It's easy to mess up the scoping so that (a) the label is pushed and then
immediately popped, and/or (b) the string doesn't live long enough. It's also
easy to do a utf16-to-utf8 conversion unnecessarily when the profiler is
inactive. This patch splits that macro into three new ones that are harder to
mess up.
- AUTO_PROFILER_LABEL_DYNAMIC_CSTR: same as current.
- AUTO_PROFILER_LABEL_DYNAMIC_NSCSTRING: for nsCStrings.
- AUTO_PROFILER_LABEL_DYNAMIC_LOSSY_NSSTRING: for nsStrings.
--HG--
extra : rebase_source : 3e2bbec4737b696e1c86579ae54be4cb3186c100
2017-10-13 16:12:57 +11:00
Matt Woodrow ext:(%2C%20Miko%20Mynttinen%20%3Cmikokm%40gmail.com%3E%2C%20Timothy%20Nikkel%20%3Ctnikkel%40gmail.com%3E)
After the fix to bug 1294442 and bug 1324499, ResizeReflow began to be called
twice for each DOM update in webext popups, and we also artificially re-set the
scroll outside of ResizeReflow to counter the DidDoReflow callback in
nsHTMLScrollFrame setting scrolltop to zero due to the first reflow, which is
done with unconstrained height.
Because of the scrollport being reset we get spurious DOM scroll events.
Replacing the scrollport also interrupts smooth scrolling.
Move the double-reflow down one level into PresShell, doing it before
DidDoReflow is called. The scrollport is no longer reset (causing a spurious
scroll event), and we don't need to replace it (interrupting smooth scrolling).
Also partially fixes bug 1396034.
MozReview-Commit-ID: HzYITyH4UeW
--HG--
extra : rebase_source : 567056300bc81c9e4c197783f48636caf67cde34
extra : intermediate-source : ef7322dfd99d79a403bb0804638a46c70ece1b45
extra : source : 4f1761bb955473026b4deba76a5e76e93b7ede35
Push pointer event pref before testing test_assign_event_data.html and remove redundant assertions since the pushed pref will be restore after testing.
MozReview-Commit-ID: 22nTMLGoMFj
It's easy to mess up the scoping so that (a) the label is pushed and then
immediately popped, and/or (b) the string doesn't live long enough. It's also
easy to do a utf16-to-utf8 conversion unnecessarily when the profiler is
inactive.
This patch splits that macro into three new ones that are harder to mess up.
- AUTO_PROFILER_LABEL_DYNAMIC_CSTR: same as current.
- AUTO_PROFILER_LABEL_DYNAMIC_NSCSTRING: for nsCStrings.
- AUTO_PROFILER_LABEL_DYNAMIC_LOSSY_NSSTRING: for nsStrings.
--HG--
extra : rebase_source : 53c8b43b6a1be06d00618a133e28bf95c46a3ba3
It's easy to mess up the scoping so that (a) the label is pushed and then
immediately popped, and/or (b) the string doesn't live long enough. It's also
easy to do a utf16-to-utf8 conversion unnecessarily when the profiler is
inactive.
This patch splits that macro into three new ones that are harder to mess up.
- AUTO_PROFILER_LABEL_DYNAMIC_CSTR: same as current.
- AUTO_PROFILER_LABEL_DYNAMIC_NSCSTRING: for nsCStrings.
- AUTO_PROFILER_LABEL_DYNAMIC_LOSSY_NSSTRING: for nsStrings.
--HG--
extra : rebase_source : 59f77df0124249bfd11fee3585420a17b4201d37
Currently the Gecko Profiler defines a moderate amount of stuff when
MOZ_GECKO_PROFILER is undefined. It also #includes various headers, including
JS ones. This is making it difficult to separate Gecko's media stack for
inclusion in Servo.
This patch greatly simplifies how things are exposed. The starting point is:
- GeckoProfiler.h can be #included unconditionally;
- everything else from the profiler must be guarded by MOZ_GECKO_PROFILER.
In practice this introduces way too many #ifdefs, so the patch loosens it by
adding no-op macros for a number of the most common operations.
The net result is that #ifdefs and macros are used a bit more, but almost
nothing is exposed in non-MOZ_GECKO_PROFILER builds (including
ProfilerMarkerPayload.h and GeckoProfiler.h), and understanding what is exposed
is much simpler than before.
Note also that in BHR, ThreadStackHelper is now entirely absent in
non-MOZ_GECKO_PROFILER builds.
(Path is actually r=froydnj.)
Bug 1400459 devirtualized nsIAtom so that it is no longer a subclass of
nsISupports. This means that nsAtom is now a better name for it than nsIAtom.
MozReview-Commit-ID: 91U22X2NydP
--HG--
rename : xpcom/ds/nsIAtom.h => xpcom/ds/nsAtom.h
extra : rebase_source : ac3e904a21b8b48e74534fff964f1623ee937c67
By returning true from WebRenderLayerManager::EndEmptyTransaction, we
avoid doing a full paint in cases where the caller decides an empty
transaction would be sufficient. WebRenderLayerManager already rejects
attempts to set some forms of empty-transaction data (specifically
transform and scroll offset updates). This means that we will never get
a call to EndEmptyTransaction where the caller is expecting a transform
or scroll offset update to be sent over to the compositor. So if we have an
implementation of EndEmptyTransaction that ignores that data, it will not
break expectations.
There is still one piece of information that WebRenderLayerManager
doesn't reject in this manner, the APZ focus state. That is, if the
layout code sets a pending APZ focus state on the WRLM, followed by a
all to EndEmptyTransaction, it expects the focus state to get propagated
to the compositor. This patch makes sure that it does happen by using
the new API added in the previous patch.
