` span and event macros. This syntax also works with
the `Debug` and `Display` field shorthand.
The span macros previously used a field name with no value to indicate
an uninitialized field. A new issue, #1138, has been opened for finding a
replacement syntax for uninitialized fields. Until then, the `tokio-trace`
macros will no longer provide a way to create fields without values,
although the `-core` API will continue to support this.
Closes#1062
Signed-off-by: Eliza Weisman <[email protected]>
## Motivation
Currently, the primary way to use a span is to use `.enter` and pass a
closure to be executed under the span. While that is convenient in many
settings, it also comes with two decently inconvenient drawbacks:
- It breaks control flow statements like `return`, `?`, `break`, and
`continue`
- It require re-indenting a potentially large chunk of code if you wish
it to appear under a span
## Solution
This branch changes the `Span::enter` function to return a scope guard
that exits the span when dropped, as in:
```rust
let guard = span.enter();
// code here is within the span
drop(guard);
// code here is no longer within the span
```
The method previously called `enter`, which takes a closure and
executes it in the span's context, is now called `Span::in_scope`, and
was reimplemented on top of the new `enter` method.
This is a breaking change to `tokio-trace` that will be part of the
upcoming 0.2 release.
Closes#1075
Signed-off-by: Eliza Weisman <[email protected]>
This updates tests to track a fix applied in Mio. Previously, Mio
incorrectly fired HUP events. This was due to Mio mapping `RDHUP` to
HUP. The test is updated to correctly generate a HUP event.
Additionally, HUP events will be removed from all platforms except for
Linux. This is caused by the inability to reliably map kqueue events to
the epoll HUP behavior.
## Motivation
In `tokio-trace`, field values may be recorded as either a subset of
Rust primitive types or as `fmt::Display` and `fmt::Debug`
implementations. Currently, `tokio-trace` provides the `field::display`
and `field::debug` functions which wrap a type with a type that
implements `Value` using the wrapped type's `fmt::Display` or
`fmt::Debug` implementation. However, importing and using these
functions adds unnecessary boilerplate.
In #1081, @jonhoo suggested adding shorthand syntax to the macros,
similar to that used by the `slog` crate, as a solution for the
wordiness of the current API.
## Solution
This branch adds `?` and `%` sigils to field values in the span and
event macros, which expand to the `field::debug` and `field::display`
wrappers, respectively. The shorthand sigils may be used in any position
where the macros take a field value.
For example:
```rust
trace_span!("foo", my_field = ?something, ...); // shorthand for `debug`
info!(foo = %value, bar = false, ...) // shorthand for `display`
```
Adding this shorthand required a fairly large change to how field
key-value pairs are handled by the macros --- since `%foo` and `%foo`
are not valid Rust expressions, we can no longer match repeated
`$ident = $expr` patterns, and must now match field lists as repeated
token trees. The inner helper macros for constructing `FieldSet`s and
`ValueSet`s have to parse the token trees recursively. This added a
decent chunk of complexity, but fortunately we have a large number of
compile tests for the macros and I'm quite confident that all existing
invocations will still work.
Closes#1081
Signed-off-by: Eliza Weisman <[email protected]>
Callers may not always have `futures` available at the root of the
crate. Re-exporting dependencies makes them available to the macro at a
deterministic location.
The signal-hook library got split into lower-level and higher-level
parts. The tokio-signal uses only API from the lower-level one, so it
can depend on it directly.
The only effect of this change is smaller amount of compiled (and
unused) code during compilation. There's no change in the code actually
used.
This PR introduces `Lock`: A concurrency primitive built on top of `Semaphore` that provides a `Mutex`-like primitive that interacts nicely with futures. Specifically, `LockGuard` (in contrast to `MutexGuard`) does _not_ borrow the `Lock`, and can thus be passed into a future where it will later be unlocked.
This replaces #958, which attempted to introduce a less generic version. The primitive proposed there will instead live in [`async-lease`](https://github.com/jonhoo/async-lease).
This branch changes `dispatcher::get_default` to unset the thread's
current dispatcher while the reference to it is held by the closure.
This prevents infinite loops if the subscriber calls code paths which
emit events or construct spans.
Note that this also means that nested calls to `get_default` inside of a
`get_default` closure will receive a `None` dispatcher rather than the
"actual" dispatcher. However, it was necessary to unset the default in
`get_default` rather than in dispatch methods such as `Dispatch::enter`,
as when those functions are called, the current state has already been
borrowed.
Before:
```
test enter_span ... bench: 3 ns/iter (+/- 0)
test span_no_fields ... bench: 51 ns/iter (+/- 12)
test span_repeatedly ... bench: 5,073 ns/iter (+/- 1,528)
test span_with_fields ... bench: 56 ns/iter (+/- 49)
test span_with_fields_record ... bench: 363 ns/iter (+/- 61)
```
After:
```
test enter_span ... bench: 3 ns/iter (+/- 0)
test span_no_fields ... bench: 35 ns/iter (+/- 12)
test span_repeatedly ... bench: 4,165 ns/iter (+/- 298)
test span_with_fields ... bench: 48 ns/iter (+/- 12)
test span_with_fields_record ... bench: 363 ns/iter (+/- 91)
```
Closes#1032
Signed-off-by: Eliza Weisman <[email protected]>
* Fix crate path in `Cargo.toml` of examples
* Add `edition2018` to examples in the documentation to make it compiled
on Rust 2018
* Fix an example in the documentation
## Motivation
To ease the implementation of `Subscriber::register_callsite`, a field
should be added to `Metadata` to indicate if this callsite is an event or
a span.
## Solution
A new struct, `Kind`, is added to the `metadata` module in
`tokio-trace-core`, and a `Kind` field is added to the `Metadata`
struct. Macros which construct `metadata` now require a `Kind`.
`Kind` is represented as a struct with a private inner enum to allow new
`Kind`s to be added without breaking changes. However, the _addition_ of
the kind field _is_ a breaking change. While this could be done in a
backward-compatible way, it would permit the construction of metadata
with unknown kinds, and since the next `tokio-trace-core` release will
be a breaking change, I opted to make the breaking change instead.
New API tests for the `callsite!` and `metadata!` macros have been added
to guard against future API breakage.
Fixes: #986Closes: #1008
Co-Authored-By: csmoe <[email protected]>
## Motivation
Currently, `tokio-trace-core` permits `Subscriber`s to indicate that
they are "always", "sometimes", or "never" interested in a particular
callsite. When "always" or "never" is returned, then the interest is
cached and the subscriber will not be asked again about that callsite.
This is much more efficient than requiring the filter to be re-evaluated
every time the callsite is hit.
However, if a subscriber wishes to change its filter configuration
dynamically at runtime, it cannot benefit from this caching. Instead, it
must always return `Interest::sometimes`. Even when filters change very
infrequently, they must still always be re-evaluated every time.
In order to support a use-case where subscribers may change their filter
configuration at runtime (e.g. tokio-rs/tokio-trace-nursery#42),
but do so infrequently, we should introducing a new function to
invalidate the cached interest.
## Solution
This branch adds a new function in the `callsite` module, called
`rebuild_interest_cache`, that will invalidate and rebuild all cached
interest.
## Breaking Change
In order to fix a race condition that could occur when rebuilding
interest caches using `clear_interest` and `add_interest`, these methods
have been replaced by a new `set_interest` method. `set_interest` should
have the semantics of atomically replacing the previous cached interest,
so that the callsite does not enter a temporary state where it has no
interest.
Closes#1038
Co-Authored-By: yaahallo <[email protected]>
## Motivation
The `Span::enter` function previously required an `&mut` reference to
enter a span. This is a relic of an earlier design where span closure
logic was determined by dropping an inner span component, and is no
longer strictly necessary.
Requiring `&mut self` to enter a span leads to awkward patterns in cases
when a user wishes to enter a span and then call methods on the span
(such as recording field values). For example, we cannot say
```rust
let mut span = span!("foo", bar);
span.enter(|| {
span.record("bar" &false);
});
```
since the span is mutably borrowed by `enter`. Instead, we must clone
the span, like so:
```rust
let mut span = span!("foo", bar);
span.clone().enter(|| {
span.record("bar" &false);
});
```
Having to clone the span is somewhat less ergonomic, and it has
performance disadvantages as well: cloning a `Span` will clone the
span's `Dispatch` handle, requiring an `Arc` bump, as well as calling
the `Subscriber`'s `clone_span` and `drop_span` functions. If we can
enter spans without a mutable borrow, we don't have to update any of
these ref counts.
The other reason we may wish to require mutable borrows to enter a span
is if we want to disallow entering a span multiple times before exiting
it. However, it is trivially possible to re-enter a span on the same
thread regardless, by cloning the span and entering it twice. Besides,
there may be a valuable semantic meaning in entering a span from inside
itself, such as when a function is called recursively, so disallowing
this is not a goal.
## Solution
This branch rewrites the `Span::enter`, `Span::record`, and
`Span::record_all` functions to no longer require mutable borrows.
In the case of `record` and `record_all`, this was trivial, as borrowing
mutably was not actually *necessary* for those functions. For `enter`,
the `Entered` guard type was reworked to consist of an `&'a Inner`
rather than an `Inner`, so it is no longer necessary to `take` the
span's `Inner`.
## Notes
In addition to allowing spans to be entered without mutable borrows,
`Entered` was changed to exit the span automatically when the guard is
dropped, so we may now observe correct span exits even when unwinding.
Furthermore, this allows us to simplify the `enter` function a bit,
leading to a minor performance improvement when entering spans.
Before:
```
test enter_span ... bench: 13 ns/iter (+/- 1)
```
...and after:
```
test enter_span ... bench: 3 ns/iter (+/- 1)
```
Note that this branch also contains a change to make the
`subscriber::enter_span` benchmark more accurate. Previously, this
benchmark constructed a new span inside of `b.iter(|| {...})`. This
means that the benchmark was measuring not only the time taken to enter
a span, but the time taken to construct a `Span` handle as well.
However, we already have benchmarks for span construction, and the
intention of this particular benchmark was to measure the overhead of
constructing a span.
I've updated the benchmark by moving the span construction out of the
`iter` closure. Now, the span is constructed a single time and entered
on every iteration. This allows us to measure only the overhead of
actually entering a span. The "before" benchmark numbers above were
recorded after backporting this change to master, so they are "fair" to
the previous implementation. Prior to this change the benchmark took
approximately 53 ns.
