Make the stylesheet cache respect the same headers as the image cache
does. This makes no-cache stylesheets work as they do now, which is
useful for developers that want to develop sites locally, and for
shift-reloads, etc.
Differential Revision: https://phabricator.services.mozilla.com/D78659
This patch implements a per-process cache of parsed stylesheets for
non-inline sheets. The entries are evicted when the document gets
destroyed and there's no other document with the same principal around.
This works fine in practice even when navigating because CC happens
pretty late, but we could add an extra timer if we deem it worth it.
I had to adapt some tests so that they keep passing. They were already
clearing various image / network caches so it seems fine to also clear
this one.
Note that there's a very subtle change in the load data key: We only
miss the cache if the referrer _policy_ is different, not if the
referrer is different. While that is slightly dubious, that is the only
think that makes this effort somewhat worth it. Otherwise stylesheets
would have to be re-fetched if the referrer is different, which
effectively would mean to re-parse it if the document URI is different,
which is bad.
It seems like the network cache only keys on the referrer policy, so it
seems fine to do the same.
Differential Revision: https://phabricator.services.mozilla.com/D77457
The 'asyncStack' flag on JS execution contexts is used as a general switch
to enable async stack capture across all locations in SpiderMonkey, but
this causes problems because it can at times be too much of a performance
burden to general and track all of these stacks.
Since the introduction of this option, we have only enabled it on Nightly
and DevEdition for non-mobile builds, which has left a lot of users unable
to take advantage of this data while debugging.
This patch enables async stack traces across all of Firefox, but introduces
a new pref to toggle the scope of the actual expensive part of async stacks,
which is _capturing_ them and keeping them alive in memory. The new pref
limits the capturing of async stack traces to only debuggees, unless an
explicit pref is flipped to capture async traces for all cases.
This means that while async stacks are technically enabled, and code could
manually capture a stack and pass it back to SpiderMonkey and see that stack
reflected in later captured stacks, SpiderMonkey itself and related async
DOM APIs, among others, will not capture stacks or pass them to SpiderMonkey,
so there should be no general change in performance by enabling the broader
feature itself, unless the user is actively debugging the page.
One effect of this patch is that if you have the debugger open and then close
it, objects that have async stacks associated with them will retain those
stacks and they will continue to show up in stack traces, no _new_ stacks
will be captured. jorendorff and I have decided that this is okay because
the expectation that the debugger fully revert every possible effect that it
could have on a page is a nice goal but not a strict requirement.
Differential Revision: https://phabricator.services.mozilla.com/D68503
When trying to get the current focused element during reflow, when the current
focused element in the document is a frame which has yet to be initialized,
attempting to walk through that frame's BrowsingContext causes immediate frame
loader initialization. Since frame loader initialization often requires running
scripts, that's unsafe, which is why we defer initialization until the end of
any in-progress reflow or document update.
This patch fixes the focus manager code to ignore any uninitialized frame
loaders it comes across, since by definition they cannot have any
currently-focused element until their subdocuments have been created.
Differential Revision: https://phabricator.services.mozilla.com/D79392
There are some issues with this code. It's using doubles, but SetRect
only takes floats, so we lose precision there already. Instead of
getting fancy, use the same conversion that the rest of Gecko uses, so
that pages that compute style values out of getBoundingClientRect values
don't get minor floating point changes.
Differential Revision: https://phabricator.services.mozilla.com/D79326
This removes the diagnostic warnings which used to be logged when the
Large-Allocation header was present, but failed to switch into a
Large-Allocation process. Due to the low adoption of the header, this shouldn't
be too large of a problem, but we can look into re-adding the diagnostics if
needed in the future.
The new codepath no longer performs multiple network requests for
Large-Allocation resources, and now relies on the battle-tested
DocumentLoadListener codepath for process switching.
Differential Revision: https://phabricator.services.mozilla.com/D78998
These were only used by process switches triggered via DocumentChannel, which
now take the PROCESS_BEHAVIOR_STANDARD codepath.
