This allows loads to be tracked as they are ongoing on a per-context basis in
the parent process, and for events to be generated for each subframe as it is
destroyed.
This patch also stops sending the `IsLoadingDocument` flag on the request to
the main process and removes RemoteWebProgress, as they are no longer necessary
due to being tracked directly.
Finally this patch also adds some logging to BrowsingContextWebProgress
to make it easier to diagnose this type of issue in the future.
Differential Revision: https://phabricator.services.mozilla.com/D115706
This allows loads to be tracked as they are ongoing on a per-context basis in
the parent process, and for events to be generated for each subframe as it is
destroyed.
This patch also stops sending the `IsLoadingDocument` flag on the request to
the main process and removes RemoteWebProgress, as they are no longer necessary
due to being tracked directly.
Finally this patch also adds some logging to BrowsingContextWebProgress
to make it easier to diagnose this type of issue in the future.
Differential Revision: https://phabricator.services.mozilla.com/D115706
Use the newly added session storage data getter to access the session
storage in the parent and store it in session store without a round
trip to content processes.
Depends on D111433
Differential Revision: https://phabricator.services.mozilla.com/D111434
For inserting text from OS in special cases, e.g., when inserting 2 or more characters
per keydown or inserting text without key press, we use a set of composition events on
macOS, but the other browsers don't use composition events. Instead, they expose only
`beforeinput` event and `input` event. We should follow their behavior for web-compat
because `beforeinput` events for IME composition are never cancelable, but the
`beforeinput` events for the cases are cancelable of the other browsers.
Differential Revision: https://phabricator.services.mozilla.com/D114826
For inserting text from OS in special cases, e.g., when inserting 2 or more characters
per keydown or inserting text without key press, we use a set of composition events on
macOS, but the other browsers don't use composition events. Instead, they expose only
`beforeinput` event and `input` event. We should follow their behavior for web-compat
because `beforeinput` events for IME composition are never cancelable, but the
`beforeinput` events for the cases are cancelable of the other browsers.
Differential Revision: https://phabricator.services.mozilla.com/D114826
We were using the top BrowsingContextGroup id in this case, which is
obviously wrong. Also make the API take a BrowsingContext directly,
rather than passing outerwindowids around.
Differential Revision: https://phabricator.services.mozilla.com/D112413
By resolving the relevant promises, instead of crashing (and if we
didn't crash we'd leave the window registered as a refresh driver
observer, which would be bad).
I wanted to reject them, since that's what we do when the page has no
pres shell, but that'd make this test fail:
https://searchfox.org/mozilla-central/rev/d8194cbbeaec11962ed67f83aea9984bf38f7c63/dom/base/test/browser_promiseDocumentFlushed.js#165-186
For this, we modify the OneShotPostRefreshObserver API to be more
generic (and rename it OneShotManagedRefreshObserver).
We fix APZ's usage of this API, which was doing something extremely
weird (returning a refcounted object in a UniquePtr). This seems like an
artifact from recent OneShotPostRefreshObserver cleanup.
Differential Revision: https://phabricator.services.mozilla.com/D111851
If we double tap on an element that is narrower than the viewport at maximum we can get into a situation where we zoom on part (say the bottom half) of that element but we could easily fit the entire element.
This happens because the code that calculate the rect to zoom to (CalculateRectToZoomTo) doesn't know about the maximum zoom. So it proceeds as though we will fit the width of the element. Under that assumption we will have to cut off part of the element vertically, so the code centers the rect on the users tap point.
This ends up cutting off part of the element vertically when it is clear that the whole element can fit on screen, which is a pretty ugly result.
This is not my favourite patch. This seemed to make the most sense. Another option I considered was passing the tap point through to AsyncPanZoomController::ZoomToRect but I think this approach came out better: the calculation all happens in one place at one time.
Differential Revision: https://phabricator.services.mozilla.com/D110538
This patch contains a large number of changes around the process switching
mechanism in order to avoid issues which are caused by a mismatched
understanding of the state of the process switch between processes in the
presence of nested event loops.
