RunExpiredTimeouts has "fudging" code to always ensure that we execute at least one timeout. This is intended to cover cases where an nsITimer fires slightly early, but it means we must be careful not to fire a timer more times than we intend to or we'll execute a timeout prematurely.
Consider a sequences of setTimeout calls alternating in delay between 0ms and 1000ms. When the 1000ms timeout fires, it schedules a 0ms timeout. The setTimeout call itself calls RescheduleTimeoutTimer, which schedules the timer for a 0 ms delay. And once we unwind the 1000ms timeout RunExpiredTimeouts will also schedule the timer for a 0 ms delay. If the timer has fired (remember, it's processed on a completely different thread) in the meantime, we ultimately will get two callbacks from nsITimer for our 0 ms timeout. The first will run the 0 ms timeout and schedule a 1000 ms timeout, and the second will run the 1000 ms timeout (remember, RunExpiredTimeouts always runs at least one timeout!) ~999 ms ahead of schedule.
The solution is to cancel the timer in RescheduleTimeoutTimer, so that when we call it the second time it will cause any pending events from the first scheduling to be canceled. But this actually doesn't work at all, because of how we use nsITimer. Before worker threads were capable of accepting arbitrary runnables we created TimerThreadEventTarget, which translates the timer firing to the special worker event queue when the timer thread attempts to *dispatch* a runnable to the worker. We still need this for some of the other types of timers (which use control runnables that interrupt JS, and not the regular event queue). But setTimeout can simply run like a normal nsITimer callback now. We need that here, or calling nsITimer::Cancel won't actually do anything, because the timer's event was ignored and TimerThreadEventTarget created its own event.
The bulk of this commit was generated with a script, executed at the top
level of a typical source code checkout. The only non-machine-generated
part was modifying MFBT's moz.build to reflect the new naming.
CLOSED TREE makes big refactorings like this a piece of cake.
# The main substitution.
find . -name '*.cpp' -o -name '*.cc' -o -name '*.h' -o -name '*.mm' -o -name '*.idl'| \
xargs perl -p -i -e '
s/nsRefPtr\.h/RefPtr\.h/g; # handle includes
s/nsRefPtr ?</RefPtr</g; # handle declarations and variables
'
# Handle a special friend declaration in gfx/layers/AtomicRefCountedWithFinalize.h.
perl -p -i -e 's/::nsRefPtr;/::RefPtr;/' gfx/layers/AtomicRefCountedWithFinalize.h
# Handle nsRefPtr.h itself, a couple places that define constructors
# from nsRefPtr, and code generators specially. We do this here, rather
# than indiscriminantly s/nsRefPtr/RefPtr/, because that would rename
# things like nsRefPtrHashtable.
perl -p -i -e 's/nsRefPtr/RefPtr/g' \
mfbt/nsRefPtr.h \
xpcom/glue/nsCOMPtr.h \
xpcom/base/OwningNonNull.h \
ipc/ipdl/ipdl/lower.py \
ipc/ipdl/ipdl/builtin.py \
dom/bindings/Codegen.py \
python/lldbutils/lldbutils/utils.py
# In our indiscriminate substitution above, we renamed
# nsRefPtrGetterAddRefs, the class behind getter_AddRefs. Fix that up.
find . -name '*.cpp' -o -name '*.h' -o -name '*.idl' | \
xargs perl -p -i -e 's/nsRefPtrGetterAddRefs/RefPtrGetterAddRefs/g'
if [ -d .git ]; then
git mv mfbt/nsRefPtr.h mfbt/RefPtr.h
else
hg mv mfbt/nsRefPtr.h mfbt/RefPtr.h
fi
--HG--
rename : mfbt/nsRefPtr.h => mfbt/RefPtr.h
When workers shut down we discard the event queue rather than running it to completion. Originally workers managed their event queue themselves and would simply iterate through the array of events and cancel them all. After bug 914762 this was done by setting a (thread-)global "canceling" flag and then calling NS_ProcessPendingEvents. But this neglects that a shut down request can be received while the worker is in a sync queue. In this case, calling NS_ProcessPendingEvents will process any events pending in the sync queue, which is *not* the queue we need to cancel.
The fix is, if we are in a sync queue when NotifyInternal is called, to defer clearing the queue until the top-most sync queue is destroyed and we are about to return to the regular event queue. Only then can we call NS_ProcessPendingEvents to clear out the queue. Because we can never process any events from this queue while sync queues are active, the timing of the mass cancellation is unchanged from the perspective of events in the regular queue.