Signed-off-by: Eliza Weisman <[email protected]>
## Motivation
In order to support conventions that add namespacing to `tokio-trace`
field names, it's necessary to accept at least one type of separator
character. Currently, the `tokio-trace` macros only accept valid Rust
identifiers, so there is no clear separator character for namespaced
conventions. See also #1018.
## Solution
This branch changes the single `ident` fragment matcher for field names
to match *one or more* `ident` fragments separated by `.` characters.
## Notes
The resulting key is still exposed to `tokio-trace-core` as a string
constant created by stringifying the dotted expression. However, if
`tokio-trace-core` were later to adopt a first class notion of
hierarchical field keys, we would be able to track that change in
`tokio-trace` as an implementation detail.
Closes#1018.
Closes#1022.
Signed-off-by: Eliza Weisman <[email protected]>
This branch fixes the `tokio-trace` Subscriber benchmarks panicking due
to constructing spans with ID 0. They will now use an arbitrary constant
instead.
Signed-off-by: Eliza Weisman <[email protected]>
## Motivation
Was determined that having the span! macro default to the TRACE level is
probably not ideal (see discussion on #952).
Closes#1013
## Solution
Remove default trace level and make log lvl mandatory on span! macro,
and add the respective `trace_span!`, `debug_span!`, `info_span!`,
`warn_span!` and `error_span!` macros that behave as span! macro, but
with defined log levels
## Notes
I think this is it, also removed some captures that were repeated, and
some testcases that also seemed repeated after adding the mandatory log
level, but please review it, if more tests or examples are needed happy
to provide (tried to find a way to get the generated macros log level,
but didn't find one, if there is a way i can add tests to assert that
the generated macro has the matching log level ). thanks
- Use `Handle::default` over `Handle::current` for consistent semantics
- Make all `windows::Event` constructors lazily invoke `global_init`
so they can be safely constructed off-task
- Don't assume the reactor is alive and event registration will be done
when calling `global_init`
Add windows regression tests. Unfortunately, Windows doesn't have a
reliable way of programmatically sending CTRL_C or CTRL_BREAK events
to a progress, so the tests can only exercise our internal machinery by
invoking the handler that we register with the OS
Fixes#999
This branch modifies the `tokio_trace::Span` API functions that take
span IDs (the `Span::child_of` constructor, and the `Span::follows_from`
method) so that more types bearing a span ID can be passed as an
argument. Span IDs may now be passed directly without requiring them to
be passed as `Some(id)`. This should make the API slightly more
ergonomic.
Also, it changes the `Span::field` method to take an `AsField` rather
than a `Borrow<str>`.
Signed-off-by: Eliza Weisman <[email protected]>
This branch improves the `fmt::Debug` implementation for `Metadata`,
and adds `fmt::Display` implementations for `FieldSet` and `ValueSet`.
When formatting a `Metadata`, only present fields are formatted --- if
optional fields, such as the file, line number, and module path are
`None`, they will be excluded. In addition, `Metadata` now formats its
`FieldSet` using `FieldSet`'s `fmt::Display` implementation, which is a
bit less noisy. Finally, the `Debug` output for `Metadata` now includes
the callsite that the metadata originates from.
The intention behind these changes is to make the output from failed
tests somewhat easier to interpret.
Signed-off-by: Eliza Weisman <[email protected]>
## Motivation
`tokio-trace` currently offers a strategy for compatibility with the
`log` crate: its macros can be dropped in as a replacement for `log`'s
macros, and a subscriber can be used that translates trace events to log
records. However, this requires the application to be aware of
`tokio-trace` and manually set up this subscriber.
Many libraries currently emit `log` records, and would like to be able
to emit `tokio-trace` instrumentation instead. The `tokio` runtimes are
one such example. However, with the current log compatibility strategy,
replacing existing logging with trace instrumentation would break
`tokio`'s logs for any downstream user which is using only `log` and not
`tokio-trace`. It is desirable for libraries to have the option to emit
both `log` _and_ `tokio-trace` diagnostics from the same instrumentation
points.
## Solution
This branch adds a `log` feature flag to the `tokio-trace` crate, which
when set, causes `tokio-trace` instrumentation to emit log records as well
as `tokio-trace` instrumentation.
## Notes
In order to allow spans to log their names when they are entered and
exited even when the span is disabled, this branch adds an
`&'static Metadata` to the `Span` type. This was previously stored in
the `Inner` type and was thus only present when the span was enabled.
This makes disabled spans one word longer, but enabled spans remain
the same size.
Fixes: #949
Signed-off-by: Eliza Weisman <[email protected]>
## Motivation
`tokio-trace` should have static verbosity level filtering, like the
`log` crate. The static max verbosity level should be controlled at
compile time with a set of features. It should be possible to set a
separate max level for release and debug mode builds.
## Solution
We can do this fairly similarly to how the `log` crate does it:
`tokio-trace` should export a constant whose value is set based on the
static max level feature flags. Then, we add an if statement to the
`span!` and `event!` macros which tests if that event or span's level
is enabled.
Closes#959
## Motivation
In order to implement "out of band" `Subscriber` APIs in third-party
subscriber implementations (see [this comment]) users may want to
downcast the current `Dispatch` to a concrete subscriber type.
For example, in a library for integrating `tokio-trace` with a fancy new
(hypothetical) distributed tracing technology "ElizaTracing", which uses
256-bit span IDs, we might expect to see a function like this:
```rust
pub fn correlate(tt: tokio_trace::span::Id, et: elizatracing::SpanId) {
tokio_trace::dispatcher::with(|c| {
if let Some(s) = c.downcast_ref::<elizatracing::Subscriber>() {
s.do_elizatracing_correlation_magic(tt, et);
}
});
}
```
This allows users to correlate `tokio-trace` IDs with IDs in the
distributed tracing system without having to pass a special handle to
the subscriber through application code (as one is already present in
thread-local storage, but with its type erased).
## Solution
This branch makes the following changes:
* Add an object-safe `downcast_raw` method to the `Subscriber` trait,
taking a `TypeId` and returning an `*const ()` if the type ID
matches the subscriber's type ID, or `None` if it does not, and
* Add `is<T>` and `downcast_ref<T>` functions to `Subscriber`
and `Dispatch`, using `downcast_raw`.
Unlike the approach implemented in #950, the `downcast_raw` method is
object-safe, since it takes a `TypeId` rather than a type _parameter_
and returns a void pointer rather than an `&T`. This means that
`Subscriber` implementations can override this method if necessary. For
example, a `Subscriber` that fans out to multiple component subscribers
can downcast to their component parts, and "chained" or "middleware"
subscribers, which wrap an inner `Subscriber` and modify its behaviour
somehow, can downcast to the inner type if they choose to.
[this comment]: https://github.com/tokio-rs/tokio/issues/932#issuecomment-469473501
[`std::error::Error`'s]: https://doc.rust-lang.org/1.33.0/src/std/error.rs.html#204
Refs: #950, #953, https://github.com/tokio-rs/tokio/issues/948#issuecomment-469444293
Signed-off-by: Eliza Weisman <[email protected]>
* chore: Fix examples not working with `cargo run`
## Motivation
PR #991 moved the `tokio` crate to its own subdirectory, but did not
move the `examples` directory into `tokio/examples`. While attempting to
use the examples for testing another change, I noticed that #991 had
broken the ability to use `cargo run`, as the examples were no longer
considered part of a crate that cargo was aware of:
```
tokio on master [$] via 🦀v1.33.0 at ☸️ aks-eliza-dev
➜ cargo run --example chat
error: no example target named `chat`
Did you mean `echo`?
```
## Solution
This branch moves the examples into the `tokio` directory, so cargo is
now once again aware of them:
```
tokio on eliza/fix-examples [$] via 🦀v1.33.0 at ☸️ aks-eliza-dev
➜ cargo run --example chat
Compiling tokio-executor v0.1.7 (/Users/eliza/Code/tokio/tokio-executor)
Compiling tokio-reactor v0.1.9
Compiling tokio-threadpool v0.1.13
Compiling tokio-current-thread v0.1.6
Compiling tokio-timer v0.2.10
Compiling tokio-uds v0.2.5
Compiling tokio-udp v0.1.3
Compiling tokio-tcp v0.1.3
Compiling tokio-fs v0.1.6
Compiling tokio v0.1.18 (/Users/eliza/Code/tokio/tokio)
Finished dev [unoptimized + debuginfo] target(s) in 7.04s
Running `target/debug/examples/chat`
server running on localhost:6142
```
Signed-off-by: Eliza Weisman <[email protected]>
Signed-off-by: Eliza Weisman <[email protected]>
#993 introduces changes in a sub crate that other Tokio crates depend
on. To make CI pass, a `[patch]` statement and `path` dependencies are
used.
When releasing, these must be removed. However, the commit that
removes them and prepares the crates for release will not be able to
pass CI.
This commit adds a conditional on a special `[ci-release]` snippet in
the commit message. If this exists, CI is only run with the full "patched"
dependencies.
Adds a `TypedExecutor` trait that describes how to spawn futures of a specific
type. This is useful for implementing functions that are generic over an executor
and wish to support both `Send` and `!Send` cases.
This branch makes the following changes to `tokio-trace`'s `Span` type:
* **Remove manual close API from spans**
In practice, there wasn't really a use-case for this, and it
complicates the implementation a bit. We can always add it back later.
* **Remove generic lifetime from `Span`**
Again, there wasn't actually a use-case for spans with metadata that
doesn't live for the static lifetime, and it made using `Span`s in
other types somewhat inconvenient. It's also possible to implement an
alternative API for non-static spans on top of the `tokio-trace-core`
primitives.
Signed-off-by: Eliza Weisman <[email protected]>
PR #973 changed the `tokio_trace_core::span::Id::from_u64` function to
require that the provided `u64` be greater than zero. However, I had
forgotten that the implementation of `Subscriber` for the `NoSubscriber`
type (which is used when no default subscriber is set) always returned
`span::Id::from_u64(0)` from its `new_span` method. In combination with
the assert added in #973, this means that every time a span is hit when
no subscriber is set, `tokio-trace-core` will panic.
This branch fixes the panics by having `NoSubscriber` construct span IDs
using a different (arbitrarily chosen) non-zero constant.
Signed-off-by: Eliza Weisman <[email protected]>
This branch changes `tokio_trace_core::span::Id::from_u64` to assert
that the integer from which the span ID is constructed is greater than
zero. This is to enable future use of non-zero optimization.
