The meaning of "possibly-changed" is provided by the big comment above
MustSendToContentProcesses.
On a new profile this reduces the number of prefs sent like so:
- Command-line: 222 --> 3
- IPC: 3129 --> 130
On an older profile:
- Command-line: 222 --> 3
- IPC: 3165 --> 180
MozReview-Commit-ID: DcgedhXhZd8
--HG--
extra : rebase_source : acef424fab5031347cbcbd5c3e6a24ee66895ef9
When ContentChild::RecvInitRendering is received, it tries to setup the
IPDL actors related to rendering. If the GPU process crashes before or
during this process, it will fail, and cause the content process to
crash as well. This is unnecessary because the UI process will either
restart the GPU process, or subsume its job into itself, and trigger
ContentChild::RecvReinitRendering. It is a similar case for failures in
ContentChild::RecvReinitRendering.
Since the GPU process crashing should be a recoverable scenario, we now
check if the remote IPDL actor is in the UI or the GPU process. If it is
in the UI process, it will fail/crash as it does today. If it is in the
GPU process, it will wait for the next
ContentChild::RecvReinitRendering.
For failures that are not IPDL related (e.g. failed to get some resource
like spawning a thread), we release assert specifically for those
failures. They are not recoverable.
With this change, the macOS content sandbox has no ability to create files
anywhere on disk (in release builds). If the content process needs a file to
write to, it needs to obtain a file descriptor from the parent process.
MozReview-Commit-ID: 7LoG1PW0UDR
--HG--
extra : rebase_source : 4ac0a7f187d45c9b6c0f8a658edfdae0509054ac
These conditions are rare and do indicate a problem which breaks accessibility.
However, we aren't getting any closer to diagnosing these as a result of these crashes, so they cause user pain without any gain to us.
MozReview-Commit-ID: D9U4et3Bg7d
--HG--
extra : rebase_source : a81263a0ef97a8ed87129d15ef30ded3005e740c
These conditions are rare and do indicate a problem which breaks accessibility.
However, we aren't getting any closer to diagnosing these as a result of these crashes, so they cause user pain without any gain to us.
MozReview-Commit-ID: D9U4et3Bg7d
--HG--
extra : rebase_source : a81263a0ef97a8ed87129d15ef30ded3005e740c
nsIDOMWindowUtils::sendKeyEvent() is already replaced with nsITextInputProcessor
for making callers set any attributes of KeyboardEvent and guaranteeing
consistency behavior with keyboard events caused by native key events. E.g.,
whether keypress event should be dispatched or not is automatically decided.
nsIFrameLoader::sendCrossProcessKeyEvent() is similart to
nsIDOMWindowUtils::sendKeyEvent() but it dispatches keyboard events in
child process directly. Currently, nsITextInputProcessor doesn't have this
feature but nobody wants/uses this feature. So, for removing actual
implementation of nsIDOMWindowUtils::sendKeyEvent(), i.e.,
nsContentUtils::SendKeyEvent(), which is shared by both
nsDOMWindowUtils::SendKeyEvent() and nsFrameLoader::SendCrossProcessKeyEvent(),
we should remove this unused API too. (FYI: it's implemented for old Fennec,
by bug 553149.)
MozReview-Commit-ID: 9n0UVo8Me8k
--HG--
extra : rebase_source : e9b117f5b9afec76e63d57ab8cd86dafb5873789
This patch was autogenerated by my decomponents.py
It covers almost every file with the extension js, jsm, html, py,
xhtml, or xul.
It removes blank lines after removed lines, when the removed lines are
preceded by either blank lines or the start of a new block. The "start
of a new block" is defined fairly hackily: either the line starts with
//, ends with */, ends with {, <![CDATA[, """ or '''. The first two
cover comments, the third one covers JS, the fourth covers JS embedded
in XUL, and the final two cover JS embedded in Python. This also
applies if the removed line was the first line of the file.
It covers the pattern matching cases like "var {classes: Cc,
interfaces: Ci, utils: Cu, results: Cr} = Components;". It'll remove
the entire thing if they are all either Ci, Cr, Cc or Cu, or it will
remove the appropriate ones and leave the residue behind. If there's
only one behind, then it will turn it into a normal, non-pattern
matching variable definition. (For instance, "const { classes: Cc,
Constructor: CC, interfaces: Ci, utils: Cu } = Components" becomes
"const CC = Components.Constructor".)
