mirror of
https://github.com/mozilla/gecko-dev.git
synced 2024-11-30 00:01:50 +00:00
76e3009ab8
This is motivated by three separate but related problems: 1. Our concept of recursion depth is broken for things that run from AfterProcessNextEvent observers (e.g. Promises). We decrement the recursionDepth counter before firing observers, so a Promise callback running at the lowest event loop depth has a recursion depth of 0 (whereas a regular nsIRunnable would be 1). This is a problem because it's impossible to distinguish a Promise running after a sync XHR's onreadystatechange handler from a top-level event (since the former runs with depth 2 - 1 = 1, and the latter runs with just 1). 2. The nsIThreadObserver mechanism that is used by a lot of code to run "after" the current event is a poor fit for anything that runs script. First, the order the observers fire in is the order they were added, not anything fixed by spec. Additionally, running script can cause the event loop to spin, which is a big source of pain here (bholley has some nasty bug caused by this). 3. We run Promises from different points in the code for workers and main thread. The latter runs from XPConnect's nsIThreadObserver callbacks, while the former runs from a hardcoded call to run Promises in the worker event loop. What workers do is particularly problematic because it means we can't get the right recursion depth no matter what we do to nsThread. The solve this, this patch does the following: 1. Consolidate some handling of microtasks and all handling of stable state from appshell and WorkerPrivate into CycleCollectedJSRuntime. 2. Make the recursionDepth counter only available to CycleCollectedJSRuntime (and its consumers) and remove it from the nsIThreadInternal and nsIThreadObserver APIs. 3. Adjust the recursionDepth counter so that microtasks run with the recursionDepth of the task they are associated with. 4. Introduce the concept of metastable state to replace appshell's RunBeforeNextEvent. Metastable state is reached after every microtask or task is completed. This provides the semantics that bent and I want for IndexedDB, where transactions autocommit at the end of a microtask and do not "spill" from one microtask into a subsequent microtask. This differs from appshell's RunBeforeNextEvent in two ways: a) It fires between microtasks, which was the motivation for starting this. b) It no longer ensures that we're at the same event loop depth in the native event queue. bent decided we don't care about this. 5. Reorder stable state to happen after microtasks such as Promises, per HTML. Right now we call the regular thread observers, including appshell, before the main thread observer (XPConnect), so stable state tasks happen before microtasks.
77 lines
2.8 KiB
Plaintext
77 lines
2.8 KiB
Plaintext
/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*-
|
|
*
|
|
* This Source Code Form is subject to the terms of the Mozilla Public
|
|
* License, v. 2.0. If a copy of the MPL was not distributed with this
|
|
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
|
|
|
|
#include "nsISupports.idl"
|
|
|
|
interface nsIRunnable;
|
|
%{ C++
|
|
template <class T> struct already_AddRefed;
|
|
%}
|
|
|
|
/**
|
|
* Interface for the native event system layer. This interface is designed
|
|
* to be used on the main application thread only.
|
|
*/
|
|
[uuid(7cd5c71d-223b-4afe-931d-5eedb1f2b01f)]
|
|
interface nsIAppShell : nsISupports
|
|
{
|
|
/**
|
|
* Enter an event loop. Don't leave until exit() is called.
|
|
*/
|
|
void run();
|
|
|
|
/**
|
|
* Exit the handle event loop
|
|
*/
|
|
void exit();
|
|
|
|
/**
|
|
* Give hint to native event queue notification mechanism. If the native
|
|
* platform needs to tradeoff performance vs. native event starvation this
|
|
* hint tells the native dispatch code which to favor. The default is to
|
|
* prevent native event starvation.
|
|
*
|
|
* Calls to this function may be nested. When the number of calls that pass
|
|
* PR_TRUE is subtracted from the number of calls that pass PR_FALSE is
|
|
* greater than 0, performance is given precedence over preventing event
|
|
* starvation.
|
|
*
|
|
* The starvationDelay arg is only used when favorPerfOverStarvation is
|
|
* PR_FALSE. It is the amount of time in milliseconds to wait before the
|
|
* PR_FALSE actually takes effect.
|
|
*/
|
|
void favorPerformanceHint(in boolean favorPerfOverStarvation,
|
|
in unsigned long starvationDelay);
|
|
|
|
/**
|
|
* Suspends the use of additional platform-specific methods (besides the
|
|
* nsIAppShell->run() event loop) to run Gecko events on the main
|
|
* application thread. Under some circumstances these "additional methods"
|
|
* can cause Gecko event handlers to be re-entered, sometimes leading to
|
|
* hangs and crashes. Calls to suspendNative() and resumeNative() may be
|
|
* nested. On some platforms (those that don't use any "additional
|
|
* methods") this will be a no-op. Does not (in itself) stop Gecko events
|
|
* from being processed on the main application thread. But if the
|
|
* nsIAppShell->run() event loop is blocked when this call is made, Gecko
|
|
* events will stop being processed until resumeNative() is called (even
|
|
* if a plugin or library is temporarily processing events on a nested
|
|
* event loop).
|
|
*/
|
|
void suspendNative();
|
|
|
|
/**
|
|
* Resumes the use of additional platform-specific methods to run Gecko
|
|
* events on the main application thread. Calls to suspendNative() and
|
|
* resumeNative() may be nested. On some platforms this will be a no-op.
|
|
*/
|
|
void resumeNative();
|
|
|
|
/**
|
|
* The current event loop nesting level.
|
|
*/
|
|
readonly attribute unsigned long eventloopNestingLevel;
|
|
};
|