Unfortunately, we can't actually use a `NonZeroU64` _now_, as that type
was only stabilized in Rust 1.28.0, and `tokio`'s current minimum
supported Rust version is 1.26.0.
Adding and documenting the assertion now allows us to change the
internal representation to `NonZeroU64` later (when 1.28.0 is the
minimum supported Rust version), without causing a breaking change.
Signed-off-by: Eliza Weisman <[email protected]>
* trace-core: Pass dispatcher by ref to `dispatcher::with_default`
As requested by @carllerche in https://github.com/tokio-rs/tokio/pull/966#discussion_r264380005, this branch changes the
`dispatcher::with_default` function in `tokio-trace-core` to take the
dispatcher by ref and perform the clone internally. This makes this
function more consistant with other `with_default` functions in other
crates.
Signed-off-by: Eliza Weisman <[email protected]>
* trace: Don't set the default dispatcher on entering a span
Setting the default dispatcher on span entry is a relic of when spans
tracked their parent's ID. At that time, it was necessary to ensure that
any spans created inside a span were observed by the same subscriber
that originally provided the entered span with an ID, as otherwise, new
spans would be created with parent IDs that did not originate from that
subscriber.
Now that spans don't track their parent ID, this is no longer necessary.
However, removing this behavior does mean that if a span is entered
outside of the subscriber context it was created in, any subsequent
spans will be observed by the current default subscriber and thus will
not be part of the original span's trace tree. Since subscribers are not
expected to change frequently, and spans are not expected to move
between them, this is likely acceptable.
I've removed the tests for the old behavior.
Note that this change improves the performance of span entry/exit fairly
significantly. Here are the results of running a benchmark that enters
a span, does nothing, and immediately exits it, before this change:
```
test enter_span ... bench: 93 ns/iter (+/- 14)
```
...and after:
```
test enter_span ... bench: 51 ns/iter (+/- 9)
```
Signed-off-by: Eliza Weisman <[email protected]>
## Motivation
Currently, it isn't possible to import individual macros from
`tokio-trace` using the macros 1.2 syntax:
```rust
use tokio_trace::{debug, info, span};
```
This is because these macros require that `callsite` and `enabled` are
imported as well.
## Solution
This branch resolves the problem by adding the [`local_inner_macros`]
attribute to the instrumentation API's macros. This allows other macros
from within the crate to be used without requiring them to be explicitly
imported.
However, this also requires duplicating any macros from other sources
(such as std and `tokio-trace-core`) with wrappers due to the behaviour
of `local_inner_macros`. I've added these wrapper macros as well.
Since the macros got even longer as a result of this, I've moved them
to a separate file to make `lib.rs` easier to read. I've also wrapped
some very long lines in the macros, and removed the explicit drop of
the result of evaluating some event macros (it's no longer necessary
as all event macros now evaluate to `()`).
[`local_inner_macros`]: https://doc.rust-lang.org/nightly/edition-guide/rust-2018/macros/macro-changes.html#local-helper-macrosFixes#968
Signed-off-by: Eliza Weisman <[email protected]>
This branch makes a handful of `tokio-trace-core` API improvements, mostly
around naming. In particular:
* Rename `dispatcher::with` to `dispatcher::get_default`
* Rename `Event::observe` to `Event::dispatch`
* Make `field::ValidLen` trait private
Closes#948Closes#960
Signed-off-by: Eliza Weisman <[email protected]>
This branch changes the `Subscriber::record` method to take a new
arguments struct, `span::Record`. The `field::Record` trait was renamed
to `field::Visit` to prevent name conflicts.
In addition, the `ValueSet::is_empty`, `ValueSet::contains`, and
`ValueSet::record` methods were made crate-private, as they are exposed
on the `Attributes` and `Record` types.
Signed-off-by: Eliza Weisman <[email protected]>
This branch allows users of `tokio-trace` to explicitly set a span's
parent, or indicate that a span should be a new root of its own trace
tree. A `parent: ` key has been added to the `span!` macros. When a span
is provided, that span will be set as the parent, while `parent: None`
will result in a new root span. No `parent:` key results in the current
behaviour.
A new type, `span::Attributes`, was added to `tokio-trace-core` to act
as an arguments struct for the `Subscriber::new_span` method. This will
allow future fields to be added without causing breaking API changes.
The `Attributes` struct currently contains the new span's metadata,
`ValueSet`, and parent.
Finally, the `span::Span` type in `-core` was renamed to `span::Id`, for
consistency with `tokio-trace` and to differentiate it from
`span::Attributes`. This name was chosen primarily due to precedent in
other tracing systems.
Closes#920
Signed-off-by: Eliza Weisman <[email protected]>
A single-producer, multi-consumer channel that only retains the _last_ sent
value. Values are broadcasted out.
This channel is useful for watching for changes to a value from multiple
points in the code base (for example, changes to a configuration value).
- Rewrite noop_waker with items from the new API and replaces
LocalWaker with Waker.
- Bump the minimum required version for `tokio-async-await` to
1.34.0-nightly.
- `Unpin` was added to std prelude.
- Add `cargo check` to .travis.yml
Fixes: #908
This branch adds links to the master RustDoc published by CI to the
`tokio-trace` and `tokio-trace-core` README. In addition, it fixes a
broken links in the RustDoc for `tokio-trace` and updates the
`tokio-trace-core` RustDoc to match the README.
Signed-off-by: Eliza Weisman <[email protected]>
## Motivation
As described in #905, subscribers have no way to get the numeric value of a span ID back _out_ of a `Span`.
## Solution
Add a `Span::into_u64` method that returns the inner `u64` span ID
Closes#905
Signed-off-by: kleimkuhler <[email protected]>
In order to support nesting a tokio reactor within another event
system exposing the file descriptor for the underlying reactor
is useful and is already implemented for mio::Poll.
Signed-off-by: Paul Osborne <[email protected]>
This patch fixes Semaphore by adding a missing code path to the release
routine that handles the case where the waiter's node is queued in the
sempahore but has not yet been assigned the permit.
This fix is used by mpsc to handle the case when the Sender has called
`poll_ready` and is dropped before the permit is acquired.
Fixes#900
This change adds an extension trait to `BufStream` and puts the core
trait behind a feature flag for optional use.
This mainly adds the additional functions in an extension trait to
allow the user to select if they want just the core trait or the fully
featured version. Now the user can add the core feature to _not_
include the extension trait. By deafult, this feature is disabled.
<!-- Thank you for your Pull Request. Please provide a description above
and review the requirements below.
Bug fixes and new features should include tests.
Contributors guide:
https://github.com/tokio-rs/tokio/blob/master/CONTRIBUTING.md -->
## Motivation
In asynchronous systems like Tokio, interpreting traditional log
messages can often be quite challenging. Since individual tasks are
multiplexed on the same thread, associated events and log lines are
intermixed making it difficult to trace the logic flow. Currently, none
of the available logging frameworks or libraries in Rust offer the
ability to trace logical paths through a futures-based program.
There also are complementary goals that can be accomplished with such a
system. For example, metrics / instrumentation can be tracked by
observing emitted events, or trace data can be exported to a distributed
tracing or event processing system.
In addition, it can often be useful to generate this diagnostic data in
a structured manner that can be consumed programmatically. While prior
art for structured logging in Rust exists, it is not currently
standardized, and is not "Tokio-friendly".
## Solution
This branch adds a new library to the tokio project, `tokio-trace`.
`tokio-trace` expands upon logging-style diagnostics by allowing
libraries and applications to record structured events with additional
information about *temporality* and *causality* --- unlike a log
message, a span in `tokio-trace` has a beginning and end time, may be
entered and exited by the flow of execution, and may exist within a
nested tree of similar spans. In addition, `tokio-trace` spans are
*structured*, with the ability to record typed data as well as textual
messages.
The `tokio-trace-core` crate contains the core primitives for this
system, which are expected to remain stable, while `tokio-trace` crate
provides a more "batteries-included" API. In particular, it provides
macros which are a superset of the `log` crate's `error!`, `warn!`,
`info!`, `debug!`, and `trace!` macros, allowing users to begin the
process of adopting `tokio-trace` by performing a drop-in replacement.
## Notes
Work on this project had previously been carried out in the
[tokio-trace-prototype] repository. In addition to the `tokio-trace` and
`tokio-trace-core` crates, the `tokio-trace-prototype` repo also
contains prototypes or sketches of adapter, compatibility, and utility
crates which provide useful functionality for `tokio-trace`, but these
crates are not yet ready for a release. When this branch is merged, that
repository will be archived, and the remaining unstable crates will be
moved to a new `tokio-trace-nursery` repository. Remaining issues on the
`tokio-trace-prototype` repo will be moved to the appropriate new repo.
The crates added in this branch are not _identical_ to the current head
of the `tokio-trace-prototype` repo, as I did some final clean-up and docs
polish in this branch prior to merging this PR.
[tokio-trace-prototype]: https://github.com/hawkw/tokio-trace-prototypeCloses: #561
Signed-off-by: Eliza Weisman <[email protected]>
ATOMIC_BOOL_INIT is deprecated since 1.34 because the const fn
AtomicUsize::new is now preferred. As deny(warnings) is set,
tokio fails to build on latest nightly. This will fix it.
Signed-off-by: Yilin Chen <[email protected]>
Following from https://github.com/tokio-rs/tokio/pull/865, this PR
removes `#[derive(Debug)]` on `mpsc` sender and receiver types in favor
of explicit `impl fmt::Debug` blocks that don't have a `T: fmt::Debug`
bound.
`#[derive(Clone)]` on a type `struct Foo<T>` adds an impl that requires that
`T: Clone`:
```rust
impl<T: Clone> Clone for Foo<T>
```
which is unfortunate in the case of senders, because we don't want to require
that the items being sent are `Clone` for the channel sender to be `Clone`.
This PR adds an explicit `impl Clone` for the bounded and unbounded sender
types which does not have the `T: Clone` bound.
Note that this is _also_ an issue with `#[derive(Debug)]`, but that one is
harder to work around as `chan::Tx` _also_ has `#[derive(Debug)]`, as does
`chan::Chan`, so we'd have to add explicit impls for all of them to make
progress.
Introduce a tokio-sync crate containing useful synchronization primitives for programs
written using Tokio.
The initial release contains:
* An mpsc channel
* A oneshot channel
* A semaphore implementation
* An `AtomicTask` primitive.