This includes:
1. The "InFlightProcessId" value is no longer recorded. All remaining uses
were removed in part 1, and the new mechanism tracks this information in
a better way.
2. The current BrowserParent instance is now tracked on
CanonicalBrowsingContext, meaning that logic which needs to work with this
information can now access it without depending on the current
WindowGlobalParent instance.
3. When doing a process switch, the previous host process for the
BrowsingContext is tracked until the process switch is completed, allowing
for future attempts to switch into that process to be delayed until the
previous unload event has finished running.
4. The process switch logic was refactored to simplify some of the
error-handling logic, and share more code between different cases.
Differential Revision: https://phabricator.services.mozilla.com/D110002
This patch contains a large number of changes around the process switching
mechanism in order to avoid issues which are caused by a mismatched
understanding of the state of the process switch between processes in the
presence of nested event loops.
This includes:
1. The "InFlightProcessId" value is no longer recorded. All remaining uses
were removed in part 1, and the new mechanism tracks this information in
a better way.
2. The current BrowserParent instance is now tracked on
CanonicalBrowsingContext, meaning that logic which needs to work with this
information can now access it without depending on the current
WindowGlobalParent instance.
3. When doing a process switch, the previous host process for the
BrowsingContext is tracked until the process switch is completed, allowing
for future attempts to switch into that process to be delayed until the
previous unload event has finished running.
4. The process switch logic was refactored to simplify some of the
error-handling logic, and share more code between different cases.
Differential Revision: https://phabricator.services.mozilla.com/D110002
Instead of collecting data from the entire tree of documents, we
collect data per document. The collected data is sent to the
corresponding parent window context and is applied incrementally to
the tab state cache.
Differential Revision: https://phabricator.services.mozilla.com/D107814
Instead of collecting data from the entire tree of documents, we
collect data per document. The collected data is sent to the
corresponding parent window context and is applied incrementally to
the tab state cache.
Differential Revision: https://phabricator.services.mozilla.com/D107814
Note that this patch only transforms the use of the nsDataHashtable type alias
to a directly equivalent use of nsTHashMap. It does not change the specification
of the hash key type to make use of the key class deduction that nsTHashMap
allows for in some cases. That can be done in a separate step, but requires more
attention.
Differential Revision: https://phabricator.services.mozilla.com/D106008
Previously, we would only send web progress events from the toplevel
BrowserParent, as other frames would never have the browser-child.js framescript
loaded in them, and so would never start sending events. This change moves the
decision to begin sending events into BrowserChild itself around the same time
as it would've happened previously with the framescript.
This new callsite should still avoid sending events for the creation of the
initial about:blank document in the BrowserChild, while not skipping any other
events, as before.
Differential Revision: https://phabricator.services.mozilla.com/D105558
This change allows associating individual web progress events with which frame
they originate from, rather than only recording the `isToplevel` information as
we were before, which is necessary in order to use the OnLocationChange events
from content to track the current URI on CanonicalBrowsingContext.
Due to events in ContentChild being filtered through nsBrowserStatusFilter, some
of the callbacks will never be passed nsIRequest or nsIWebProgress pointers, and
this patch also simplifies them by removing information which is not necessary
from the IPC message.
Finally, this patch adds a number of checks to the relevant Recv callbacks in
the parent process which help ensure that it does not accept web progress events
from a content process which is no longer hosting the target BrowsingContext.
This may allow for us to stop manually suspending web progress events on process
switches, as these checks will handle this automatically based on the existing
BrowsingContext and WindowContext objects.
Differential Revision: https://phabricator.services.mozilla.com/D105556
Inversed logic has been proven to be more difficult to read,
so use the simple positive variant.
Also add a simple sanity check for `WAYLAND_DISPLAY` so if people
set `MOZ_ENABLE_WAYLAND` in a X11 session don't get undesired behavior.
While on it, change a check for `XDG_SESSION_TYPE` to also use
`WAYLAND_DISPLAY`, improving behavior when launching FF from a TTY
or a TTY-launched session (e.g. via `weston-launch`).