When workers shut down we discard the event queue rather than running it to completion. Originally workers managed their event queue themselves and would simply iterate through the array of events and cancel them all. After bug 914762 this was done by setting a (thread-)global "canceling" flag and then calling NS_ProcessPendingEvents. But this neglects that a shut down request can be received while the worker is in a sync queue. In this case, calling NS_ProcessPendingEvents will process any events pending in the sync queue, which is *not* the queue we need to cancel.
The fix is, if we are in a sync queue when NotifyInternal is called, to defer clearing the queue until the top-most sync queue is destroyed and we are about to return to the regular event queue. Only then can we call NS_ProcessPendingEvents to clear out the queue. Because we can never process any events from this queue while sync queues are active, the timing of the mass cancellation is unchanged from the perspective of events in the regular queue.
--HG--
extra : rebase_source : f67fbee27c0751068a4e7aaf692cbfc1d3c9aa7c
Per the product discussion, the Notification API should be disabled in
ServiceWorker in release builds for 42 since the UX isn't great [1].
The aim is to release in 44.
Apologies for the code duplication for pref checking in Notification and
ServiceWorkerRegistration. There isn't a easy way to get
ServiceWorkerRegistration's generated binding to include Notification.h without
having an attribute/method that uses Notification.
[1]: https://mana.mozilla.org/wiki/x/TgAJAw
--HG--
extra : commitid : 5dtc2E63kuM
extra : rebase_source : 4265dcd154462aa4f3b915e9e898fe7b82bf9afc
Right now, synthetic Responses did not have a valid channel info. When these
were saved in the Cache, and then restored, the restored Response did have
a ChannelInfo, but that ChannelInfo did not have a valid security info.
Passing this to respondWith() then caused the interception to fail.
This patch modifies Response::Constructor() to initialize its ChannelInfo from
the global. ChannelInfo can now initialize itself from a nsIDocument. All
workers now store their ChannelInfo on the WorkerLoadInfo.
--HG--
extra : commitid : L1wltwPICd8
extra : rebase_source : 8dab4c414eb50e02a00dd2cb3ee848b811060e70
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.
client.focus() now directly uses the DOMServiceWorkerFocusClient event. The
platform popup checking is not available on service workers. Instead each
worker maintains a counter of if it is allowed to interact with windows. This
counter is currently only incremented by the notificationclick event and
dropped after the event has been dispatched.
Since acquiring a client is an async operation most service workers will
perform in notificationclick, an additional extension is granted after the event
during which the service worker may focus a client. This extension is only granted
if the script invokes NotificationEvent.waitUntil() at which point the timer begins.
The extension is terminated when the Promise passed to waitUntil() is fulfilled, or
the timer expires, whichever comes first.
--HG--
extra : commitid : 5wavKTRZWcy
extra : rebase_source : fc8ab4ef6c9bf384b5525b0bc979b3cedc4e1d6c
client.focus() now directly uses the DOMServiceWorkerFocusClient event. The
platform popup checking is not available on service workers. Instead each
worker maintains a counter of if it is allowed to interact with windows. This
counter is currently only incremented by the notificationclick event and
dropped after the event has been dispatched.
Since acquiring a client is an async operation most service workers will
perform in notificationclick, an additional extension is granted after the event
during which the service worker may focus a client. This extension is only granted
if the script invokes NotificationEvent.waitUntil() at which point the timer begins.
The extension is terminated when the Promise passed to waitUntil() is fulfilled, or
the timer expires, whichever comes first.
--HG--
extra : commitid : 5DUAOg4j41K
extra : rebase_source : 775d36e41c3d3b62e5189b032220fa2469ac237d
extra : source : 8300859f8e9751ca663f96fae3375dfda8b2ad13
Currently we don't check the dom.push.enabled pref in some cases for
some of these interfaces. This patch unifies how all of these
interfaces are exposed to Window, Worker, and ServiceWorker.
Boris Zbarsky
Andrea Marchesini
Kyle Huey
Eddy Bruel
Carsten "Tomcat" Book
Phil Ringnalda
Olli Pettay
Ehsan Akhgari
Nathan Froyd
Morris Tseng
Wes Kocher
Catalin Badea
Nigel Babu
Nikhil Marathe
Michael Layzell
Francisco Jordano
Ryan VanderMeulen