The current code assumes that it can get the ServiceWorkerManager in the child
process to send a message to the proper service worker. That isn't true, we
need to ask the parent to do it for us.
Differential Revision: https://phabricator.services.mozilla.com/D14773
--HG--
extra : moz-landing-system : lando
Some prefs need to be available before IPC is started, so we serialize a
snapshot when we start launching the process, and then stream further
changes over IPC messages. However, async launch introduces a window
between the snapshot and when the parent can start sending messages,
during which other code can run on the main thread and change prefs.
In order to not lose those updates, they're queued and sent when the
launch is complete.
Depends on D14089
Differential Revision: https://phabricator.services.mozilla.com/D14090
--HG--
extra : moz-landing-system : lando
We need content processes that are created but not finished launching
(not "alive" yet) to be treated differently from ones that have exited
(no longer "alive"), so the boolean mIsAlive is expanded to a 3-state
enumeration, which could be expanded more in the future if needed.
(This is similar to GeckoChildProcessHost::mProcessState, but it's
synchronized with the rest of the ContentParent's state, which can lag
the GeckoChildProcessHost state due to runnable dispatch.)
This patch also removes mIsAvailable/IsAvailable/MarkAsTroubled, which
are unused as of bug 1459212.
Differential Revision: https://phabricator.services.mozilla.com/D14089
--HG--
extra : moz-landing-system : lando
This patch adds some telemetry histograms:
* CONTENT_PROCESS_LAUNCH_IS_SYNC - boolean, true if the content process
was launched synchronously (blocking the main thread)
* CONTENT_PROCESS_SYNC_LAUNCH_MS - the time consumed by sync launch;
the main thread will be busy or blocked for this entire time
* CONTENT_PROCESS_LAUNCH_TOTAL_MS - the total time elapsed from the
start of async content process launch until the launch promise is
resolved and the ContentParent can be sent IPDL messages
* CONTENT_PROCESS_LAUNCH_MAINTHREAD_MS - the time consumed on the parent
process main thread during async content process launch; typically this
is due to ContentParent::Init.
* CHILD_PROCESS_LAUNCH_MS - for any kind of Gecko child process
(including plugins, GPU, etc.), the time taken in the common process
launch code (which is run off-main-thread)
The probes restricted to async content process launch don't have "async"
in the name because that will eventually become the only kind of content
process launch.
Depends on D8943
Differential Revision: https://phabricator.services.mozilla.com/D8944
--HG--
extra : moz-landing-system : lando
There are several layers to this patch:
1. GeckoChildProcessHost now exposes a promise that's resolved when
the process handle is available (or rejected if launch failed), as a
nonblocking alternative to LaunchAndWaitForProcessHandle.
2. ContentParent builds on this with the private method
LaunchSubprocessAsync and the public method PreallocateProcessAsync;
synchronous launch continues to exist for the regular on-demand launch
path, for the time being.
3. PreallocatedProcessManager now uses async launch, and handles the new
"launch in progress" state appropriately.
Depends on D8942
Differential Revision: https://phabricator.services.mozilla.com/D8943
--HG--
extra : moz-landing-system : lando
The first attempt at async launch tried to hide the asynchrony inside
IPC, by making the process seem to be launched enough to construct new
channels and send it messages, and lazily blocking on the pid/handle.
Unfortunately, in practice we wind up needing the pid/handle immediately,
and this requirement is too deeply embedded in IPC for that to be viable.
(The alternative that will be used instead -- exposing process launch via
an explicitly asynchronous promise interface -- is made simpler by
Project Fission's upcoming rewrite of how the DOM requests new content
processes.)
Depends on D8941
Differential Revision: https://phabricator.services.mozilla.com/D8942
--HG--
extra : moz-landing-system : lando
By replacing nsWebBrowser's implementation of the
nsIBaseWindow.initWindow and nsIBaseWindow.create with a new static
nsWebBrowser::Create method we make it possible to pass arguments
directly when creating an nsWebBrowser, for example the opener
BrowsingContext. As a bonus we can do away with
nsWebBrowser::mInitInfo!