MozReview-Commit-ID: 596UgW9ZWAF
--HG--
extra : rebase_source : e0f4f201a76747d6e29cde5da26fe760fd7f770b
Revise the pointer event implementation, includes
- Revise the implementation to fetch preference values.
- Separate the logic to get the pointer capturing frame to PointerEventHandler.
MozReview-Commit-ID: 7pdAr0XFNT2
This isn't a super essential feature, and is just a change to try to bring us in
line with chromium and the spec. As this has apparent web compat issues, and
DataTransfer is a hard to test area, this patch moves the changes behind a pref,
which we can come back to turning on after we ship 57.
This ensures that, if the touch event is over a scrollbar thumb and
makes it into nsSliderFrame::StartAPZDrag(), nsSliderFrame knows
that the event went through APZ and that therefore APZ will handle
the drag.
MozReview-Commit-ID: 92wAc1l9Pqz
--HG--
extra : rebase_source : 94ac60bc8b38dad3d8abaa39b5a94de88ec0f6b0
This is just simplifying things by eliminating a layer of now-unnecessary
abstraction.
MozReview-Commit-ID: 1MBagjYXbCD
--HG--
extra : rebase_source : a0e7e78ef4925efd00eac969ae126d1ba3f4f607
These functions are now dead code. And really, they were already dead code
even before this patch-stack -- they were only called in never-evaluated
(and now-deleted) #else clauses.
MozReview-Commit-ID: AeY6Z3ybmsv
--HG--
extra : rebase_source : deb51373f4d32580ab2ffb71c5c078132e4a76a6
With previous change, KeyboardEvent is dispatched even when invisible window
has focus. However, nsRootWindow::GetControllerForCommand() returns controller
for focused window even when the window is invisible because it uses
nsFocusManager::GetFocusedDescendant() to retrieve focused window.
Perhaps, we can assume that users won't expect to do something with invisible
window when they type some keys. Then, nsRootWindow::GetControllerForCommand()
should return controller for visible ancestor window if focused window is
invisible.
This patch makes nsFocusManager::GetFocusedDescendant() can return only visible
descendants. However, it already has a bool argument. Therefore, it should
have a flag instead of adding new flag. Most changes of this patch is replacing
its callers.
Then, nsRootWindow::GetControllerForCommand() and nsRootWindow::GetControllers()
should have a bool flag if it should return controller(s) for visible window.
This patch adds a bool flag for it. Fortunately, the interface isn't scriptable.
Finally, this patch makes nsXBLPrototypeHandler::DispatchXBLCommand() and
EventStateManager::DoContentCommandEvent() retrieve controller for visible
window since they are always handles user input.
MozReview-Commit-ID: GygttTHuKRm
--HG--
extra : rebase_source : 1341273c4606298cb9b890b9312d9f5c8a75d144
When focused document is invisible, PresShell::HandleEvent() won't dispatch
any events. However, if it's KeyboardEvent, user cannot do anything only with
keyboard. So, in this case, PresShell should retarget keyboard events to
first visible ancestor.
MozReview-Commit-ID: CvCBcYwvfbe
--HG--
extra : rebase_source : f7a59be80137449bf4c7486b069ebb3057f46b05
When there is no selection and no focused content, aContent of
nsIPresShell::GetScrollableFrameToScrollForContent() is nullptr. Then, it'll
use root scrollable frame to scroll. However, the root scrollable frame may
be not scrollable. In such case, ancestor document's scrollable frame should
be scrolled instead.
MozReview-Commit-ID: 1kOPtDX6G5L
--HG--
extra : rebase_source : 4b2c48bb187e40dbddd2760b8ddd8bf4abf5b178
We currently use the aFlags argument of ContentRemoved for two purposes:
(1) To determine when a frame is being removed due to its element being removed
from the DOM, so we reframe its now-possibly-no-longer-suppressed
whitespace siblings as needed.
In other cases, our ContentRemoved call will be followed by a
ContentInserted call, which will end up doing AddTextItemIfNeeded() to
generate the relevant textframes if they're necessary.
Since we only need to tell apart the "DOM removal" and "anything else"
cases, we don't need to thread the aFlags argument through all the ways
ContentRemoved can call itself (on an ancestor).
All those cases should just be treated as "not DOM removal". In particular,
even if the original call _was_ for a DOM removal, if we convert it to an
ancestor reframe, which will call ContentInserted on the ancestor as well,
we don't need to do anything with text siblings of the ancestor.
(2) To save the frame tree state from DestroyFramesFor, but the frame tree
state is unconditionally captured on RecreateFramesForContent, so we only
need to care about it in the original ContentRemoved call.
Because of that, we can move that to the callsite, patch incoming for that.
MozReview-Commit-ID: Gy5IhUysBlz
Signed-off-by: Emilio Cobos Álvarez <emilio@crisal.io>
This patch splits up the existing "layout/style-sets" measurement into
"layout/gecko-style-sets", or "layout/servo-style-sets/stylist/rule-tree" and
"layout/servo-style-sets/other". (Additional things will be measured under
"layout/servo-style-sets/" later, such as cascade data.)
This requires introducing a new type, ServoStyleSetSizes, for transferring the
multiple measurements from Rust code to C++ code.
MozReview-Commit-ID: FbmzpsjBpgI
--HG--
extra : rebase_source : f2d1441705139e6674d355792255302fcd89f748
For parent process, users may expect the UI is interactable after they saw the
first tab has restored/shown.
So this patch added a new topic "sessionstore-one-or-no-tab-restored" which
represents the parent process has finished a tab restoring. If there is nothing
to restore, it is effectively equal to "sessionstore-windows-restored".