The `oneshot` and `mpsc` channels are new implementations providing improved
performance characteristics. In some benchmarks, the new mpsc channel shows
up to 7x improvement over the version provided by the `futures` crate. Unfortunately,
the `oneshot` implementation only provides a slight performance improvement as it
is mostly limited by the `futures` 0.1 task system. Once updated to the `std` version
of `Future` (currently nightly only), much greater performance improvements should
be achievable by `oneshot`.
Additionally, he implementations provided here are checked using
[Loom](http://github.com/carllerche/loom/), which provides greater confidence of
correctness.
## Motivation
When the thread pool shuts down, futures that have been polled at least once but not completed yet are simply leaked. We should drop them instead.
## Solution
Multiple changes are introduced:
* Tasks are assigned a home worker the first time they are polled.
* Each worker contains a set of tasks (`Arc<Task>`) it is home to. When a task is assigned a home worker, it is registered in that worker's set of tasks. When the task is completed, it is unregistered from the set.
* When the thread pool shuts down and after all worker threads stop, the remaining tasks in workers' sets are aborted, i.e. they are switched to the `Aborted` state and their `Future`s are dropped.
* The thread pool shutdown process is refactored to make it more robust. We don't track the number of active threads manually anymore. Instead, there's `Arc<ShutdownTrigger>` that aborts remaining tasks and completes the `Shutdown` future once it gets destroyed (when all `Worker`s and `ThreadPool` get dropped because they're the only ones to contain strong references to the `ShutdownTrigger`).
Closes#424Closes#428
* docs: fixed links to tokio_timer::clock::Now in tokio-timer/src/timer/mod.rs
* docs: fixed links to std::time::Instant in tokio-timer/src/timer/mod.rs
This reverts commit 7a49ebb65e.
The commit conflicted with another change that was merged, causing CI to fail. The public API
also requires a bit more refinement (#833) and Tokio crates need to be released.
Disabling all features means the only dependency is `futures`.
Relevant pieces of the API can then be enabled with the following features:
- `codec`
- `fs`
- `io`
- `reactor`
- `tcp`
- `timer`
- `udp`
- `uds`
This also introduces the beginnings of enabling only certain pieces of the `Runtime`. As a start, the entire default runtime API is enabled via the `rt-full` feature.
Error::cause is deprecated in Rust 1.33, but this allows Error::cause
until the minimum supported version of tokio is Rust 1.30.
When the minimum support version of tokio reaches Rust 1.30,
replace Error::cause with Error::source.
Fixes: #817
`inner` is a fitting name for variables of type named `Inner`, but in other cases I find them confusing - sometimes `inner` refers to a `Pool`, sometimes to a `Sender`. I renamed a bunch of variables named `inner` to be more descriptive.
This PR is the first step in an effort of splitting https://github.com/tokio-rs/tokio/pull/722#issuecomment-439552671 into multiple PRs.
When spawning using `Handle` while on the executor, tasks were being
double counted. This prevented the number of active tasks to reach zero,
thus preventing the executor from shutting down.
This changes `spawn` to check if being called from the executor
**before** incrementing the number of active tasks.
Fixes#760
Should hopefully fix the underlying bug that was causing tokio-tls tests to occasionally fail on Windows.
Signed-off-by: Toby Lawrence <[email protected]>
* async-await: fix README example dependencies
As per commit "async-await: track nightly changes (#661)" ( commit
2f690d30bc)
> The `tokio-async-await` crate is no longer a facade. Instead, the
> `tokio` crate provides a feature flag to enable async/await support.
Ensure the example in the async-await README file also works by
correctly declaring this updated dependency
* async-await: remove unnecessary 'edition' declaration from README
As the "edition" feature was stabilized in rust v1.30 and async-await
specifies that the nightly toolchain must be used, remove the use of the
"edition" feature gate since it is enabled by default.
* Minimize allocation needed for channels
* Use a newtype for signal ids
* We can just cast the raw pointer to a `usize` and still perform a
simple identity check, without incurring any implications of storing a
raw pointer (e.g. previously Signal was !Sync and had an unsafe impl of
Send, and now it is naturally Sync+Send)
* Broadcast with `try_send` instead of `start_send`
The `Stream::start_send` method uses backpressure and schedules the
current task to be notified whenever the channel has additional room,
which means we'll generate a lot of unnecessary wakeups whenever a
channel gets full
By changing to `try_send` and handling any errors, we ensure the
Driver's task won't get woken up when a Signal finally consumes its
notification, since we're coalescing things anyway
* udp: add `into_parts` to `RecvDgram`
If `RecvDgram` can not be driven to completion it may become necessary to get back the `UdpSocket` it contains which is currently not possible.
This adds`into_parts` to get the socket as well as the buffer back. Both methods consume `RecvDgram`.
Note that after the future has completed, `into_parts` must not be used, or else a panic will happen.
## Motivation
tokio depends on an out of date version of crossbeam-utils, which results in multiple versions of that package being linked in binaries which use other popular libraries.
## Solution
Bump the version; there's no API changes and tests still pass.
* rt: fix `Runtime::reactor()` as used by tokio-core
Up until Tokio v0.1.11, the handle returned by `Runtime::reactor()`
pointed to a reactor instance running in a background thread. The thread
was eagerly spawned.
As of v0.1.12, a reactor instance is created per runtime worker thread.
`Runtime::reactor()` was deprecated and updated to point to the reactor
for one of the worker threads.
A problem occurs when attempting to use the reactor before spawning a
task. Worker threads are spawned lazily, which means that the reactor
referenced by `Runtime::reactor()` is not yet running.
This patch changes `Runtime::reactor` back to a dedicated reactor
running on a background thread. However, the background thread is now
spawned lazily when the deprecated function is first called.
Fixes#720
* Fix comment
Co-Authored-By: carllerche <[email protected]>
- `tokio::run` checks Enter before creating a new threadpool and
spawning the main future.
- `Runtime::block_on` now checks Enter
- `Runtime::block_on_all` now checks Enter
<!--
Thank you for your Pull Request. Please provide a description above and review
the requirements below.
Bug fixes and new features should include tests.
Contributors guide: https://github.com/tokio-rs/tokio/blob/master/CONTRIBUTING.md
-->
## Motivation
Now that each worker thread drives its own reactor, reactors have to be driven until the threadpool shuts down. We mustn't use the `keep_alive` setting to shut down a worker thread if it doesn't receive an event from the reactor for a certain duration of time.
<!--
Explain the context and why you're making that change. What is the problem
you're trying to solve? In some cases there is not a problem and this can be
thought of as being the motivation for your change.
-->
## Solution
Just ignore the `keep_alive` setting when parking in `Worker::sleep`.
<!--
Summarize the solution and provide any necessary context needed to understand
the code change.
-->
Fixes: #681
## Motivation
Currently, a potential panic exists in `LengthDelimitedCodec::encode`.
Writing the length field to the `dst` buffer can exceed the buffer
capacity, as `BufMut::put_uint_{le,be}` doesn't reserve more capacity.
## Solution
This branch adds a call to `dst.reserve` to ensure that there's
sufficient remaining buffer capacity to hold the length field and
the frame, prior to writing the length field. Previously, capacity
was only reserved later in the function, when writing the frame
to the buffer, and we never reserved capacity for the length field.
I've also added a test that reproduces the issue. The test panics on
master, but passes after making this change.
Signed-off-by: Eliza Weisman <[email protected]>
* io: ensure ReadHalf/WriteHalf do not return WouldBlock directly
These facades were passing back WouldBlock when the internal BiLock
couldn't be acquired, which does not fit the intended behavior.
Signed-off-by: Toby Lawrence <[email protected]>
* io: pull from the local crate, not crates.io
## Motivation
Currently, there is a potential denial of service vulnerability in the
`lines` codec. Since there is no bound on the buffer that holds data
before it is split into a new line, an attacker could send an unbounded
amount of data without sending a `\n` character.
## Solution
This branch adds a `new_with_max_length` constructor for `LinesCodec`
that configures a limit on the maximum number of bytes per line. When
the limit is reached, the the overly long line will be discarded (in
`max_length`-sized increments until a newline character or the end of the
buffer is reached. It was also necessary to add some special-case logic
to avoid creating an empty line when the length limit is reached at the
character immediately _before_ a `\n` character.
Additionally, this branch adds new tests for this function, including a
test for changing the line limit in-flight.
## Notes
This branch makes the following changes from my original PR with
this change (#590):
- The whole too-long line is discarded at once in the first call to `decode`
that encounters it.
- Only one error is emitted per too-long line.
- Made all the changes requested by @carllerche in
https://github.com/tokio-rs/tokio/pull/590#issuecomment-420735023Fixes: #186
Signed-off-by: Eliza Weisman <[email protected]>
## Motivation
Currently, the `RUST_BACKTRACE` environment variable is set to `1` on
Travis CI builds:
https://github.com/tokio-rs/tokio/blob/0ca973a7ebc5b8a29beac1ccb6c73ef26ddcbf22/.travis.yml#L49
However, it's not set on AppVeyor. This can make debugging
Windows-specific CI failures challenging for developers on other
operating systems.
## Solution
This branch sets `RUST_BACKTRACE=1` on AppVeyor.
Signed-off-by: Eliza Weisman <[email protected]>
Since the CI runs all tests for all tokio crates, it is possible that a
sporadic failure in one crate can mask failures/successes of other
crates' tests.
Using the `--no-fail-fast` flag instructs cargo to run *all* tests
before failing the build. This will allow checking to see if any
relevant test cases still pass even if an unrelated test has failed.
* Don't use tokio-core any more for tests. That one brings tokio from
crates.io instead of the current workspace and two versions of that
don't want to cooperate.
* Guard unix-specific examples on windows.
* Leave CI setup to top-level directory.
* Originally reported in alexcrichton/tokio-process#42
* The root cause appears to be due to two different PollEvented
instances trying to consume readiness events from the same file
descriptor.
* Previously we would simply swallow any `AlreadyExists` errors when
attempting to register the pipe receiver with the event loop. I'm not
sure if this means the PollEvented wrapper wasn't fully registered to
receive events, or maybe there is a potential race condition with how
PollEvented consumes mio readiness events. Using a fresh/duplicate file
descriptor appears to mitigate the issue, however.