`WAYLAND_DISPLAY` and `DISPLAY` are not expected to be set if
no Wayland or X11 server is available, so using them makes us behave
more predictable.
Differential Revision: https://phabricator.services.mozilla.com/D106726
If we don't do this, then just moving the mouse over a window experiencing a
slow script will cause it to show the notification. We could try to
deserialize the message inside nsContentUtils::IsMessageCriticalInputEvent, but
that seems overcomplicated compared to just adding a new message which proxies
to the original message handlers.
Differential Revision: https://phabricator.services.mozilla.com/D106016
This makes the naming more consistent with other functions called
Insert and/or Update. Also, it removes the ambiguity whether
Put expects that an entry already exists or not, in particular because
it differed from nsTHashtable::PutEntry in that regard.
Differential Revision: https://phabricator.services.mozilla.com/D105473
If we don't do this, then just moving the mouse over a window experiencing a
slow script will cause it to show the notification. We could try to
deserialize the message inside nsContentUtils::IsMessageCriticalInputEvent, but
that seems overcomplicated compared to just adding a new message which proxies
to the original message handlers.
Differential Revision: https://phabricator.services.mozilla.com/D106016
If we don't do this, then just moving the mouse over a window experiencing a
slow script will cause it to show the notification. We could try to
deserialize the message inside nsContentUtils::IsMessageCriticalInputEvent, but
that seems overcomplicated compared to just adding a new message which proxies
to the original message handlers.
Differential Revision: https://phabricator.services.mozilla.com/D106016
It should be called "Get" rather than "Lookup" because it returns
UserDataType. "Add" is called "Insert" in the other methods.
Differential Revision: https://phabricator.services.mozilla.com/D105470
There are no code changes, only #include changes.
It was a fairly mechanical process: Search for all "AUTO_PROFILER_LABEL", and in each file, if only labels are used, convert "GeckoProfiler.h" into "ProfilerLabels.h" (or just add that last one where needed).
In some files, there were also some marker calls but no other profiler-related calls, in these cases "GeckoProfiler.h" was replaced with both "ProfilerLabels.h" and "ProfilerMarkers.h", which still helps in reducing the use of the all-encompassing "GeckoProfiler.h".
Differential Revision: https://phabricator.services.mozilla.com/D104588
Note that there's still a little plugin related code in
widget/ and gfx/ etc after this. That can be removed
once we remove plugin support from dom/ etc.
The removal from layout/ should be pretty complete though.
Differential Revision: https://phabricator.services.mozilla.com/D102140
I flagged this as sketchy before (though it was trying to preserve
existing behavior).
However now that that state propagates to the parent process and races
with the state that the parent process reads, it started causing
correctness issues.
Just remove this line, it shouldn't be needed. I'm not sure how to write
a test for this, unfortunately :(
Differential Revision: https://phabricator.services.mozilla.com/D100971
This fixes a bunch of regressions:
- a wrong calculation in `GetIdleDeadlineHint()`, leading to pageload
regressions.
- in certain situations we'd use `StartupRefreshDriverTimer` instead
of `VsyncRefreshDriverTimer` when initializing timers early
- unnecessary use of `BrowserChild` on backends that don't opt for
per-browser-child vsync - i.e. all but Wayland.
This is partly done by reverting to pre-1645528 behaviour, although
with some code simplifications.
FTR: I also played with some more radical changes, but given the
complexity of the code involved I found the regression potential too
big. Thus this is the most conservative solution I could come up with.
Differential Revision: https://phabricator.services.mozilla.com/D100471
Csoregi Natalia
Nika Layzell
Butkovits Atila
Alexandru Michis
Andreas Farre
Masayuki Nakano
Edgar Chen
Valentin Gosu
Emilio Cobos Álvarez
Timothy Nikkel
David Parks
Neil Deakin
Andreea Pavel
Bogdan Tara
Simon Giesecke
Kagami Sascha Rosylight
Olli Pettay
Robert Mader
Doug Thayer
Brindusan Cristian
Gerald Squelart
smolnar
Mats Palmgren