For centent processes, users may expect web content is interactable when the
top-level-content-document has finished loading, which is different from the
parent case.
MozReview-Commit-ID: AtEUW80Ea6n
--HG--
extra : rebase_source : d920975bf95545ea9e3127d3f570b814fe301be9
CLOSED TREE
--HG--
extra : amend_source : 7618f27e1cf71347f12384935374d303d5c1ab23
extra : histedit_source : 1143d226673ace563bb5d9ff3c420ded33439791%2C9746a30097304bd2214a0072773dc00c767d31cd
There's only one case of sync frame construction from ContentRemoved now, and
it's not on the element being removed, but on the whitespace siblings if needed,
and _only_ when they don't support lazy frame construction.
Basically, this switches all the RecreateFramesForContent calls to use
`aAsyncInsert` (which I changed to an enum class for readability), except when
we're already reframing.
Also, it switches ReframeTextIfNeeded to opt-in into lazy frame construction,
since it's used only when aFlags == CONTENT_REMOVED.
This allows to simplify the DestroyFramesFor API (which I'm happy to rename to
something more meaningful, since now it's something like
DestroyFramesForAndRecreateThemAsync), and do some other consistency cleanups.
A bunch of the ContentRemoved callsites were pretty random at passing
aAsyncInsert, and that was some kind of a mess. This patch ensures consistency,
and makes it impossible to do O(n^2) work when removing DOM nodes, which is
nice.
The underlying reason for this is explained in the description of bug 1377848,
and basically allows us to remove a bunch of Servo hacks on the longer term (a
few of them are going away already, yay!).
MozReview-Commit-ID: 2DrUTxGV8RX
--HG--
extra : rebase_source : f428d839a5482477dea22c0fea600d54f3e8799c
This patch changes UpdateAnimationOnlyStyles to only flush animation styles if
there are throttled animation styles that could affect hit-testing and renames
the function to UpdateAnimationStylesForHitTesting at the same time.
For GeckoRestyleManager, the original UpdateAnimationOnlyStyles which flushes
animation styles if there are any pending animation styles, is renamed to
UpdateAnimationStyles for consistency.
MozReview-Commit-ID: 89UleXjI2OE
--HG--
extra : rebase_source : cac59f9d48b096aee718f89ffe203d38d39027fd
This patch moves measurement of ComputedValues objects from Rust to C++.
Measurement now happens (a) via DOM elements and (b) remaining elements via
the frame tree. Likewise for the style structs hanging off ComputedValues
objects.
Here is an example of the output.
> ├──27,600,448 B (26.49%) -- active/window(https://en.wikipedia.org/wiki/Barack_Obama)
> │ ├──12,772,544 B (12.26%) -- layout
> │ │ ├───4,483,744 B (04.30%) -- frames
> │ │ │ ├──1,653,552 B (01.59%) ── nsInlineFrame
> │ │ │ ├──1,415,760 B (01.36%) ── nsTextFrame
> │ │ │ ├────431,376 B (00.41%) ── nsBlockFrame
> │ │ │ ├────340,560 B (00.33%) ── nsHTMLScrollFrame
> │ │ │ ├────302,544 B (00.29%) ── nsContinuingTextFrame
> │ │ │ ├────156,408 B (00.15%) ── nsBulletFrame
> │ │ │ ├─────73,024 B (00.07%) ── nsPlaceholderFrame
> │ │ │ ├─────27,656 B (00.03%) ── sundries
> │ │ │ ├─────23,520 B (00.02%) ── nsTableCellFrame
> │ │ │ ├─────16,704 B (00.02%) ── nsImageFrame
> │ │ │ ├─────15,488 B (00.01%) ── nsTableRowFrame
> │ │ │ ├─────13,776 B (00.01%) ── nsTableColFrame
> │ │ │ └─────13,376 B (00.01%) ── nsTableFrame
> │ │ ├───3,412,192 B (03.28%) -- servo-style-structs
> │ │ │ ├──1,288,224 B (01.24%) ── Display
> │ │ │ ├────742,400 B (00.71%) ── Position
> │ │ │ ├────308,736 B (00.30%) ── Font
> │ │ │ ├────226,512 B (00.22%) ── Background
> │ │ │ ├────218,304 B (00.21%) ── TextReset
> │ │ │ ├────214,896 B (00.21%) ── Text
> │ │ │ ├────130,560 B (00.13%) ── Border
> │ │ │ ├─────81,408 B (00.08%) ── UIReset
> │ │ │ ├─────61,440 B (00.06%) ── Padding
> │ │ │ ├─────38,176 B (00.04%) ── UserInterface
> │ │ │ ├─────29,232 B (00.03%) ── Margin
> │ │ │ ├─────21,824 B (00.02%) ── sundries
> │ │ │ ├─────20,080 B (00.02%) ── Color
> │ │ │ ├─────20,080 B (00.02%) ── Column
> │ │ │ └─────10,320 B (00.01%) ── Effects
> │ │ ├───2,227,680 B (02.14%) -- computed-values
> │ │ │ ├──1,182,928 B (01.14%) ── non-dom
> │ │ │ └──1,044,752 B (01.00%) ── dom
> │ │ ├───1,500,016 B (01.44%) ── text-runs
> │ │ ├─────492,640 B (00.47%) ── line-boxes
> │ │ ├─────326,688 B (00.31%) ── frame-properties
> │ │ ├─────301,760 B (00.29%) ── pres-shell
> │ │ ├──────27,648 B (00.03%) ── pres-contexts
> │ │ └─────────176 B (00.00%) ── style-sets
The 'servo-style-structs' and 'computed-values' sub-trees are new. (Prior to
this patch, ComputedValues under DOM elements were tallied under the the
'dom/element-nodes' sub-tree, and ComputedValues not under DOM element were
ignored.) 'servo-style-structs/sundries' aggregates all the style structs that
are smaller than 8 KiB.