* I was also not able to reproduce the issue as an isolated test case so
there is no regression test available within this crate (but we can add
one in tokio-process)
* codec: add new constructor `with_max_length ` to `LinesCodec`
* codec: add security note to docs
Signed-off-by: Eliza Weisman <[email protected]>
* Fix Rust 1.25 compatibility
* codec: Fix incorrect line lengths in tests (and add assertions)
Signed-off-by: Eliza Weisman <[email protected]>
* codec: Fix off-by-one error in lines codec
Signed-off-by: Eliza Weisman <[email protected]>
* codec: Fix call to decode rather than decode_eof in test
Signed-off-by: Eliza Weisman <[email protected]>
* codec: Fix incorrect LinesCodec::decode_max_line_length
This bug was introduced after the fix for the off-by-one error.
Fortunately, the doctests caught it.
Signed-off-by: Eliza Weisman <[email protected]>
* codec: Minor style improvements
Signed-off-by: Eliza Weisman <[email protected]>
* codec: Don't allow LinesCodec length limit to be set after construction
Signed-off-by: Eliza Weisman <[email protected]>
* codec: change LinesCodec to error and discard line when at max length
* codec: Fix build on Rust 1.25
The slice patterns syntax wasn't supported yet in that release.
Signed-off-by: Eliza Weisman <[email protected]>
* codec: Add test for out-of-bounds index when peeking
Signed-off-by: Eliza Weisman <[email protected]>
* codec: Fix out of bounds index
* codec: Fix incomplete comment
Signed-off-by: Eliza Weisman <[email protected]>
* codec: Add test for line decoder buffer underrun
@jonhoo reported a panic in the call to `LocalKey::with`, which occurs
when the reactor is dropped in the middle of TLS teardown. This PR
changes the call to `LocalKey::try_with` and handles the case when the
thread-local value has already been destroyed.
* Added timeouts to all tests that were missing them
- any issue we have will likely result in deadlocks/starvation so its
best if all tests quickly timeout rather than require getting killed
or have the CI timeout itself
* Added a `support` module and put a bunch of helpers there to DRY the
tests
* As observed in alexcrichton/tokio-signal#38, Signal instances can starve based on the order
they are created in, and this ordering appears to be platform/OS
specific
* The crux of the issue is that we woud only *attempt* to broadcast any
pending signals if we successfully read out at least one byte from the
global pipe.
* For reasons unclear to me, the affected Signal instance would get
woken up after the signal handler writes to the global pipe, but it
would immediately hit a WouldBlock error and give up, bypassing the
broadcast attempt (even though the pending flag was correctly set).
- Maybe this has to do with OS specifics with how the bytes are
delivered (or not), or with some complex interaction with tokio and
the pipe registration. It seems fishy since strace logs didn't show
the signal handler pipe write fail either, but I'm all out of ideas
* The fix appears simple: unconditionally attempt to broadcast any
pending signals *any* time a Driver instance is woken up.
* Since we perform an atomic check for each pending signal, we know that
each (coalesced) signal broadcast will happen at most once. If we were
supuriously woken up and no signals were pending, then nothing will be
yielded to any pollers of Signal
* The down side is that since each Signal instance polls a Driver
instance, each poll to Signal will essentially perform N atomic
operations (N = number of signals we support) in an attempt to broadcast
any pending signals.
- However, we can revisit optimizing this better in the future
Fixesalexcrichton/tokio-signal#38
* We introduce a new global structure which keeps track of how many
signal streams have been registered with a given event loop (the event
loop is identified by its OS file descriptor)
* We only attempt to deregister our global evented pipe from any event
loop if and only if we are the last signal that was registered with it
* Currently, whenever a new signal stream is created we attempt to
register a global pipe with the event loop to drive events.
* We also (correctly) swallow any descriptor-already-registered errors
since the same pipe is always used
* However, we currently *deregister* the same global pipe *any time* a
Signal stream is dropped.
* This means that if 2 or more of Signal instances exist simultaneously
(even if listening for different signals) and one of them is dropped,
the remainder will starve (until any new signal is created again).
* Cargo runs each integration-style-test in its own process. Since the
tests use global data structures specific to the process, we should run
them in an isolated manner to avoid having cross-test interactions
* Fixesalexcrichton/tokio-signal#39
This allows tokio-signal to build with `-Z minimal-versions` - see
https://github.com/rust-lang/cargo/issues/5657#issuecomment-401110172
for more details.
Earlier versions depend on log 0.3.1, which itself depends on libc
0.1, which doesn't build on any post-1.0 version of rust.
This text historically was copied verbatim from rust-lang/rust's own README [1]
with the intention of licensing projects the same as rustc's own license, namely
a dual MIT/Apache-2.0 license. The clause about "various BSD-like licenses"
isn't actually correct for almost all projects other than rust-lang/rust and
the wording around "both" was slightly ambiguous.
This commit updates the wording to match more precisely what's in the
standard library [2], namely clarifying that there aren't any BSD-like licenses
in this repository and that the source is licensable under either license, at
your own discretion.
[1]: https://github.com/rust-lang/rust/tree/f0fe716dbcbf2363ab8f929325d32a17e51039d0#license
[2]: https://github.com/rust-lang/rust/blob/f0fe716dbcbf2363ab8f929325d32a17e51039d0/src/libstd/lib.rs#L5-L9
Replace the sequential counting (which might be exhausted) by an address
of an object (in a box, so it doesn't change). This is also a unique, so
it is acceptable ID.
Run multiple loops (both in parallel and sequentially) to make sure
broadcasting to multiple of them works and we work even after the
initial loop has gone away.
Add the driver task, connecting the signal handler wakeups to the
wakeups of of the streams.
It is a prototype-quality code, a lot of cleanups and similar is needed.
Which is just a wrapper around the futures::sync::mpsc. The sender is in
a global registry.
The part that connects the wakeups to the senders in the registry
doesn't yet exist.
Register the signal handler that wakes up someone through a self-pipe.
That someone doesn't yet exist, though.
Some dependencies (nix, lazy_static) added to speed up the prototyping
process. They are likely to be dropped in some future commits.
Some features (eg. preserving the previous signal handlers) are still
missing.
## Motivation
`tokio_threadpool::ThreadPool::spawn` has no return value.
## Solution
Add `ThreadPool::spawn_handle` which calls
`futures::sync::oneshot::spawn` to return a future represents the return
value.
This patch refactors `length_delimited` to be implemented as a `Codec` and
use the default `Framed` wrapper types.
The original implementation did not do this in order to support vectored writes in the
write half. However, this implementation would be more efficient with small frames anyway.
If vectored writes are to be explored in the future, then it should be done holistically.
Signed-off-by: Eliza Weisman <[email protected]>
Previously, every call to `current_thread::Handle::spawn` would go
through a `mpsc` channel. This is unnecessary when the `Handle` is still
on the same thread as the current thread executor. This patch fixes that
by storing the `ThreadId` of the executor when it is created, and then
comparing against that when `Handle::spawn` is called. If the call is
made from the same thread, `spawn_local` is used directly.
Fixes#562.
This patch keeps the primary net types in `tokio::net` and moves
secondary types to a protocol specific submodules.
Primary types are the ones that users are most likely to name (`TcpStream`,
`TcpListener`, `UdpSocket`, ...)
Secondary types are the operation futures.
This patch adds experimental async/await support to Tokio. It does this
by adding feature flags to existing libs only where necessary in order
to add nightly specific code (mostly `Unpin` implementations). It then
provides a new crate: `tokio-async-await` which is a shim layer on top
of `tokio`.
The `tokio-async-await` crate is expected to look exactly like `tokio`
does, but with async / await support. This strategy reduces the amount
of cfg guarding in the main libraries.
This patch also adds `tokio-channel`, which is copied from futures-rs
0.1 and adds the necessary `Unpin` implementations. In general, futures
0.1 is mostly unmaintained, so it will make sense for Tokio to take over
maintainership of key components regardless of async / await support.
Re-export it inside the tokio::runtime::current_thread, as the original
place (tokio::executor::current_thread) is hidden from documentation and
users need some way to spawn non-Send futures.
This also bumps a number of sub crates:
* tokio-executor (0.1.3)
* tokio-io (0.1.8)
* tokio-reactor (0.1.4)
* tokio-threadpool (0.1.6)
* tokio-timer (0.2.6)
* tokio-udp (0.1.2)
This patch introduces `Timeout`. This new type allows setting a timeout
both using a duration and an instant. Given this overlap with
`Deadline`, `Deadline` is deprecated.
In addition to supporting future timeouts, the `Timeout` combinator is
able to provide timeout functionality to streams. It does this by
applying a duration based timeout to each item being yielded.
The main reason for introducing `Timeout` is that a deadline approach
does not work with streams. Since `Timeout` needed to be introduced
anyway, keeping `Deadline` around does not make sense.
* Remove `counted` field on `timer::Entry`.
It turns out that a better indicator of whether or not the number of
active timeouts should be decremented is if the `Entry` has been
associated with a timer. In other words, if `Entry::inner` can be
upgraded, then the count should be decremented on drop.
* timer: Tweak link between `Delay` and the driver
This tweaks the struct layout / details regarding how a `Delay` instance
is linked to a driver (timer instance). Instead of lazily allocating the
`Entry` (node shared between `Delay` and the timer), `Entry` is
allocated immediately when `Delay` is created. This allows using the
entry store data used by `Delay`.
This is in anticipation of further timer improvements that would
otherwise require the size of `Delay` to grow further. Since an
allocation is already made, the idea is to shrink the size of the
`Delay` struct.
This patch adds a `DelayQueue` to tokio_timer. The `DelayQueue` allows
inserting elements as well as specifying a time at which the element
should be returned to the user. This allows handling more complex
timeout situations.
This patch implements `Default` for `tokio_timer::Handle`. It returns a
`Handle` instance that is not bound to a specific timer. Instead, it
will use the timer for the current execution context. This is the same
strategy used by `tokio_reactor::Handle`.
Fixes#547
The `lines_encoder` test is a copy/paste of `bytes_encoder` and not
testing the LinesCodec encoding at all. This updates the test to do
simple validations of the LinesCodec encoding.
The futures 0.2 crate is not intended for widespread usage. Also, the
futures team is exploring the compat shim route.
If futures 0.3 support is added to Tokio 0.1, then a different
integration route will be explored, making the current code unhelpful.