MozReview-Commit-ID: DeSHcClQ7cG
--HG--
extra : rebase_source : d9c41878036c1ef7766ef5e91a7005025bc1d72b
Modifiers of shortcut keys may be same as modifier of content access keys.
When focus is in the main process, such eKeyPress event is sent to remote
content first. Then, and if it's not handled in the remote content,
eAccessKeyNotFound is dispatched into the DOM tree in the main process.
However, nsXBLWindowKeyHandler doesn't handle it as eKeyPress event. So,
it causes that shortcut keys whose modifier conflicts with content access key
won't be handled.
This patch just makes nsXBLWindowKeyHandler treat eAccessKeyNotFound as
eKeyPress event even though other shortcut keys which are handled by JS
won't be executed. Perhaps, we should stop using eAccessKeyNotFound but
it's too risky change for now.
MozReview-Commit-ID: IJltg5gwBc5
--HG--
extra : rebase_source : f456eade18cd4fefd2eab6e06f5b00156ac8ad59
This was done using the following script:
37e3803c7a/processors/chromeutils-import.jsm
MozReview-Commit-ID: 1Nc3XDu0wGl
--HG--
extra : source : 12fc4dee861c812fd2bd032c63ef17af61800c70
extra : intermediate-source : 34c999fa006bffe8705cf50c54708aa21a962e62
extra : histedit_source : b2be2c5e5d226e6c347312456a6ae339c1e634b0
This was done using the following script:
37e3803c7a/processors/chromeutils-import.jsm
MozReview-Commit-ID: 1Nc3XDu0wGl
--HG--
extra : source : 12fc4dee861c812fd2bd032c63ef17af61800c70
This was done using the following script:
37e3803c7a/processors/chromeutils-import.jsm
MozReview-Commit-ID: 1Nc3XDu0wGl
--HG--
extra : rebase_source : c004a023389f1f6bf3d2f3efe93c13d423b23ccd
dom/time contained the TimeService and TimeManager classes, used for
setting time via Gecko on FirefoxOS. Since FirefoxOS is no longer in
the code base, the directory can be removed.
MozReview-Commit-ID: 8PEk3e6HA67
The end goal is to allow the seccomp-bpf policy to vary based on the
content sandbox level.
Rather than add yet another parameter to SetContentProcessSandbox to
pass down the sandbox level, this collects the values that have to be
computed in libxul into a struct, and moves the code that computes it so
it's not cluttering up ContentChild.
MozReview-Commit-ID: L0dyQwHQKhc
Content processes can contain ghost windows, so the debug-only ghost
window unlinker needs to send a message to child processes to get them
to run it, too.
MozReview-Commit-ID: 9Ffc3SDNDJB
--HG--
extra : rebase_source : 875891e9332cf41c4157d246b71c2c361cab4aa6
Most of the Shadow DOM related code are behind "dom.webcomponents.enabled" and
this pref is only used by Shadow DOM right now, so we should rename it to
"dom.webcomponents.shadowdom.enabled"
MozReview-Commit-ID: er1c7AsSSW
This pref does not override privacy.resistFingerprinting, but when it is set (and
privacy.resistFingerprinting is not) we will still adjust the precision of almost
all timers. The adjustment amount is the second pref, which is defaulted to
20us but now dynamically adjustable (in the scale of microseconds.)
This patch does _not_ address the performance API, which privacy.resistFingerprinting
disables.
We are landing this preffed on at the current value we clamp performance.now() at
which is 20us.
MozReview-Commit-ID: ESZlSvH9w1D
--HG--
extra : rebase_source : a8afead1bdba958c6c7b383b2216dacb3a1b135c
Calling GetPresShell() might create a content viewer, which might cause
script to run. This is bad if there's a ForcePaint message queued up,
because it could mean running the force painting code while we're still
in the midst of making a tab hidden, which would put us in an inconsistent
state.
MozReview-Commit-ID: 3rw2wGllGdk
--HG--
extra : rebase_source : 1c4b1ad28467fc6960eaa7cb744e47dd191b30f5
extra : source : 78073667ddc6e932408f49076b74c448a74bb710
This assumption also mirrors how non-remote browsers have their docShells active by default.