Other notable things done by the patch are as follows.
- It significantly changes the signatures of the methods measuring nsINode and
its subclasses, in order to handle the tallying of style structs separately
from element-nodes. Likewise for nsIFrame.
- It renames the 'layout/style-structs' sub-tree as
'layout/gecko-style-structs', to clearly distinguish it from the new
'layout/servo-style-structs' sub-tree.
- It adds some FFI functions to access various Rust-side data structures from
C++ code.
- There is a nasty hack used twice to measure Arcs, by stepping backwards from
an interior pointer to a base pointer. It works, but I want to replace it
with something better eventually. The "XXX WARNING" comments have details.
- It makes DMD print a line to the console if it sees a pointer it doesn't
recognise. This is useful for detecting when we are measuring an interior
pointer instead of a base pointer, which is bad but easy to do when Arcs are
involved.
- It removes the Rust code for measuring CVs, because it's now all done on the
C++ side.
MozReview-Commit-ID: BKebACLKtCi
--HG--
extra : rebase_source : 4d9a8c6b198a0ff025b811759a6bfa9f33a260ba
This patch does the following.
- Moves nsWindowSizes from nsWindowMemoryReporter.h to its own file,
nsWindowSizes.h, so it can be included more widely without exposing
nsWindowMemoryReporter.
- Merges nsArenaMemoryStats.h (which defines nsTabSizes and nsArenaMemoryStats)
into nsWindowSizes.h.
- Renames nsArenaMemoryStats as nsArenaSizes, and nsWindowSizes::mArenaStats as
nsWindowSizes::mArenaSizes. This is the more usual naming scheme for such
types.
- Renames FRAME_ID_STAT_FIELD as NS_ARENA_SIZES_FIELD.
- Passes nsWindowSizes to PresShell::AddSizeOfIncludingThis() and
nsPresArena::AddSizeOfExcludingThis(), instead of a bunch of smaller things.
One nice consequence is that the odd nsArenaMemoryStats::mOther field is no
longer necessary, because we can update nsWindowSizes::mLayoutPresShellSize
directly in nsPresArena::AddSizeOfExcludingThis().
- Adds |const| to a few methods.
MozReview-Commit-ID: EpgFWKFqy7Y
On macOS, we fall back eKeyPress event to native menu. Therefore, widget always requests a reply from remote process because it's difficult to check if the eKeyPress event will be sent to a remote process actually. If it's not sent to any remote processes, PresShell needs to dispatch the event into the DOM tree. Additionally, even if it's marked as "waiting reply from remote process", it needs to dispatch the DOM event in the main process first because we need to check if the key combination is reserved by chrome (if it's reserved, the eKeyPress event shouldn't be fired in the remote process).
Therefore, this patch makes EventStateManager::PreHandleEvent() resets the state when focused content isn't in any remote processes and the event's propagation hasn't been stopped.
Additionally, this patch makes PresShell::HandleEventInternal() checks WidgetEvent::PropgationStopped() with WidgetEvent::IsWaitingReplyFromRemoteProcess() before dispatching the event into the DOM tree.
MozReview-Commit-ID: FmgL3rCuQ8y
--HG--
extra : rebase_source : aa8d6b924fc78d1d9dd35a35c92976c35c758657
Currently, access key is handled in EventStateManager::PreHandleEvent() with eKeyPress event, i.e., before dispatching it into the DOM tree, if the access key is registered in EventStateManager. So, the main process does not check if the preceding eKeyDown event is consumed in focused remote process.
When preceding eKeyDown event is consumed in the main process, eKeyPress event won't be dispatched by widget. However, if remote process has focus, it's impossible widget to stop dispatching eKeyPress event because preceding eKeyDown event hasn't been handled in the focused remote process yet. Therefore, main process needs to post eKeyPress event to check if preceding eKeyDown event was consumed. When eKeyPress event is marked as "waiting reply from remote process", TabChild sends it back to the main process only when preceding eKeyDown event wasn't consumed. So, only when eKeyPress event is back to the main process, main process should handle accesskey with it.
This patch makes EventStateManager::PreHandleEvent() check if a remote target has focus before handling accesskey. If a remote process has accesskey and there is an accesskey matching with eKeyPress event, it marks the event as "waiting reply from remote content" and stop propagation in the process.
Finally, when eKeyPress event is sent back to TabParent, TabParent::RecvReplyKeyEvent() calls EventStateManager::HandleAccessKey() before dispatching the reply event into the DOM tree.
MozReview-Commit-ID: KsOkakaIVzb
--HG--
extra : rebase_source : 7e0c6966a1bde085e34d45bca4b0166b9fc2f3f1
DOM Standard defines that offset of Range is unsigned long. However, nsRange uses int32_t to them.
This patch makes nsRange use uint32_t instead. However, this patch does NOT allow to set over INT32_MAX as offset values since a lot of users of nsRange cannot treat the values as over INT32_MAX because a lot of internal APIs take int32_t as offsets.
For easier to search such points, this patch adds static_cast<int32_t> to uint32_t variables when they are used for int32_t arguments.
And note that nsContentUtils::ComparePoints() behaves odd. It accepts negative offset and compares such value with valid offset simply. This patch still uses int32_t offset variables in nsRange::CompareNodeToRange() even though it may be negative value if nsINode::IndexOf() returns -1 because the caller of it depends on this behavior.
