Previously, dropping the Write handle would issue a `shutdown(Both)`. However,
shutting down the read half is not portable and not the correct action to take.
This changes the behavior of OwnedWriteHalf to only perform a `shutdown(Write)`
on drop.
Previously, in the broadcast channel, receiver wakers were passed to the
sender via an atomic stack with allocated nodes. When a message was
sent, the stack was drained. This caused a problem when many receivers
pushed a waiter node then dropped. The waiter node remained indefinitely
in cases where no values were sent.
This patch switches broadcast to use the intrusive linked-list waiter
strategy used by `Notify` and `Semaphore.
In some cases, when a call to `block_in_place` completes, the runtime is
reinstated on the thread. In this case, the task budget must also be set
in order to avoid starving other tasks on the worker.
Do not export the `scoped_thread_local` macro outside of the Tokio
crate. This is not considered a breaking change as the macro never
worked if used from outside of the crate due to the generated code
referencing crate-private types.
Previously, the function picking the default number of threads for the
threaded runtime did not factor in `max_threads`. Instead, it only used
the value returned by `num_cpus`. However, if `num_cpus` returns a value
greater than `max_threads`, then the function would panic.
This patch fixes the function by limiting the default number of threads
by `max_threads`.
Fixes#2452
Fixes a couple bugs in the work-stealing queue introduced as
part of #2315. First, the cursor needs to be able to represent more
values than the size of the buffer. This is to be able to track if
`tail` is ahead of `head` or if they are identical. This bug resulted in
the "overflow" path being taken before the buffer was full.
The second bug can happen when a queue is being stolen from concurrently
with stealing into. In this case, it is possible for buffer slots to be
overwritten before they are released by the stealer. This is harder to
happen in practice due to the first bug preventing the queue from
filling up 100%, but could still happen. It triggered an assertion in
`steal_into`. This bug slipped through due to a bug in loom not
correctly catching the case. The loom bug is fixed as part of
tokio-rs/loom#119.
Fixes: #2382
The new queue uses `u8` to track offsets. Cursors are expected to wrap.
An operation was performed with `+` instead of `wrapping_add`. This was
not _obviously_ issue before as it is difficult to wrap a `usize` on
64bit platforms, but wrapping a `u8` is trivial.
The fix is to use `wrapping_add` instead of `+`. A new test is added
that catches the issue.
Fixes#2361
The work-stealing scheduler includes an optimization where each worker
includes a single slot to store the **last** scheduled task. Tasks in
scheduler's LIFO slot are executed next. This speeds up and reduces
latency with message passing patterns.
Previously, this optimization was susceptible to starving other tasks in
certain cases. If two tasks ping-ping between each other without ever
yielding, the worker would never execute other tasks.
An early PR (#2160) introduced a form of pre-emption. Each task is
allocated a per-poll operation budget. Tokio resources will return ready
until the budget is depleted, at which point, Tokio resources will
always return `Pending`.
This patch leverages the operation budget to limit the LIFO scheduler
optimization. When executing tasks from the LIFO slot, the budget is
**not** reset. Once the budget goes to zero, the task in the LIFO slot
is pushed to the back of the queue.
Fixes a test from a PR that was written before the recent loom upgrade.
A change in the details how loom executes models resulted in the test to
start failing. The fix is to reduce the number of iterations performed
by the test.
Loom is having a big refresh to improve performance and tighten up the
concurrency model. This diff tracks those changes.
Included in the changes is the removal of `CausalCell` deferred checks.
This is due to it technically being undefined behavior in the C++11
memory model. To address this, the work-stealing queue is updated to
avoid needing this behavior. This is done by limiting the queue to have
one concurrent stealer.
Since the original shell runtime was implemented, utilities have been
added to encapsulate `unsafe`. The shell runtime is now able to use
those utilities and not include its own `unsafe` code.
A refactor of the scheduler internals focusing on simplifying and
reducing unsafety. There are no fundamental logic changes.
* The state transitions of the core task component are refined and
reduced.
* `basic_scheduler` has most unsafety removed.
* `local_set` has most unsafety removed.
* `threaded_scheduler` limits most unsafety to its queue implementation.
Allow storing the intrusive linked-list pointers in an arbitrary
location in the node. This is in preparation for using the linked list
in the scheduler.
In order to make using the intrusive linked list more flexible, a trait
is introduced to abstract mapping an entry to raw pointers and the next
/ prev pointers. This also pushes more unsafety onto the user.
`StreamMap` is similar to `StreamExt::merge` in that it combines source
streams into a single merged stream that yields values in the order that
they arrive from the source streams. However, `StreamMap` has a lot more
flexibility in usage patterns.
`StreamMap` can:
- Merge an arbitrary number of streams.
- Track which source stream the value was received from.
- Handle inserting and removing streams from the set of managed streams
at any point during iteration.