Differential Revision: https://phabricator.services.mozilla.com/D12634
--HG--
extra : moz-landing-system : lando
This patch adds some telemetry histograms:
* CONTENT_PROCESS_LAUNCH_IS_SYNC - boolean, true if the content process
was launched synchronously (blocking the main thread)
* CONTENT_PROCESS_SYNC_LAUNCH_MS - the time consumed by sync launch;
the main thread will be busy or blocked for this entire time
* CONTENT_PROCESS_LAUNCH_TOTAL_MS - the total time elapsed from the
start of async content process launch until the launch promise is
resolved and the ContentParent can be sent IPDL messages
* CONTENT_PROCESS_LAUNCH_MAINTHREAD_MS - the time consumed on the parent
process main thread during async content process launch; typically this
is due to ContentParent::Init.
* CHILD_PROCESS_LAUNCH_MS - for any kind of Gecko child process
(including plugins, GPU, etc.), the time taken in the common process
launch code (which is run off-main-thread)
The probes restricted to async content process launch don't have "async"
in the name because that will eventually become the only kind of content
process launch.
Depends on D8943
Differential Revision: https://phabricator.services.mozilla.com/D8944
--HG--
extra : moz-landing-system : lando
There are several layers to this patch:
1. GeckoChildProcessHost now exposes a promise that's resolved when
the process handle is available (or rejected if launch failed), as a
nonblocking alternative to LaunchAndWaitForProcessHandle.
2. ContentParent builds on this with the private method
LaunchSubprocessAsync and the public method PreallocateProcessAsync;
synchronous launch continues to exist for the regular on-demand launch
path, for the time being.
3. PreallocatedProcessManager now uses async launch, and handles the new
"launch in progress" state appropriately.
Depends on D8942
Differential Revision: https://phabricator.services.mozilla.com/D8943
--HG--
extra : moz-landing-system : lando
The first attempt at async launch tried to hide the asynchrony inside
IPC, by making the process seem to be launched enough to construct new
channels and send it messages, and lazily blocking on the pid/handle.
Unfortunately, in practice we wind up needing the pid/handle immediately,
and this requirement is too deeply embedded in IPC for that to be viable.
(The alternative that will be used instead -- exposing process launch via
an explicitly asynchronous promise interface -- is made simpler by
Project Fission's upcoming rewrite of how the DOM requests new content
processes.)
Depends on D8941
Differential Revision: https://phabricator.services.mozilla.com/D8942
--HG--
extra : moz-landing-system : lando
This patch also takes the timeout for killing a content process back to its
original value and removes a related but long unused field.
Differential Revision: https://phabricator.services.mozilla.com/D11739
--HG--
extra : moz-landing-system : lando
Cache the sandboxing command line parameters used when starting a new content process, avoiding calls to realpath(3) on the main thread in the parent process for each content process that is started.
Differential Revision: https://phabricator.services.mozilla.com/D10529
--HG--
extra : moz-landing-system : lando
This commit removes the PRenderFrame protocol, while keeping the same ordering
and semantics of graphics IPC initialization.
To do this, some messages are added to PBrowser to simulate the constructor
and destructor of PRenderFrame. Messages that expected a nullable PRenderFrame
are updated to get a boolean instead.
One tricky area is the destruction of PRenderFrame. I've tried to keep it the
same as much as possible, but it's possible it might be slightly semantically
different than IPDL destruction. Destruction will be touched up in a later
patch, so I'm not too concerned.
Differential Revision: https://phabricator.services.mozilla.com/D11057
--HG--
extra : rebase_source : bb8a7896bb4aefb6e9957d8808b755fa76cc00ed
extra : histedit_source : 6377819a946b5b6bc18b15f748229360e42a6f3a
Content process of Android uses sync IPC when initializing LookAndFeel. But
current e10s has LookAndFeel cache for start up of content process.