MozReview-Commit-ID: 8RbOgA86JuT
--HG--
extra : rebase_source : 46d526c6d50dfa2f104439b19b8691477b17a4af
This commit changes async keyboard scrolling to be enabled only if the content to
scroll is from a selection. This works around the problem of detecting whether
an arbitrary element has key listeners that should prevent async key scrolling,
because when they have the focus we will have disabled async key scrolling.
MozReview-Commit-ID: 6HhSuGZNsMX
--HG--
extra : rebase_source : 98a6449dd1e913136ca66532a67df8e0bb717e52
This does NOT change variable names like |endNode| because it's not odd and somebody use it for nsINode and endContent for nsIContent. So, changing them needs more work.
MozReview-Commit-ID: 22imUltlu5R
--HG--
extra : rebase_source : 6c93069d0586b37c5084eaa71861085c01da7a7d
This does NOT change variable names like |startNode| because it's not odd and somebody use it for nsINode and startContent for nsIContent. So, changing them needs more work.
MozReview-Commit-ID: H19pTDprRuT
--HG--
extra : rebase_source : 7a7454ac14da48a597ff19a50c863d04dcaddd6e
Web standards use "Container" instead of "Parent". So, nsRange shouldn't use "Parent" for its members and methods.
MozReview-Commit-ID: Ho6N0diuWtE
--HG--
extra : rebase_source : ee4eb7068a68b118c7fe98e9e9e7fa9c9e7f13e2
Most of this patch is updating a few places that use gfxMatrix to use
the equivalent-but-differently-named functions on MatrixDouble:
- Translate/Rotate/Scale get turned into PreTranslate/PreRotate/PreScale
- Transform(Point) gets turned into TransformPoint(Point)
- gfxMatrix::TransformBounds(gfxRect) gets turned into
gfxRect::TransformBoundsBy(gfxMatrix).
- gfxMatrix::Transform(gfxRect) gets turned into
gfxRect::TransformBy(gfxMatrix).
The last two functions are added in this patch as convenience wrappers
to gfxRect instead of Matrix.h because we don't want Matrix.h to "know"
about gfxRect (to avoid adding gecko dependencies on Moz2D). Once we
turn gfxRect into a typedef for RectDouble these will be eliminated
anyway.
MozReview-Commit-ID: BnOjHzmOSKn
--HG--
extra : rebase_source : cf1692d1f0d44a4b05d684a66678739181a426d5
We need to call nsBindingManager's method from ServoStyleSet in a later
patch.
Also, we call nsStyleSet::SetBindingManager() immediately after we call
nsStyleSet::Init(). Therefore, we could pass binding manager to Init()
directly, and delete SetBindingManager().
MozReview-Commit-ID: 43WTpW3mSKX
--HG--
extra : rebase_source : f24f6085c51dfae1a969aee1e073720097d240d5
This commit also schedules empty transactions on input events that trigger a
focus sequence number update.
MozReview-Commit-ID: 1FWXfStLsGv
--HG--
extra : rebase_source : 02c6f07287b6909493f6f88345a3cca93ab3e45b
When keyboard apz is disabled, we don't need to calculate focus targets and
we don't need to update focus state. It should be harmless even if it's done,
but I think it's good to not add something on this critical path that doesn't
do anything.
This commit also disable keyboard map generation in this case too for similar
reasoning. This has the side effect that you can't turn on keyboard apz without
doing a restart.
MozReview-Commit-ID: LxmofT2g7qs
--HG--
extra : rebase_source : 719d29efd80498b824fee03a5be1c1fd05c83074
extra : histedit_source : 7ad71a19782fc6dd203207afbdc7a73a936b3e04
Focus can change at any moment in a document. This causes non-determinism and
correctness problems for doing keyboard apz scrolling. To get around this, we
will maintain deterministic behavior for focus changes initiated by input events
and see if we can get away with more non-determinism for things like `setTimeout`
In order to do this, we disable async keyboard scrolling when an input event is
processed that could have a event listener. We then attach a sequence number to
that input event and dispatch it to content. In content, we record the highest
sequence number that we have processed from an event, and send that on each focus
update. Using this, we can determine in APZ if we have a current focus target or
if we are still waiting for an input event to be processed and focus to be
reconfirmed.
MozReview-Commit-ID: CWcu8YEFQz4
--HG--
extra : rebase_source : 8c54a619bd4f5ee892f0cc8768a10f3e1e4e0b59
extra : histedit_source : 601ca293a028787883841adc6b40e62c0cc829e5
This commit modifies PresShell and nsDisplayList to send a FocusTarget update on
every layer transaction. Ideally we would like to send updates as often as possible,
but this seems like it works well. This can be iterated on later, if necessary.
MozReview-Commit-ID: 8PFqIOhzH77
--HG--
extra : rebase_source : 1e2c3b5620f5d7e6e789848da57b2486c3d74f14
This commit begins the work needed for tracking focus by creating two new classes,
FocusTarget and FocusState. FocusState is created and used by APZCTreeManager to
track the global focus information, while FocusTarget is created per layer tree and
sent to APZ with local focus information. Between the two we are able to figure out
what the correct scrollable layer is to use in response to a keyboard scroll.
See the comment in `FocusState.h` for more details on the architecture and things
needed in future patches to complete this.
MozReview-Commit-ID: F75VZv3i9U2
--HG--
extra : rebase_source : 3d04bced8e13a9884f0c1b320bad8ba2205d7011
This patch makes the following changes to the macros.
- Removes PROFILER_LABEL_FUNC. It's only suitable for use in functions outside
classes, due to PROFILER_FUNCTION_NAME not getting class names, and it was
mostly misused.
- Removes PROFILER_FUNCTION_NAME. It's no longer used, and __func__ is
universally available now anyway.