Differential Revision: https://phabricator.services.mozilla.com/D78973
Due to the PPDC process switching changes, we also handle switching between
remote and non-remote browsers with DocumentLoadListener. Adds support to
CanonicalBrowsingContext::ChangeFrameRemoteness to allow it to handle non-remote
browsers as well.
Differential Revision: https://phabricator.services.mozilla.com/D78971
Add a series of extra hooks and methods to allow tabbrowser to use the new
process switching codepath. This duplicates some of the logic from
`updateBrowserRemoteness` into event handlers.
Differential Revision: https://phabricator.services.mozilla.com/D78970
This new process switching behavior is only enabled for some browser elements,
which have specified a specific attribute. Turning this on for all browsers with
a `remote` attribute causes breakage in reftests.
The initial version does not handle switching from remote to parent or
vice-versa, that is covered in a later part.
Differential Revision: https://phabricator.services.mozilla.com/D78969
Some existing tests crash if we just add the assertions and the rest of the patch
fixes those.
mServiceNotifier is now created always on the main thread and used only on the target thread.
And use of it is protected by the mutex so that at the same time when it is created, target thread
can't try to delete it.
Differential Revision: https://phabricator.services.mozilla.com/D79269
In order to delegate the permission to the top-level window, in this
patch, we pre-compute the permissions of the top-level context and set
them to the top-level WindowContext. So, the cross-origin iframe can
know the permission of the top-level window through the WindowContext.
Thus, the permission can be delegated in Fission.
Differential Revision: https://phabricator.services.mozilla.com/D79132
It will simply forward the interface to the main thread.
This is the final step required to remove AbstractThread support here.
Differential Revision: https://phabricator.services.mozilla.com/D79095
...and be still super safe by adding null checks.
Some existing tests crash if we just add the assertions and the rest of the patch
fixes those.
Differential Revision: https://phabricator.services.mozilla.com/D79269
Make the stylesheet cache respect the same headers as the image cache
does. This makes no-cache stylesheets work as they do now, which is
useful for developers that want to develop sites locally, and for
shift-reloads, etc.
Differential Revision: https://phabricator.services.mozilla.com/D78659
This patch implements a per-process cache of parsed stylesheets for
non-inline sheets. The entries are evicted when the document gets
destroyed and there's no other document with the same principal around.
This works fine in practice even when navigating because CC happens
pretty late, but we could add an extra timer if we deem it worth it.
I had to adapt some tests so that they keep passing. They were already
clearing various image / network caches so it seems fine to also clear
this one.
Note that there's a very subtle change in the load data key: We only
miss the cache if the referrer _policy_ is different, not if the
referrer is different. While that is slightly dubious, that is the only
think that makes this effort somewhat worth it. Otherwise stylesheets
would have to be re-fetched if the referrer is different, which
effectively would mean to re-parse it if the document URI is different,
which is bad.
It seems like the network cache only keys on the referrer policy, so it
seems fine to do the same.
Differential Revision: https://phabricator.services.mozilla.com/D77457
Those methods have two sources to check after call: the return value and the pointer. This can be confusing as a caller may think they should check both when they don't need to. Since the two always behaves together (a valid pointer + NS_OK, or nullptr + NS_ERROR_FAILURE), this replaces the return value with the pointer.
Differential Revision: https://phabricator.services.mozilla.com/D79196
* Majorly simplity CanvasRenderer
* Replace GLScreenBuffer with trivial GLSwapChain
* Use descriptor structs so that future SharedSurface changes aren't so painful
to propagate
* Mortgage/strip out more OffscreenCanvas code for now
Differential Revision: https://phabricator.services.mozilla.com/D75055
Henri Sivonen
Emilio Cobos Álvarez
Mirko Brodesser
Simon Giesecke
Logan Smyth
Karl Tomlinson
Narcis Beleuzu
Kris Maglione
Olli Pettay
Nika Layzell
Tim Huang
Jean-Yves Avenard
Kartikaya Gupta
Razvan Maries
Dorel Luca
Kagami Sascha Rosylight
Jeff Gilbert
Mihai Alexandru Michis