So we should use it, then remove sync IPC for start up performance
Differential Revision: https://phabricator.services.mozilla.com/D9750
--HG--
extra : moz-landing-system : lando
Pass sandbox parameters to content processes on the command line allowing for early sandbox startup.
Pref'd off behind "security.sandbox.content.mac.earlyinit" until it's ready to be enabled by default.
Once early startup is enabled by default and considered stable, the original sandbox startup code can be removed.
Depends on D6719
Differential Revision: https://phabricator.services.mozilla.com/D6720
--HG--
extra : moz-landing-system : lando
Pass sandbox parameters to content processes on the command
line allowing for early sandbox startup. Limited to Nightly
until confirmed to be stable and ready to ride the trains.
Enable early sandbox startup by default on Nightly and use
pref "security.sandbox.content.mac.earlyinit" to disable
early startup for debugging purposes.
Once early startup is stable, the original sandbox startup
code can be removed.
Depends on D6719
Differential Revision: https://phabricator.services.mozilla.com/D6720
--HG--
extra : moz-landing-system : lando
Pass sandbox parameters to content processes on the command
line allowing for early sandbox startup. Limited to Nightly
until confirmed to be stable and ready to ride the trains.
Enable early sandbox startup by default on Nightly and use
pref "security.sandbox.content.mac.earlyinit" to disable
early startup for debugging purposes.
Once early startup is stable, the original sandbox startup
code can be removed.
Depends on D6719
Differential Revision: https://phabricator.services.mozilla.com/D6720
--HG--
extra : moz-landing-system : lando
This reverts the changes in bug 1360308, bug 1390143 and bug 1469603. Minidump
generation will now only happen on the main process' main thread which might
lead to hangs but is known to be fairly robust. Asynchronous generation proved
too brittle and enormously increased the complexity of this already
hard-to-read code.
Differential Revision: https://phabricator.services.mozilla.com/D5147
--HG--
extra : moz-landing-system : lando
This moves the code that detects very low memory scenarios and grabs memory
reports from the main thread event-loop to the available memory tracker.
Besides removing the overhead of the check from the event-loop code this
increases the likeliness of the reports being gathered by sampling at a
higher frequency but only when we already detected a low-memory scenario. Last
but not least this add checks for low commit-space detection alongside low
virtual-memory detection.
Differential Revision: https://phabricator.services.mozilla.com/D3669
--HG--
extra : moz-landing-system : lando
Have BrowsingContext keep its own cache to enable caching of
BrowsingContexts, especially in the parent process.
This isn't really optimal, since it effectively duplicates the
cache in the child process. BFcache keeps a list of strong pointers to
the list of cached nsDocShells, where each nsDocShell in turn keeps a
reciprocated strong pointer to its BrowsingContext, which in turn is
held in the BrowsingContexts list of cached contexts. Ideally these
caches should be merged.
--HG--
extra : histedit_source : 094370f6d54d83728e8433ec5c47003086146476
Add BrowsingContext to allow the tree structure of docshells to exist
in several processes simultaneously. This is a first step towards
allowing a tree structure preserving separation of docshells across
processes.
--HG--
extra : histedit_source : d3c7f6ab4b9ae76f170c126d669ebd570e52f348
Right now consumers can't know when the parent process has finished talking
to the permission manager. It would be nice to enable consumers to depend
on the status of the asynchronous task using a promise.
Right now consumers can't know when the parent process has finished talking
to the permission manager. It would be nice to enable consumers to depend
on the status of the asynchronous task using a promise.
Blake Kaplan
Ehsan Akhgari
Jed Davis
Andrea Marchesini
Sylvestre Ledru
Andreas Farre
Andreea Pavel
Lina Cambridge
Gabriele Svelto
Haik Aftandilian
Ryan Hunt
Makoto Kato
Narcis Beleuzu
Brian Hackett
Tiberius Oros
Cosmin Sabou
Kartikaya Gupta
Bogdan Tara
Coroiu Cristina