- Combines the first two string literal arguments of PROFILER_LABEL and
PROFILER_LABEL_DYNAMIC into a single argument. There was no good reason for
them to be separate, and it forced a '::' in the label, which isn't always
appropriate. Also, the meaning of the "name_space" argument was interpreted
in an interesting variety of ways.
- Adds an "AUTO_" prefix to PROFILER_LABEL and PROFILER_LABEL_DYNAMIC, to make
it clearer they construct RAII objects rather than just being function calls.
(I myself have screwed up the scoping because of this in the past.)
- Fills in the 'js::ProfileEntry::Category::' qualifier within the macro, so
the caller doesn't need to. This makes a *lot* more of the uses fit onto a
single line.
The patch also makes the following changes to the macro uses (beyond those
required by the changes described above).
- Fixes a bunch of labels that had gotten out of sync with the name of the
class and/or function that encloses them.
- Removes a useless PROFILER_LABEL use within a trivial scope in
EventStateManager::DispatchMouseOrPointerEvent(). It clearly wasn't serving
any useful purpose. It also serves as extra evidence that the AUTO_ prefix is
a good idea.
- Tweaks DecodePool::SyncRunIf{Preferred,Possible} so that the labelling is
done within them, instead of at their callsites, because that's a more
standard way of doing things.
--HG--
extra : rebase_source : 318d1bc6fc1425a94aacbf489dd46e4f83211de4
This patch reduces the differences between builds where the profiler is enabled
and those where the profiler is disabled. It does this by removing numerous
MOZ_GECKO_PROFILER checks.
These changes have the following consequences.
- Various functions and classes are now defined in all builds, and so can be
used unconditionally: profiler_add_marker(), profiler_set_js_context(),
profiler_clear_js_context(), profiler_get_pseudo_stack(), AutoProfilerLabel.
(They are effectively no-ops in non-profiler builds, of course.)
- The no-op versions of PROFILER_* are now gone. The remaining versions are
almost no-ops when the profiler isn't built.
--HG--
extra : rebase_source : 8fb5e8757600210c2f77865694d25162f0b7698a
It appears that neither Chrome, Safari or Edge support this feature,
and it's causing web-compat issues for us, e.g. bug 1373417.
MozReview-Commit-ID: AP5LMgL6QmR
This patch does the following renamings, which increase consistency.
- GeckoProfilerInitRAII -> AutoProfilerInit
- GeckoProfilerThread{Sleep,Wake}RAII -> AutoProfilerThread{Sleep,Wake}
- GeckoProfilerTracingRAII -> AutoProfilerTracing
- AutoProfilerRegister -> AutoProfilerRegisterThread
- ProfilerStackFrameRAII -> AutoProfilerLabel
- nsJSUtils::mProfilerRAII -> nsJSUtils::mAutoProfilerLabel
Plus a few other minor ones (e.g. local variables).
The patch also add MOZ_GUARD_OBJECT macros to all the profiler RAII classes
that lack them, and does some minor whitespace reformatting.
--HG--
extra : rebase_source : 47e298fdd6f6b4af70e3357ec0b7b0580c0d0f50
Delete the NS_WARNING because it's not needed once this bug is fixed.
MozReview-Commit-ID: Ht3ayjSvYW8
--HG--
extra : rebase_source : 457f011ac64fcd511ec2f15ea2b8ce24a88b0d84
As I've said before, as module owner I prefer that MOZ_ASSERT_IF not be
used in the module because I consider it to be unreadable. However, a
few uses have crept in, and this patch removes them.
I consider it to be unreadable because the name looks like a name that
uses smalltalk-ish naming conventions, i.e., with a part of the name
corresponding to each parameter, in order. However, the parameters are
in the order opposite the name.
This was written primarily with the vim commands:
:%s/MOZ_ASSERT_IF(\([^,]*\),/MOZ_ASSERT(!\1 ||/
:wn
followed by manual cleanup for indentation and removal of !!.
MozReview-Commit-ID: G6rLbOn7k8d
I plan to use it for now to force a full document restyle when a standalone rule
changes or something like that.
In practice, we can do better sometimes, and we may just want to propagate to
the StyleSet all the style change notifications in order to have access to the
rule that changed and all that...
But for now this seemed easier than adding other four or five functions to
StyleSetHandle.
MozReview-Commit-ID: 2BEIliGu4mO
--HG--
extra : rebase_source : 386a1b9fe5ffc5fefbf20678068573ea195043f0
I plan to use it for now to force a full document restyle when a standalone rule
changes or something like that.
In practice, we can do better sometimes, and we may just want to propagate to
the StyleSet all the style change notifications in order to have access to the
rule that changed and all that...
But for now this seemed easier than adding other four or five functions to
StyleSetHandle.
MozReview-Commit-ID: 2BEIliGu4mO
--HG--
extra : rebase_source : 926f8442fbd17c7ffa7f72b6b4a515a28b9aa18b
The ignore-root-scroll-frame flag is generally needed on Fennec, or possibly
other zoom-enabled platforms, when we have them. It allows hit-testing things
outside the main thread's notion of what is visible, because that might occur
when the user zooms out.
For largely historical reasons, we are passing this flag around in other scenarios,
such as when doing hit-tests for touch events, because in the past touch events
and zooming only happened on Fennec, so it didn't matter. Now that we have
touch events enabled on other platforms, such as Windows, we need to make the
distinction clearer.
MozReview-Commit-ID: BlHjtjFYgzv
--HG--
extra : rebase_source : cbfeab4666b5556ba0b38f3abdf666d63945b9a1
The difference is that PostRestyleEventForCSSRuleChanges sets
mRestyleForCSSRuleChanges true. In a subsequent patch,
we propagate a new TraversalRestyleBehavior flag to servo side
if mRestyleForCSSRuleChanges is true.