All source streams held by `StreamMap` are indexed using a key. This key
is included with the value when a source stream yields a value. The key
is also used to remove the stream from the `StreamMap` before the stream
has completed streaming.
Because the `StreamMap` API moves streams during runtime, both streams
and keys must be `Unpin`. In order to insert a `!Unpin` stream into a
`StreamMap`, use `pin!` to pin the stream to the stack or `Box::pin` to
pin the stream in the heap.
The `macros` feature flag was ommitted despite the fact that these
macros require the feature flag to function. The macros are now scoped
by the `macros` feature flag.
This is *not* a breaking change due to the fact that the macros were
broken without the `macros` feature flag in the first place.
Provides a `try_join!` macro that supports concurrently driving multiple
`Result` futures on the same task and await the completion of all the
futures as `Ok` or the **first** `Err` future.
Used for stack pinning and based on `pin_mut!` from the pin-util crate.
Pinning is used often when working with stream operators and the select!
macro. Given the small size of `pin!` it makes more sense to include a
version than re-export one from a separate crate or require the user to
depend on `pin-util` themselves.
Provides a `select!` macro for concurrently waiting on multiple async
expressions. The macro has similar goals and syntax as the one provided
by the `futures` crate, but differs significantly in implementation.
First, this implementation does not require special traits to be
implemented on futures or streams (i.e., no `FuseFuture`). A design goal
is to be able to pass a "plain" async fn result into the select! macro.
Even without `FuseFuture`, this `select!` implementation is able to
handle all cases the `futures::select!` macro can handle. It does this
by supporting pre-poll conditions on branches and result pattern
matching. For pre-conditions, each branch is able to include a condition
that disables the branch if it evaluates to false. This allows the user
to guard futures that have already been polled, preventing double
polling. Pattern matching can be used to disable streams that complete.
A second big difference is the macro is implemented almost entirely as a
declarative macro. The biggest advantage to using this strategy is that
the user will not need to alter the rustc recursion limit except in the
most extreme cases.
The resulting future also tends to be smaller in many cases.
The Tokio runtime provides a "shell" runtime when `rt-core` is not
available. This shell runtime is enough to support `#[tokio::main`] and
`#[tokio::test].
A previous change disabled these two attr macros when `rt-core` was not
selected. This patch fixes this by re-enabling the `main` and `test`
attr macros without `rt-core` and adds some integration tests to prevent
future regressions.
Loom currently does not compile on windows due to a
transitive dependency on `generator`. The `generator`
crate builds have started to fail on windows CI. Loom
is not run under windows, however, so removing the
loom dependency on windows is sufficient to fix CI.
Refs: https://github.com/Xudong-Huang/generator-rs/issues/19
Provides an asynchronous equivalent to `Iterator::collect()`. A sealed
`FromStream` trait is added. Stabilization is pending Rust supporting
`async` trait fns.
Provides an equivalent to stream `select()` from futures-rs. `merge`
best describes the operation (vs. `select`). `futures-rs` named the
operation "select" for historical reasons and did not rename it back to
`merge` in 0.3. The operation is most commonly named `merge` else where
as well (e.g. ReactiveX).
Previously, when the threaded scheduler was in the shutdown process, it
would hold a lock while dropping in-flight tasks. If those tasks
included a drop handler that attempted to wake a second task, the wake
operation would attempt to acquire a lock held by the scheduler. This
results in a deadlock.
Dropping the lock before dropping tasks resolves the problem.
Fixes#2046
Previously, if an IO event was received during the runtime shutdown
process, it was possible to enter a deadlock. This was due to the
scheduler shutdown logic not expecting tasks to get scheduled once the
worker was in the shutdown process.
This patch fixes the deadlock by checking the queues for new tasks after
each call to park. If a new task is received, it is forcefully shutdown.
Fixes#2061
Tweak context to remove more fns and usage of `Option`. Remove
`ThreadContext` struct as it is reduced to just `Handle`. Avoid passing
around individual driver handles and instead limit to the
`runtime::Handle` struct.
This patch improves the behavior of frozen time (a testing utility made
available with the `test-util` feature flag). Instead of of requiring
`time::advance` to be called in order to advance the value returned by
`Instant::now`, calls to `time::Driver::park_timeout` will use the
provided duration to advance the time.
This is the desired behavior as the timeout is used to indicate when the
next scheduled delay needs to be fired.
Extend internal semaphore to support batch operations. With this PR,
consumers of the semaphore are able to atomically request more than one
permit. This is useful for implementing a RwLock.
Storing a `Runtime` value in a thread-local resulted in a panic due to
the inability to access the parker.
This fixes the bug by skipping parking if it fails. In general, there
isn't much that we can do besides not parking.
Fixes#593
Nested spawn_blocking calls would result in a panic due to the necessary
context not being setup. This patch sets the blocking pool context from
within a blocking pool.
Fixes#1982
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.
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
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.
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.
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.
Carl Lerche