This patch (currently WIP) alters the way we determine whether jank is user-visible or not.
Instead of measuring the total time spent doing JS, we now use an
indicator provided by the vsync driver: how long it takes to deliver
the signal from the vsync timer to the main thread. This lets us find
out more accurately if there is user-visible jank. In the future, this
will also let us add an observer to find out whether the process
itself is janky, regardless of JS.
--HG--
extra : rebase_source : a538e3cc9d8904f52d4a0e7bad291189986e4e6d
Displayport margins change by small amounts on almost every single scroll. We do not want to update image visibility nearly that often.
As the comment, and the original bug (bug 1169881) suggest this is only meant to catch rather large changes in display ports as we already have means to trigger an image visibility update via a scroll position change and via any style or layout flush.
This is a regression from bug 1002992 where we switch from the display list builder to the frame tree walker and didn't update mLastUpdateImagesPos in the frame walker.
Experience shows that we do not have enough profiler labels to make
BHR hang reports meaningful. This patch adds enough labels to let us
exploit hang reports matching the 25 topmost chrome hangs.
--HG--
extra : rebase_source : b9ec379c58255a250db1020377147c95c82df712
Native events come with timestamps indicating when they were created or
generated by the user. Using those, we can get a full picture of how long it
takes between the user trying to do something and us responding to it.
This is currently only for ports that populate WidgetEvent's timeStamp
(presently gtk, Windows).
Experience shows that we do not have enough profiler labels to make
BHR hang reports meaningful. This patch adds enough labels to let us
exploit hang reports matching the 25 topmost chrome hangs.
--HG--
extra : rebase_source : b9ec379c58255a250db1020377147c95c82df712
All the event interfaces changed except for nsIDOMUIEvent and its inheritors.
--HG--
extra : transplant_source : %A5U%3F%80%2B%DD%01%F4%D8%21%F2%E9z%C1%D6%AA%CC%D4%EC%F8
We store the original value of duration in AnimationTiming, and add
computed duration in ComputedTiming, so both the Timing model and
AnimationEffectTimingReadOnly can get what they want.
By the way, replace mIterationDuration with mDuration.
--HG--
extra : rebase_source : f8e1fd648572e6d7b1cbecc2ac1888a2f74bbc7e
FillMode could be 'auto', and we should treat it as 'none' in the timing model.
However, AnimationEffectTimingReadOnly should get its original value.
By the way, replace mFillMode with mFill.
--HG--
extra : rebase_source : 5a397dd7fbb22ac76fe96003d82d097e398852c7
We want to store the original value from KeyframeEffectOptions whose
iterations is unrestricted double. Therefore, we can get the original value
of iterations by AnimationEffectTimingReadOnly.
By the way, replace mIterationCount with mIterations.
--HG--
extra : rebase_source : da2953056031079c41273ed977545dc926e1b83c
When updating animations, we shouldn't unnecessarily clobber the "wins in
cascade" state of their properties since this can lead to unnecessary restyles
when we then decide we need to update the cascade.
RestyleManager currently has a piece of state for tracking if throttled
animations are up-to-date or not. Actually, it's not so much about throttled
animations but really about outstanding changes to animation styles (which
is typically expected to be due to throttling animations but there are
other cases that invalidate the animation style rule that we should be
considering here).
We now have that same information stored in the EffectCompositor so we can
remove the redundant state from RestyleManager. Furthermore, the state stored
in EffectCompositor is more accurate since it captures the case when animation
style needs to be updated twice within a tick, or when nothing needs to be
updated within a tick.
This patch, therefore, introduces EffectCompositor::HasPendingStyleUpdates in
place of setting RestyleManager::mLastUpdateForThrottledAnimations.
nsTransitionManager also uses mLastUpdateForThrottledAnimations to warn if we
have not processed throttled animations. We can't use HasPendingStyleUpdates
here however, since it will return true in the case where we have triggered new
transitions in the process of restyling. However, any new transitions will
trigger "standard" (i.e. not throttled) restyles so we introduce another
method, HasThrottledStyleUpdates, that returns true only if we have outstanding
throttled updates and use this for the warning inside nsTransitionManager.
nsPresContext contains a mLastStyleUpdateForAllAnimations flag which is simply
used to prevent unnecessarily posting restyles when throttled animations are
already up to date. Since part 13 we now accurately record whether we have
posted a restyle for each throttled animation and only post a restyle if we
have not done so already. As a result, this flag is no longer needed since
calling PostRestyleForThrottledAnimations is effectively a noop when throttled
animations are up-to-date.
This flag is no longer needed because in bug 1232563 we introduced a more
thorough optimization that detects when the animation is not changing by
comparing the progress value between samples and avoids requesting restyles
when it does not change.
Now that we track whether or not animations are up to date using the hashset in
EffectCompositor, we can remove the mStyleRuleRefreshTime flag that is, as of
part 5 of this patch series, now only used for detecting whether or not
animations are up to date.
In order to preserve the existing behavior of FlushAnimations, however, this
patch temporarily introduces a method to indicate if there are throttled
animations or not.