MozReview-Commit-ID: IKsBbm09uT9
--HG--
extra : rebase_source : 5014c0a978f98e81543ec7766d2daa415317069c
We need to request an animation-only restyle to force flush all throttled
animations on main thread when we handle an event with coordinates
(e.g. mouse event).
MozReview-Commit-ID: KkjeQVsLgTl
--HG--
extra : rebase_source : 314408062e719e9f52df9a6726e2f3dad817bbef
This patch aims to speed up the lookup and storage of DisplayItemData objects, by removing a level of indirection and preventing the previously required hashtable lookup in order to access these. Instead it stores an array of pointers on each frame that allows direct access to the DisplayItemData object by dereferencing the frame. Since most frames get either 1 or 2 DisplayItemData objects attached to them a specialized class is used that is of minimal size (2 * sizeof(void)) and that performs well for sizes 1 or 2.
MozReview-Commit-ID: HONKAmpk5H8
There are situations where events may be handled out-of-order. Allow
coalescing to happen backwards in time as well as forwards to cover them.
MozReview-Commit-ID: 3k7vNFxE1cz
--HG--
extra : rebase_source : 636a3671536747cf62cb410836b4c5a2275bbb49
This patch aims to speed up the lookup and storage of DisplayItemData objects, by removing a level of indirection and preventing the previously required hashtable lookup in order to access these. Instead it stores an array of pointers on each frame that allows direct access to the DisplayItemData object by dereferencing the frame. Since most frames get either 1 or 2 DisplayItemData objects attached to them a specialized class is used that is of minimal size (2 * sizeof(void)) and that performs well for sizes 1 or 2.
MozReview-Commit-ID: HONKAmpk5H8
This avoids conflicts with mozilla::dom::FrameType.
MozReview-Commit-ID: 7aEMbHRaTFk
--HG--
extra : rebase_source : 2d01321f5ce0ec8c0e3f70984674f82678034b3c
The call that's causing the crash seems to be [1], that is, we're trying to
recreate frames for the root element, which should always have a frame created
at the initialization of the PresShell.
So the function I removed in that bug had something like the following:
if (!mDidInitialize) {
// Nothing to do here. In fact, if we proceed and aContent is the
// root we will crash.
return NS_OK;
}
Which PostRecreateFramesFor doesn't guard against (because I thought it was not
needed, per tryserver results).
Sounds a lot like we do need that check, though I'd like to have a testcase
where it happens :(
[1]: http://searchfox.org/mozilla-central/rev/3dc6ceb42746ab40f1441e1e659ffb8f62ae78e3/layout/base/nsCSSFrameConstructor.cpp#2420
MozReview-Commit-ID: Lh6SohNmmI6
--HG--
extra : rebase_source : 5b7076f86d41f5489e47ca16ac2f3620812ee9e8
Say there's a single lag event, a GC or a busy loop, during which the user
types several characters.
Is this one (lag) event? Several (input) events?
We have INPUT_EVENT_RESPONSE_MS which will accumulate several lagged events in
this case. However, that is more of an indication of how users use Firefox than
how good we've been at eliminating sources of lag.
INPUT_EVENT_RESPONSE_COALESCED_MS records the coalesced time spend waiting for
responses to input events. So in this case it will record one value for the
entire duration of the lag.
MozReview-Commit-ID: H5rYnhwF0q3
--HG--
extra : rebase_source : b9be179dac6c6a007d2835b909ec4ce71a3f1c90
This also moves the NoteStyleSheetsChanged to RecordStylesheetChange, which
makes more sense, and stopped special-casing author styles, since it's not
needed now.
MozReview-Commit-ID: 9WKFa0JxqlU
--HG--
extra : rebase_source : 5803620c98309bd2fbb542158e69ef3ddcee9a4b
It's not only inefficient, but also prone to buggyness. Since styles may not be
up-to-date when it happens.
Post a reconstruct instead, which ensures a style flush happens before running
frame construction.
MozReview-Commit-ID: DrakHsJv5fY
Signed-off-by: Emilio Cobos Álvarez <emilio@crisal.io>
--HG--
extra : rebase_source : 11900af908654336cc2391ab9480542c5474e38f
It is possible that events created before the OS goes to sleep will remain
unhandled until after the OS wakes. They will have long response times on
platforms where TimeStamp increments during sleep (like Windows) even though
they really shouldn't. (and the user likely doesn't care if they do).
So don't record those.
MozReview-Commit-ID: 4ybjF8gjkae
--HG--
extra : rebase_source : ff48923c6589d9cbac3284dd81f28598677ca449
NotifyApproximateFrameVisibilityUpdate gets the displayport so we want the base rect set before calling it.
We also don't want to record the displayport if we ignored it in the actual visibility pass.
We want to use a similar model as Chrome on Android does for scaling our display of web content, that is use font inflation for desktop pages and plain text zooming for everything else.
Since we don't want to simply clobber any text zoom that might have been set by the user/front-end code, we allow setting and storing the system font scale separately on the PresContext. We then calculate the effective text zoom value as the product of the system font scale and the current text zoom value.
Any function that is using the PresContext's TextZoom value for layouting/rendering is switched over to this new EffectiveTextZoom value, whereas functions that are interested in the text zoom as actually set by the user/front-end (e.g. the nsDocumentViewer, or the code responsible for copying text and full zoom settings into the new PresContext on page navigation) continue using the plain TextZoom value.