* create_dir
* create_dir_all
* hard_link
* read_dir
* read_link
* remove_dir
* remove_file
* rename
* set_permissions that works with path
* symlink_metadata
* symlink on unix
* symlink_dir on windows
* symlink_file on windows
* Normalize links to docs.rs/CRATE/M.N/...
docs.rs is smart enough to show docs for the latest M.N.P release when
M.N is used in the link. For example:
https://docs.rs/mio/0.6/mio/struct.Poll.html
..will show mio 0.6.14 and later docs. While using the `M.N.*`
(ASTERISK) syntax also works, `M.N` is the more common usage, so
standarize a few existing links to that format.
* Fix missing or malformed rustdoc links
* executor lib rustdoc minor format change
* Promote tokio-threadpool crate level comments to rustdoc
* Replace hidden tokio::executor::thread_pool docs with deprecation note
* Fix typo/simplify util module rustdoc
* Reuse some tokio::executor::thread_pool rustdoc for the crate
Relates to #421
This PR enables backtraces when running tests and disables tsan for the thread pool.
The thread sanitizer was generating too many false positives. Once #329 lands, then it can
be re-enabled.
Extract `tokio::executor::current_thread` to a tokio-current-thread
crate. Deprecated fns stay in the old location. The new crate only
contains thee most recent API.
Currently, the timer uses a `Now` trait to abstract the source of time.
This allows time to be mocked out. However, the current implementation
has a number of limitations as represented by #288 and #296.
The main issues are that `Now` requires `&mut self` which prevents a
value from being easily used in a concurrent environment. Also, when
wanting to write code that is abstract over the source of time, generics
get out of hand.
This patch provides an alternate solution. A new type, `Clock` is
provided which defaults to `Instant::now` as the source of time, but
allows configuring the actual source using a new iteration of the `Now`
trait. This time, `Now` is `Send + Sync + 'static`. Internally, `Clock`
stores the now value in an `Arc<Now>` value, which introduces dynamism
and allows `Clock` values to be cloned and be `Sync`.
Also, the current clock can be set for the current execution context
using the `with_default` pattern.
Because using the `Instant::now` will be the most common case by far, it
is special cased in order to avoid the need to allocate an `Arc` and use
dynamic dispatch.
The thread pool optimizes cases where a task currently running on the
pool spawns a new future. However, the optimization did not factor in
cases where two thread pools interacted.
This patch fixes the optimization and includes a test.
Fixes#342
Currently, not specifying a `Handle` is different than using
`Handle::default()`. This is because `Handle::default()` will
immediately bind to the reactor for the current context vs. not
specifying a `Handle`, which binds to a reactor when it is polled.
This patch changes the `Handle::default()` behavior, bringing it inline
with actual defaults.
`Handle::current()` still immediately binds to the current reactor.
Fixes#307
This patch adds a new crate: tokio-fs. This crate provides a wrapper
around `std` functionality that can only be performed using blocking
operations. This primarily includes filesystem operations, but it also
includes standard input, output, and error access as these streams
cannot be safely switched to non-blocking mode in a portable way.
These wrappers call the `std` functions from within a `blocking`
annotation which allows the runtime to compensate for the fact that the
thread will potentially remain blocked in a system call.
Currently, the state of the futures2 integration is pretty broken. This
patch removes the feature flag, preventing users from trying to use it.
In the future, it can be brought back when the implementation is fixed.
This patch introduces a version of `Runtime` that runs all components on
the current thread. This allows users to spawn futures that do not implement
`Send`.
This ensures that all fd-based futures are put into the queue for the
current tick, if the CurrentThread is parking via the Reactor.
Otherwise, if there are queued up futures already, only those would be
polled in the turn. These futures could then notify others/themselves to
have the queue still non-empty on the next turn. Which then potentially
allows the reactor to never be polled, and thus fd-based futures are
never queued up and polled.
Also return in the Turn return value whether any futures were polled at
all, which allows the caller to know if any work was done at all in this
turn and based on that adjust behavior.
This patch adds a `blocking` to `tokio-threadpool`. This function serves
as a way to annotate sections of code that will perform blocking
operations. This informs the thread pool that an additional thread needs
to be spawned to replace the current thread, which will no longer be
able to process the work queue.
Replaces homegrown Arc with std Arc
Is this safer? Unknown. At least we don't have to maintain an arc
implementation anymore. This will also make it easier to filter out tsan
false positives.
Also split task/mod.rs into multiple files.
This patch renames `Sleep` from tokio-timer and the tokio facade to
`Delay`. Given that the future does not actually put anything to sleep,
the `Delay` name feels more appropriate.
Fixes#263
This patch integrate the new timer implementation with the runtime by
initializing a timer per worker thread. This allows minimizing the
amount of synchronization needed for using timers.
* Allow customizing the threadpool's parker
This patch allows the user of threadpool to customize how the worker
threads park themselves. This allows custom parking logic to be
injected. For example, this allows embedding a timer on each worker
thread.
* Call `park` instance every so often.
Since the `park` is now customizable, it might have logic that must be
called every so often. For example, a timer might have timeouts that it
must expire.
Currently, if a worker is very busy, it won't call into the `park`
instance. This patch changes this so that after every 32 task
invocations, `park` is called with a duration of zero.
This patch adds a new crate: tokio-timer. This crate provides an
efficient timer implemeentation designed for use in Tokio based
applications.
The timer users a hierarchical hashed timer wheel algorithm with six
levels, each having 64 slots. This allows the timer to have a resolution
of 1ms while maintaining O(1) complexity for insert, removal, and firing
of timeouts.
There already exists a tokio-timer crate. This is a complete rewrite
which solves the outstanding problems with the existing tokio-timer
library.
Closes#146.
* Fix races.
This mostly pulls in changes from rust-lang-nursery/futures-rs#881, but
also updates Registration to be a bit more obvious as to what is going
on.
* Reduce spurious wakeups caused by Reactor
This patch adds an ABA guard on token values before registering them
with Mio. This allows catching token reuse and avoid the notification.
This is needed for OS X as the notification is used to determine that a
TCP connect has completed. A spurious notification can potentially cause
write failures.
Currently, the runtime does not shutdown if the runtime handle is
dropped. This can happen during a panic or when the value is simply
dropped.
This patch forces the runtime to shutdown if it is not explicitly
shutdown.
Fixes#209
If a future panics from within the context of a thread pool, the pool
should not be impacted. To do this, polling the future is wrapped with a
catch_unwind. Extra care is taken to ensure that `thread::panicking()`
is set from within the future's drop handle.
Fixes#209
This patch relicenses the Tokio project exclusively under the MIT
license. Before this, the project was dual licensed under MIT and Apache
2. As such, switching to only MIT is permitted.
Fixes#202
This patch fixes a bug where `CurrentThread::turn` is expected to block
even if the executor is idle.
The `turn` API is the low level interface for callers to interact with
the `Sleep` instance used by the `CurrentThread` instance. As such, a
call to `turn` is expected to call `sleep` once if the executor did not
perform any work.
This patch updates `poll_read_ready` to take a `mask` argument, enabling
the caller to specify the desired readiness. `need_read` is renamed to
`clear_read_ready` and also takes a mask.
This enables a caller to listen for HUP events without requiring reading
from the I/O resource.
Currently, if a thread pool instance is dropped without being shutdown,
the workers will run indefinitely. This is not ideal as it leaks the
threadpool.
This patch forces the thread pool to shutdown on drop.
Closes#151
Mio will be requiring `deregister` to be called explicitly in order to
guarantee that Poll releases any state associated with the I/O resource.
See carllerche/mio#753.
This patch adds an explicit `deregister` function to `Registration` and
updates `PollEvented` to call this function on drop.
`Registration::deregister` is also called on `PollEvented::into_inner`.
Closes#168
Some of the benchhmarks were broken and/or using deprecated APIs. This
patch updates the benches and requires them all to compile without
warnings in order to pass CI.
The exampes included in the repository have lagged behind the changes
made. Specifically, they do not use the new runtime construct.
This patch updates examples to use the latest features of Tokio.
Tokio is moving away from using `WouldBlock`, instead favoring
`Async::NotReady`.
This patch updates the TCP and UDP types, deprecating any function that
returns `WouldBlock` and adding a poll_ prefixed equivalent.
Currently, `tokio::spawn` matched the `spawn` function from futures 0.2.
However, this adds additional ergonomic overhead and removes the ability
to spawn from a drop fn. See rust-lang-nursery/futures-rs#830.
This patch switches the behavior to access the thread-local variable
referencing the default executor directly in the `spawn` function.
This allows libraries that require access to reactor related types to
depend on this crate without having to depend on the entirety of Tokio.
For example, libraries that implement their custom I/O resource will
need to access `Registration` or `PollEvented`.
This patch updates the documentation for a number of APIs. It also
introduces a prelude module and an io facade module, re-exporting types
from tokio-io.
Sometimes, passing ownership to an executor is necessary. For example,
some libraries require taking ownership of one.
This patch adds a function that returns an executor associated with a
runtime.
This patch makes a significant change to how I/O resources bind to a
reactor. Currently, an I/O resource (TCP, UDP, PollEvented) will bind
itself with a reactor upon creation.
First, some history.
Originally, tokio-core required that I/O resources be explicitly
associated with a reactor upon creation by passing in a `&Handle`. Tokio
reform introduced a default reactor. If I/O resources do not specify a
reactor upon creation, then the default reactor is used.
However, futures tend to favor being lazy. Creating a future should do
no work, instead it is defining a computation to be performed once the
future is executed. Binding an I/O resource with a reactor on creation
goes against this pattern.
This patch fixes this by allowing I/O resources to lazily bind to a
reactor. An explicit `&Handle` can still be used on creation, but if no
reactor is specified, then the default reactor is used. However, this
binding happens during execution time (read / write) and not creation.
Currently, the thread-local tracking the current thread executor is not
set when a task is dropped. This means that one cannot spawn a new
future from within the drop implementation of another future.
This patch adds support for this by setting the thread-local before
releasing a task.
This implementation is a bit messy. It probably could be cleaned up, but
this is being put off in favor of trying a more comprehensive
reorganization once the current thread executor is feature complete.
The logic that enables `CurrentThread::turn` to avoid unbounded
iteration was incorrect. It was possible for unfortunate timing to
result in a dead lock.
This patch provides a fix as well as a test.
FreeBSD uses a separate kqueue filter type for lio_listio. This change
adds support for that filter type. Full functionality will be provided
by the mio-aio and tokio-file crates.
* Add PollEvented::into_inner
Consumes a PollEvented and returns its inner io object. Useful for io
types that have exclusive ownership of a resource.