In order to do this, I also have to increase the initial epoch's for the TabParent and TabChild,
as if a RenderLayers has been called. Originally, since the epochs initted at 0, and the epochs
stored by the [Layer|WebRender]TransactionParent were also initted at 0, we'd hit this branch:
https://searchfox.org/mozilla-central/rev/c633ffa4c4611f202ca11270dcddb7b29edddff8/gfx/layers/ipc/LayerTransactionParent.cpp#703
and then we'd never alert the TabParent about the layer upload.
MozReview-Commit-ID: 6PP1eCnisYK
--HG--
extra : rebase_source : 4ca9f692635e3a3a1ec8e626d0db24f2e512ec9a
extra : source : b9b2895b11a32f3da0f4c8fe364bf3bdfc7defb6
This is necessary because sometimes the async tab switcher will instantiate when
there already exists some background tabs that are rendering via print preview. When
that happens, it's important for the state to be set correctly for them so that we
don't accidentally treat them as still loading, and wait (forever) for them to report
having finished loading.
MozReview-Commit-ID: 2dwo5WlXlgJ
--HG--
extra : rebase_source : 4b1e2fb945f1dc4591c56f836a1caa9adeea9eb4
extra : source : 53bb2bc2b67673572dafc3093280fa72973b3d32
Originally, setting the active state on the DocShell for remote browsers also did the
work of making the TabChild render its layers and upload them to the compositor.
This patch adds a new renderLayers method to nsITabParent which allows more fine-grained
control - we can now, for example, cause layers to be rendered and uploaded without
activating the DocShell.
Note that if one activates or deactivates the DocShell, we'll still do the work of
attempting to render / clear the layers if it hasn't already been done.
MozReview-Commit-ID: KkLaMDTzfHi
--HG--
extra : rebase_source : 6c53729ae4b07c67d9e3de99c12e1165a85ffaac
extra : source : 261be8ec05542a9e50c0ca03a59a8a44997269b7
This patch requires that each instance of IPC's RunnableFunction is
passed in a name, like the non-IPC RunnableFunction.
MozReview-Commit-ID: Atu1W3Rl66S
--HG--
extra : rebase_source : f932d7597a26a3f0c4246b3a95df638860d3d32d
New content processes get prefs in three ways.
- They read them from greprefs.js, prefs.js and other such files.
- They get sent "early prefs" from the parent process via the command line
(-intPrefs/-boolPrefs/-stringPrefs).
- They get sent "late prefs" from the parent process via IPC message.
(The latter two are necessary for communicating prefs that have been added or
modified in the parent process since the file reading occurred at startup.)
We have some machinery that detects if a late pref is accessed before the late
prefs are set, which is good. But it has a big exception in it; late pref
accesses that occur early via Add*VarCache() and RegisterCallbackAndCall() are
allowed.
This exception was added in bug 1341414. The description of that bug says "We
should change AddBoolVarCache so that it doesn't look at the pref in the
content process until prefs have been received from the parent." Unfortunately,
the patch in that bug added the exception to the checking without changing
Add*VarCache() in the suggested way!
This means it's possible for late prefs to be read early via VarCaches (or
RegisterCallbackAndCall()) when their values are incorrect, which is bad.
Changing Add*VarCache() to delay the reading as bug 1341414 originally
suggested seems difficult. A simpler fix is to just remove the exception in the
checking and extend the early prefs list as necessary. This patch does that,
lengthening the early prefs list from ~210 to ~300. Fortunately, most (all?) of
the added prefs are ints or bools rather than strings, so it doesn't increase
the size of the command line arguments for content processes by too much.
--HG--
extra : rebase_source : 5ea5876c206401d23a368ef9cb5040522c9ca377
We implement PointerEvent.getCoalescedEvents as
1. Clone the widget events we coalesced.
2. Convert them to dom::PointerEvent when user calls getCoalescedEvents.
MozReview-Commit-ID: 8IKw4PbUsDD
During history import, sending NotifyVisited messages from the
chrome process to the content processes in order to change link
colors can take a significant portion of the parent process's
main thread time. Batching it seems to have very significant
results on jank time during history imports.
MozReview-Commit-ID: BHAXpIMa7ly
--HG--
extra : rebase_source : f43c653e6945d7775cc9dd7bca4c1e84099c2673
Right now the only parameter will be sent via the IPC message is form URI.
IPC is triggered when a password field is focusd (See P2.)
MozReview-Commit-ID: J8lVwRhTFIr
--HG--
extra : rebase_source : b948cf1a719c9a06100c54f3eda526ea6f7cf848