It might not be obvious that FlushAnimations is only concerned with throttled
animations due to its name. FlushAnimations is simply intended to post
animation restyles for out-of-date animations. Any animations that are *not*
throttled will either be up to date, or we will have already posted an
animation restyle so we only need to consider throttled animations in this case.
This patch continues to migrate functionality from
AnimationCollection::RequestRestyle to EffectCompositor::RequestRestyle.
In order to post the animation restyle from the EffectCompositor, this patch
also moves the PostRestyleForAnimation method to EffectCompositor.
The GetElementToRestyle method is temporarily duplicated in both
EffectCompositor and AnimationCollection however we will remove the version in
AnimationCollection later in this patch series.
In this patch series we are gradually migrating style rule updating
functionality from AnimationCollection to EffectCompositor. This patch moves
part of the RequestRestyle method from one class to the other.
Note that in both cases we only call SetNeedsStyleFlush if we haven't already
posted an animation restyle. (In the case of AnimationCollection we check this
using the mHasPendingAnimationRestyle flag, and in EffectCompositor case we
simply check if the element is already in the "needs restyle" hashmap. If it is,
either we already have a throttled restyle and have called SetNeedsStyleFlush or
we have a standard restyle and have posted an animation restyle.)
The added check for a null pres context matches the behavior of
AnimationCollection::RequestRestyle which has an equivalent early return at the
beginning of the function.
This patch uses the presence/absence of (pseudo-)elements in the "needs
animation rule update" hashmap on EffectCompositor to detect if a style update
is required. The various flags in AnimationCollection that do a similar job
still remain so that we can remove them one-by-one in subsequent patches in
this series.
This is in preparation for moving RequestRestyle to EffectCompositor (and
because we'll run into compile issues if we don't since AnimationCommon.h
includes too many interdependent definitions).
Since we want to track elements needing a restyle on EffectCompositor we need
to scope it to an nsPresContext rather than just making if a collection of
static methods.
Even though the content of the root scroll frame is the root element, the primary frame of the root element is never the root scroll frame. This is even true if the normal primary frame of the root element is not created (say because it is display: none). Leaving the primary frame of the root element to be null even though there are frames (the root scroll frame, and the canvas frame) that have the root element as their content.
This behaviour is more consistent by not ignoring a root scroll frame when it exists.
And to several functions above it in the callgraph, ones that no longer need an
nsRenderingContext.
--HG--
extra : rebase_source : 331d00be421316a7071ab4b93894431c36a031d5
We change the color and anti-aliasing on the gfxContext but never do anything
with it while those values are changed.
--HG--
extra : rebase_source : 39ab84c17f8b4b482383464b2b8e0552369fb94c
Instead do it when we first encounter the root scroll frame.
Doing this goes back to bug 635053, where we did it because fixed position items weren't getting included. However in bug 974643 we learned that this was wrong. Displayports aren't relevant to fixed pos content, displayports are only relevant to scrolled content. And we set the dirty rects of fixed pos content specially. The only other thing that should be affected is scrollbars, and we already carefully set their dirty rects too.
The flag SkipParentDisplayBasedStyleFixup is for flex/grid fixup, and it is set
for all pseudo elements other than before/after. This is not desirable for ruby
case.
Moving the code out also means elements will inherit the suppress flag directly
from "display: contents" parent (instead of the container), which is fine since
the parent should have had its flag set properly as well.
--HG--
extra : source : ab8d1c650ec6aa41cb165c278de9e06182a0c733
Temporarily rename GetDisplayPort to GetDisplayPortRelativeToScrollPort for the duration of this patchset.
This means that every caller of GetDisplayPort is guaranteed to be touched by this patchset (assuming it compiles), and thus each call site can be checked in review to make sure it is relative to the correct coordinate system.
This also has the side effect of making the computation that takes place in GetDisplayPortFromMarginsData happen relative to the scrollport. As opposed to some things relative to the scroll port, and some things relative to the scroll frame.
This makes the one caller that needs the displayport rect have to ask for it seperately.
The reason for this is later in the patch series we need to add "RelativeToScrollFrame/Port" to all displayport getters, but there is no semantically good way to do that to the name GetOrMaybeCreateDisplayPort.
This makes the one caller that needs the displayport rect have to ask for it seperately.
The reason for this is later in the patch series we need to add "RelativeToScrollFrame/Port" to all displayport getters, but there is no semantically good way to do that to the name ViewportHasDisplayPort.
XUL documents don't have a root scroll frame, so this would early return and prevent any of the tiling and max texture size code from running.
The root element of XUL documents likely only ever has zero-margin displayports, and they are not scrollable, so it may not be important.
This patch rearranges Paint() so that the next nsRenderingContext is created
via the one-arg constructor rather than via the zero-arg constructor + Init().
This means the SetLineWidth(1.0) call in Init() no longer occurs but that's ok;
it was redundant because 1.0 is the default line width in a new gfxContext.
--HG--
extra : rebase_source : d3aa44310e180f8a297feab281a92233bd4859ce
It doesn't need to store the nsIContent. It just needs to record if mPointerId
has been set. This is because ReleasePointerCapturingContent() doesn't use its
second parameter.
--HG--
extra : rebase_source : d31036c74b06143d88848f63165cd1f41b433102