As long as font inflation is enabled in principle (e.g. font.size.inflation.minTwips != 0), every page starts out as eligible for font inflation until the relevant meta viewport tags marking the page as "mobile friendly" have been detected. Since the PresShell caches the font inflation state and only recalculates it when necessary, we make use of that and set the PresContext's system font scale as necessary whenever the font inflation state has been refreshed.
MozReview-Commit-ID: 2InyE04wKAW
--HG--
extra : rebase_source : 3f6d7128f37c1dc18f67a6655f86d9a3003fe90b
extra : source : 6100458b97289f9aea5ac8fda57ded045e6860b7
This patch measures the time from an input event creation, to it been
dispatched to PresShell, but just before any handler been executed.
MozReview-Commit-ID: 6ZYra7YYICY
--HG--
extra : rebase_source : fb864ab2ec7af3d59f4a4ccf52667f34dcd588d6
Instead of copying and concatenating strings into an mDest buffer in
SamplerStackFramePrintfRAII, require callers to keep the string buffer alive
for the duration of the current scope, and store the pointer to the annotation
string in the ProfileEntry. During stackwalking, concatenate the label and the
annotation (separated by a space) and store the resulting string in the
profile buffer.
MozReview-Commit-ID: GEjcLrhhdvb
--HG--
extra : rebase_source : 683749421ee2122805a249cf413e882ee5f33331
This patch measures the time from an input event creation, to it been
dispatched to PresShell, but just before any handler been executed.
MozReview-Commit-ID: 6ZYra7YYICY
--HG--
extra : rebase_source : 019521545132e72113a87c06cfbe562a9915522d
We set the base rect for root scroll frames of root content documents and top level documents because it is quick and easy to calculate it.
Otherwise we ignore the display port if we are missing a base rect, and treat it like a scrollframe without a displayport. This is likely better behaviour as if the base rect is missing the computed displayport may not even cover what is currently visible.
This patch is written by the following script with some manual adjustment to
the comment in nsRubyTextContainerFrame.cpp and nsRubyFrame.cpp, and
nsColumnSetFrame's constructor.
function rename() {
find layout\
-type f\
\( -name "*.cpp" -or\
-name "*.h" \)\
-exec sed -i -r "s/$1/$2/g" "{}" \;
}
rename "nsReflowStatus *([a-zA-Z0-9]*) = NS_FRAME_COMPLETE" "nsReflowStatus \1"
rename "([a-zA-Z0-9.*]*) *= NS_FRAME_COMPLETE;" "\1.Reset();"
rename "([a-zA-Z0-9.*]*) == NS_FRAME_COMPLETE" "\1.IsEmpty()"
MozReview-Commit-ID: 9tqQAHvdQex
--HG--
extra : rebase_source : 3119776946dc2c8350098b7bf9f3ceff29bdffb5
This adds markers for FireScrollEvent, DispatchSynthMouseMove, ForwardTransaction, and NotifyDidPaint.
MozReview-Commit-ID: 9XahFGzIOls
--HG--
extra : rebase_source : 1985b2a11777a17dac621a35632f138a76224b0f
This patch does a few things:
* Buckets invalidations by transaction ID, and sends MozAfterPaints events for them when the associated composite completes.
* Creates a separate EventualDidPaint timer for each transaction ID we have invalidations for rather than just using one.
* Removes NotifyDidPaintForSubtree(PAINT_LAYERS), as it was only necessary for the existing bucketing mechanism.
MozReview-Commit-ID: JERMsgxhPQd
--HG--
extra : rebase_source : aad23cb4e77afe12ddf4ebf21db36ae6edec9692
We added this so that MozAfterPaint events would be delivered to hidden documents as part of bug 539356, but I don't remember what needed it.
It doesn't appear to be necessary for any tests any more, so let's just get rid of it
MozReview-Commit-ID: HcmIjstZyLQ
--HG--
extra : rebase_source : 3f58129b3b1588e5d39e00c157e4bdba1afeed0c
This is the bulk of the changes.
- DisplayItemScrollClip is removed. Instead, we will have 1) ActiveScrolledRoot
and 2) DisplayItemClipChain.
- ActiveScrolledRoot points to a scroll frame and allows traversing up the
scroll frame chain.
- DisplayItemClipChain is a linked list of clips, each clip being associated
with the ActiveScrolledRoot that moves this clip.
- Each display item has an ActiveScrolledRoot and a clip chain.
- nsDisplayItem::GetClip returns the item of the clip chain that scrolls with
the item's ASR. The separation between "regular clip" and "scroll clips"
mostly goes away.
- Tracking clips in the display list builder's clip state happens very
similarly to how regular clips used to be tracked - there's a clip chain for
content descendants and a clip chain for containing block descendants. These
clip chains are intersected to create the combined clip chain.
- There are strict rules for the ASR of a container item: A container item's
ASR should be the innermost ASR which the item has finite clipped bounds with
respect to.
- At some point in the future, ASRs and AGRs should be reunified, but I haven't
done that yet, because I needed to limit the scope of the change.
MozReview-Commit-ID: KYEpWY7qgf2
--HG--
extra : rebase_source : c727f6300a35463750639e165bfa37374c06b851
Prior to this patch, dumping of style contexts in the layout
debugger would only dump the subtree associated with the root
frame, which just covers the viewport, canvas, and associated
scrollbars.
I think this has been broken since bug 236921 (2004) and
bug 378693 (2007) properly disconnected the root element from the
styles for the viewport and canvas.
This patch just manually dumps the root frame and the root
element's frame, since the style set no longer maintains a list
of all of the style context roots.
MozReview-Commit-ID: 5LEnbNtngN2
--HG--
extra : transplant_source : v%BB%09%C8M8%FE%25%A9%A6%3D%EA%BD%BE%DF7%08%9A%85m