See also https://github.com/tokio-rs/tokio-core/commit/9400ffb
CurrentThread::turn uses a turn count strategy to allow `turn` to not
run infinitely. Currently, there is a bug where spawned tasks will not
get executed in calls to `turn`.
This patch fixes the bug by correctly setting the turn count for newly
spawned tasks.
This patch is an intial implementation of the Tokio runtime. The Tokio
runtime provides an out of the box configuration for running I/O heavy
asynchronous applications.
As of now, the Tokio runtime is a combination of a work-stealing thread
pool as well as a background reactor to drive I/O resources.
This patch also includes tokio-executor, a hopefully short lived crate
that is based on the futures 0.2 executor RFC.
* Implement `Park` for `Reactor`
This enables the reactor to be used as the thread parker for executors.
This also adds an `Error` component to `Park`. With this change, a
`Reactor` and a `CurrentThread` can be combined to achieve the
capabilities of tokio-core.
* move src/codec.rs -> src/codec/mod.rs
* Move traits Encoder and Decoder from src/framed_{read|write}.rs into src/codec/{encoder,decoder}.rs
* Move LinesCodec and BytesCodec from src/codecs.rs into src/codec/{lines,bytes}_codec.rs
The `Handle` type is intended to be used by end users of Tokio. The
wakeup functionlity is needed by executor implementations. It doesn't
make sense to put executor specific functionality on a type that is
intended for end users.
* Limit the max number of registered resources
This allows some token space to be used internally. Also, Mio 0.7 will
be limiting the token space some as well.
Mio: https://github.com/carllerche/mio/issues/788Closes#42
Generally speaking, it is unsafe to access to perform asynchronous
operations using `&self`. Taking `&self` allows usage from a `Sync`
context, which has unexpected results.
Taking `&mut self` to perform these operations prevents using these
asynchronous values from across tasks (unless they are wrapped in
`RefCell` or `Mutex`.
The caller should be required to have control of a reactor when setting
it as a fallback. A `Handle` is used to pass into libraries and
functions and should not grant capability of using the associated
reactor as a fallback.
Closes#71
std's `Incoming` iterator yields `TcpStream` instances. This patch
updates the `Incoming` future to match this signature.
This changes the yielded value from `(TcpStream, SocketAddr)` ->
`TcpStream`.
This was added oh-so-long ago and nowadays is just used by one consumer,
`PollEvented`. Let's just inline the already small implementation directly into
`PollEvented` which should make it easier to modify in the future as well
* Refactor UDP SendDgram & RecvDgram
Get rid of unnamed structs in the favor of private structs with named fields
* Change the signature of UdpCodec::encode
Now it is:
```
fn encode(&mut self, msg: Self::Out, buf: &mut Vec<u8>) -> Result<SocketAddr, Self::Error>;
```
Closes https://github.com/tokio-rs/tokio/issues/79
* Fix compilation error from `mio` crate
* Change a return value of reactor::poll to io::Result.
* Revert "Change a return value of reactor::poll to io::Result."
This reverts commit 281d8c32d4.
* Return a result from reactor::poll.
* Drop a reactor if any error occurs. Fix warnings in tests.
* Update a documentation for reactor::turn.
* Unwrap the last turn() call in tests.
This commit removes the `Handle` argument from the following constructors
* `TcpListener::bind`
* `TcpStream::connect`
* `UdpSocket::bind`
The `Handle` argument remains on the various `*_std` constructors as they're
more low-level, but this otherwise is intended to set forth a precedent of by
default not taking `Handle` arguments and instead relying on the global
`Handle::default` return value when necesary.
This method is intended to be used to wake up the reactor from a remote thread
if necessary, forcing it to return from a blocked call of `turn` or otherwise
prevent the next call to `turn` to from blocking.
This commit removes the `Reactor::run` method which has previously been used to
execute futures and turn the reactor at the same time. The tests/examples made
heavy usage of this method but they have now all temporarily moved to `wait()`
until the futures dependency is upgraded. In the meantime this'll allow us to
further trim down the `Reactor` APIs to their final state.
This commit starts to add support for a global event loop by adding a
`Handle::default` method and implementing it. Currently the support is quite
rudimentary and doesn't support features such as shutdown, overriding the return
value of `Handle::default`, etc. Those will come as future commits.
This commit changes the `PollEvented::deregister` signature from
fn deregister(self, handle: &Handle) -> io::Result<()>
to
fn deregister(&self) -> io::Result<()>
Now that the handles are `Send` and `Sync` there's no longer any need to pass it
in (it's already stored in the `PollEvented` itself). Additionally this switches
to `&self` instead of `self` to allow reclamation of the internal resources if
necessary.
This should allow configuration over what reactor accepted streams go on to by
giving back a libstd-bound object that can then be used later in conjunction
with `TcpStream::from_std`.
This commit is targeted at solving tokio-rs/tokio-core#12 and incorporates the
solution from tokio-rs/tokio-core#17. Namely the `need_read` and `need_write`
functions on `PollEvented` now return an error when the connected reactor has
gone away and the task cannot be blocked. This will typically naturally
translate to errors being returned by various connected I/O objects and should
help tear down the world in a clean-ish fashion.
This commit renames the various constructors of networking types to have a
`_std` suffix instead of a smorgasboard of other suffixes, canonicalizing on
`_std` as the suffix for constructors which take the libstd corresponding types.
* small doc cleanups in PollEvented
* small doc cleanups in IoToken
* improve crate level documentation
- Add links to the futures, mio and tokio-uds crates.
- Add links to various structs and types mentioned.
- use eprintln for error reporting in the example.
* improvements to the UdpSocket documentation
- Fixed links usage.
- Removed references to a no longer existing `Window` struct.
- Made notes about using functions in context of a future.
* documentation improvements to UdpFramed and UdpCodec
- Since HTTP uses TCP (QUIC aside) using it as an example in an UDP
protocol feels wrong.
- Make the note of tampering with the underlying streams more explicit.
* update reactor module level documentation
Adds an explanation of every public struct.
* expand Handle and Remote documentation
* expand net module documentation
Adds an explanation of every public struct and how they work together.
* update TcpListener documentation
Reorder the various option methods; get first then set.
Note about panicing added to poll_read.
* remove mention of none-existing future R
* improve documentation of TcpStream
* fix UdpSocket doc
This when wrong when merging various commits.
* Make Handle `Send + Sync`.
This is an initial implementation making `Handle: Send + Sync`. It uses
a `RwLock` to coordinate access to the underlying state storage. An
implementation without the lock is left to later.
This pass also leaves a lot of dead code that can be removed in later
commits.
* Remove reactor code related to message passing
The previous commit removed the need for using message passing to
communicate with the reactor. This commit removes all the unnecessary
code.
Mio should be providing the compatibility layer so there is no need to
run CI on OS X as well as Linux.
Benches are currently being skipped as the Tokio reform work progresses.
In accordance with tokio-rs/tokio-rfcs#3, the executor functionality of
Tokio is being removed and will be relocated into futures-rs as a
"current thread" executor.
This PR removes task execution from the code base. As a temporary
mesure, all examples and tests are switched to using CpuPool.
Depends on #19.
In accordance with tokio-rs/tokio-rfcs#3, timers are being extracted
from Tokio and moved to a separate crate (probably futures-timer).
This PR removes timers from the code base.
This text historically was copied verbatim from rust-lang/rust's own README [1]
with the intention of licensing projects the same as rustc's own license, namely
a dual MIT/Apache-2.0 license. The clause about "various BSD-like licenses"
isn't actually correct for almost all projects other than rust-lang/rust and
the wording around "both" was slightly ambiguous.
This commit updates the wording to match more precisely what's in the
standard library [2], namely clarifying that there aren't any BSD-like licenses
in this repository and that the source is licensable under either license, at
your own discretion.
[1]: https://github.com/rust-lang/rust/tree/f0fe716dbcbf2363ab8f929325d32a17e51039d0#license
[2]: https://github.com/rust-lang/rust/blob/f0fe716dbcbf2363ab8f929325d32a17e51039d0/src/libstd/lib.rs#L5-L9
This example is intended to showcase sharing state between all connected clients
on a server, for example a key/value store (in-memory database)
Closes#257
Some contexts, like EC2, have `Instant::now` as a relatively expensive function
to call. To help amortize the cost of this function this commit exposes a new
function on `Interval` and `Timeout` to pass in the assumed current time.
The `consume_queue` function can be relatively slow for an empty queue (the fast
path) so optimize this a bit with a preflight check that should just touch a few
atomics.
My slab bump in #246 added a potential new panic in Core's Inner's
`cancel_timeout`.
Thankfully, the only way `cancel_timeout` can be called is from
TimeoutToken's `cancel_timeout`, which is crate-internal only and is
only called from Timeout's and Interval's drop fn's.
This change simply adds new clarifying documentation around
TokenTimeout's cancel_timeout to "future proof" anybody looking to use
cancel_timeout directly (not just on drop).
This patch disables various Unix-specific platform features that are
not enabled on Fuchsia. It also updates the mio version to 0.6.10,
which is the first release that supports Fuchsia.
This commit adds a general-purpose method for querying the readiness of a
`PollEvented` type. This new method, `poll_ready`, takes a blanket `mio::Ready`
and tests if any part of it is ready. The purpose of this is to expose
platform-specific events through `PollReady` such as `hup` and `error` events
other than just the platform-agnostic readable/writable events.
The semanatics of this method are:
* The `poll_ready` function takes a mask, and the return value is either
`Async::Ready` with a subset of these events that are ready or `None` if none
of them are ready.
* There can be up to two tasks blocked on a `PollEvented`, so we need to pick
which one is suitable for these new events. Currently all events are routed to
the `read` task unless the writable bit is set. This is mostly only relevant
for multi-task usage or if you're manually calling `need_read` and/or
`need_write`, and hopefully the docs will cover this now.
If the underlying read_bufs() call returned a WouldBlock error,
TcpStream.read_buf() was erroneously calling self.io.need_write(), when
it should actually call self.io.need_read().
* if remaining bytes are smaller then 8 * 1024, allocate 8 * 1024
* otherwise allocate as much as needed to hold the remaining bytes
without re-allocations.
The previous implementation would always use the capacity
of the previous buffer, which would effectively prevent it
from ever shrinking.
This also means that protocol with greater variance in
possible frame sizes would likely be heavily over-allocating.
If these implementations use zero-copy, this would imply
that even small frames kept alive by the client would use
large amounts of memory.
The change is motivated by the implementation of the
cassandra-protocol, which allows frames of up to 256MB
in size, which solely depend on the kind of query.
Using Linux's `epoll(7)` interface, a `EPOLLHUP` condition is
signalled for the reading end of a pipe or socket when the other end
is closed for writing. This may happen in combination with `EPOLLIN`
being signalled (if further data is available for reading), or with
`EPOLLIN`.
If `EPOLLHUP` is signalled without `EPOLLIN` it indicates an immediate
EOF condition, which will result in the next `read()` suceeding with 0
bytes read. It is thus not required to handle `EPOLLHUP` specially in
the reader, but we need to indicate readiness upon encountering it.
Looking at the `kqueue` and `windows` mio backends, they seem to be
turn a detected EOF and connection reset into `mio::Ready::hup()`
events, so it may be reasonable to speculate that this change is an
improvement for these platforms as well.
Fixes#117.
Note that this brings in a lot of machinery just for the sake of
wrapping file descriptors as PollEvented. This should maybe be moved
into the public API of some crate.
* Move to `std::net` as it's all purely UDP related
* Rename to `UdpCodec` and `UdpFramed` to give a consistent `Udp` prefix
* Add `RecvDgram`, rename `SendDGram` to `SendDgram`
* Touch up some style here and there
This commit makes a few changes to the Sink implementation on Framed:
* Backpressure is implemented for `start_send`. If the write buffer is
over 8KiB and can't be flushed, no new items are accepted.
* 0 length writes to the upstream transport are translated into a
`WriteZero` error, as with `io::Write::write_all`. `write_all` checks
for and ignores `Interrupted` errors, but I do not think this is
necessary for non-blocking writes.
* In `poll_complete`, the upstream transport is not flushed until
*after* writing the entire write buffer.
traits into `Codec`
A previous commit refactored such that `Encode` and `Decode` are
implemented directly on the types being encoded or decoded. This was
thought to be less expressive but more convenient than having a separate
notion of a (stateful) encoder or decoder.
However, there are certain situations where the approach is just too
limiting: you're required to implemented `Decode` and `Encode` for types
you don't "own" and can't newtype.
This commit moves back to a setup where `Self` represents the
encoder/decoder state; it also merges the two traits into a single
`Codec` trait, since they are currently always used together.
- Gets rid of `easy` module, instead providing framing support directly
in the `io` module.
- In particular, adds a framing adapter directly to the `Io` trait,
which gives you a Stream + Sink object. That object can then be
`split` into separate `Stream` and `Sink` objects if needed.
- Deprecates the `FramedIo` trait; that's now just Stream + Sink.
- Updates the line framing test to use the stream/sink combinators.
Typically this happens automatically as the `E` in `PollEvented<E>` is an owned
reference (e.g. a `TcpStream`) where dropping that will close the resource,
automatically unregistering it from the event loop. In some situations, however,
this isn't always the case, so the deregistering needs to happen manually.
This commit makes a few tweaks to the new `easy` module:
- Rename `Parse` to `Decode`, and `Serialize` to `Encode`.
- Don't use `Poll` for the `decode` method; we prefer to reserve
that type for actual aync events, and in particular for a `NotReady`
result to imply that some task scheduling has taken place. Instead,
use an internal `Option`.
This commit extracts the concrete implementation of `FrameIo` in tokio-proto to
tokio-core under the name `EasyFramed`. This extraction is accompanied with a
new `EasyBuf` buffer type to work with when parsing types.
The purpose of this movement is to provide a clear and easy entry point at the
`FramedIo` layer for those who need it. Eventually these buffer types will get
replaced or moved to the `bytes` crate, but in the interest of an 0.1 release
and remaining backwards compatible with the tokio-core 0.1 release this is
adding a separate module.
The other read futures (read_exact, read_until, etc.) all expose their
concrete future types so that function signatures can return them, but
until now `read()` didn't. Exposing it with the name "Read" causes
naming conflicts with the std::io::Read trait, so the easiest thing to
do is to just change the name. Importing std::io::Read under a different
name would've been an option too, but that would probably be annoying
for consumers in the same way it's annoying for us.
The original PR (https://github.com/tokio-rs/tokio-core/pull/29) decided
that "read" was a better name than "read_some", so I'm leaving the top
level functions unchanged. I don't have a strong opinion about it one
way or the other, but I *do* think it's worth bikeshedding a little bit.
Python's asyncio library actually ended up with a very similar issue
around naming inconsistency between the sync and async worlds, and we
can hopefully avoid repeating that: https://bugs.python.org/issue22279
* Send source address of message in addition to the message to connected clients.
* Move `spawn_fn` to the bottom.
* Use `map` instead of `and_then` if there is no need for blocking.
* `map` to unit where values are not needed anymore.
In general it's easier to implement and should have more predictable performance
semantics for applications in general. More serious timer usage can go through
`tokio-timer` which has properly configurable timer wheels and such.
Closes#2Closes#7
* Handle -> Remote
* Pinned -> Handle
All APIs now take a `&Handle` by default and in general can return an immediate
`io::Result` instead of an `IoFuture`. This reflects how most usage will likely
be done through handles rather than remotes, and also all previous functionality
can be recovered with a `oneshot` plus `Remote::spawn`.
Closes#15
* Remove `LoopData` as it's no longer necessary
* Add `LoopHandle::spawn` to spawn new futures onto an event loop
* Add `LoopData::spawn` to also spawn new futures onto an event loop
* Rejigger the implementation of the event loop a bit (make a slab of futures),
but otherwise everything else is pretty constant.
Split it up into a number of targeted modules for each purpose, for example loop
data, I/O sources, timeouts, and channels. No actual change is intended to be
part of this commit.
This commit contains a few refactorings, but the major goal is to remove the
`Arc` that's stored inside of each `ReadinessStream` and `Scheduled` slot in the
event loop. The original purpose of this `Arc` was to share the I/O object among
the concrete handle itself and the event loop. The event loop would then change
how the socket is registered over time and then deregister it when it gets a
"shutdown request".
Nowadays, however, once an I/O object is registered with the event loop it's
never updated. Additionally, we don't actually need to call `deregister` but can
rather just instead close the I/O object itself and let the kernel/event loop
take care of the cleanup. All we need to do on deregistering is free up the slab
entry.
The major result of this commit is that I/O objects no longer need to be `Sync`
(as they're not stored in an `Arc`). Instead they just need to be `Send +
'static` as one might otherwise expect.
Along the way this also refactors a few pieces here and there to make more sense
in this new scheme. The `ReadinessStream` type now has a type parameter
indicating an owned reference to the I/O object it wraps. This can be accessed
via the `get_ref` and `get_mut` methods. Additionally I/O tokens on the event
loop are now a full-fledged `IoToken` type which we can change in the future if
we need to.
We know that the future will never persist beyond this stack frame, so we can
just leave its ownership on the stack frame itself and receive notifications off
the event loop that we need to poll it.
The core trait and associated types no longer require the `'static` bounds, and
all associated methods have also had the `'static` bounds removed. The only
location for the `'static` bound is `forget`.
While the restriction of `'static` on `forget` is here to stay, we'll soon
enable `Loop::run` to take a non-`'static` future, allowing driving a
non-`'static` future.
Also involved yet another round of bug fixes to the timer wheel as well as an
unfortunately serious rejiggering of the level-translation into libcurl. Eew.
* Auto-register interest whenever we see WouldBlock
* Remove implementations of `Stream<Item=Ready>`, no longer needed
* Add explicit `poll_{read,write}` methods, if needed
* Remove all I/O streams, libstd ones suffice
* Update all I/O futures
A more appealing model is actually just automatically inferring what needs to be
scheduled based on what actions are done during poll. For example if during a
poll you check a oneshot channel, then the current task is registered for being
woken up if it's not ready. Similarly this will apply to I/O where if I/O is
attempted but we see EAGAIN then we'll schedule the task to get notified when
it's ready.
This may also have performance benefits in some niche situations because you
don't need to recompute where you are in the state machine both during poll and
during schedule. Instead, it now happens all at once.
This bound existed for two primary reasons, both detail below, and both of which
have now been solved.
One of the primary reasons this existed was due to the presence of `tailcall`.
Each standard combinator will call `tailcall` as appropriate, storing the
resulting trait object. Storing trait objects influences the applicatoin of the
`Send` and `Sync` bounds normally, but a key insight here is that we're not
storing trait objects but rather just pieces of otherwise internal futures.
With this insight the main storage for these futures, `Collapsed`, could simply
implement `Send` so long as the future itself originally implemented `Send`.
This in turn means that `tailcall` must be an `unsafe` method, but it seems well
worth the benefit of relaxing the `Send` bound.
The second primary reason for this bound was so the `Task` itself could be send.
This is critical for ensuring that futures can receive notifications from
multiple threads (e.g. be a future waiting on sources of multiple events).
Another key insight here, however, is that only the *outer* future needs to be
`Send`. We already have a solution, with `LoopData`, to make non-`Send` data
`Send`. By implementing `Future` directly for `LoopData<F: Future>`, this means
that it's trivial to make any future sendable by simply pinning it to an event
loop!
With these two pieces combined, it means that `Send` is no longer needed as a
bound on the `Future` and `Stream` traits. It may practically mean that
`LoopData` is used commonly in some scenarios, but that's quite a small price to
pay for relaxing the requirements of the core trait.
Some other ramifications of this change are:
* The `Future::boxed` and `Stream::boxed` methods now require `Self` to adhere
to `Send`. This is expected to be the most common case, and in the less common
case of not-`Send` `Box::new` can be used.
* Two new type aliases, `BoxFuture` and `BoxStream` have been introduced to
assist in writing APIs that return a trait object which is `Send`. Both of
these type aliases package in the `Send` bound.
* A new `LoopPin` type, added in the previous commit, can be used to easily
generate handles that can be used to pin futures to an event loop without
having a literal reference to the event loop itself.
It's already Sync because it's safe to access amongst many threads by the sheer
fact that data is only accessed on one thread. That is, all other concurrent
accessors, if any, will receive `None`.
This type acts for a handle to storage of non-`Send` data. The handle itself is
sendable across threads and is therefore suitable for storage in a `Future`.
This data uses communication internally and a new method on `Task` to ensure
that when the data needs to be accessed the future will find its way to the
right thread.
More on this type coming soon!
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Carl Lerche
