When PluginInstanceChild receives native key events, it should post the events to the chrome process first for checking if the key combination is reserved. However, posting all key events to the chrome process may make damage to the performance of text input. Therefore, this patch starts to post a key event whose key combination may be a shortcut key. However, for avoiding to shuffle the event order, it posts following key events until all posted key events are handled by the chrome process.
For receiving response from widget, this patch defines nsIKeyEventInPluginCallback. It's specified by nsIWidget::OnWindowedPluginKeyEvent() for ensuring the caller will receive the reply. Basically, the caller of nsIWidget::OnWindowedPluginKeyEvent() should reply to the child process. However, if the widget is a PuppetWidget, it cannot return the result synchronously. Therefore, PuppetWidget::OnWindowedPluginKeyEvent() returns NS_SUCCESS_EVENT_HANDLED_ASYNCHRONOUSLY and stores the callback to mKeyEventInPluginCallbacks. Then, TabParent::HandledWindowedPluginKeyEvent() will call PuppetWidget::HandledWindowedPluginKeyEvent().
MozReview-Commit-ID: G6brOU26NwQ
--HG--
extra : rebase_source : 8140456de278956d2d594e85c7b397ae366b4962
This patch separates Windows' message definition and related definitions from nsWindowDefs.h to mozilla/widget/WinMessages.h because the definitions are useful in plugin module but nsWindowDefs.h cannot be included from plugin module since it depends on some headers in under widget.
MozReview-Commit-ID: KNUAffLMKT1
--HG--
rename : widget/windows/nsWindowDefs.h => widget/windows/WinMessages.h
extra : rebase_source : ae9390e700ca97a0a4eea3b2977d4d5568ce9408
js::Class op are often all null. And when they're not all null, they're often
duplicated among classes. By pulling them out into their own struct, and using a
(possibly null) pointer in js::Class, we can save 114 KiB per process on
64-bit, and half that on 32-bit.
* * *
imported patch separate-ClassOps-2
--HG--
extra : rebase_source : bd751bf247e9491c1966a123dbeffa573657dfb1
js::ClassExtension is often all null. When it's not all null, it's often
duplicated among classes. By pulling it out into its own struct, and using a
(possibly null) pointer in js::Class, we can save 17 KiB per process on
64-bit, and half that on 32-bit.
--HG--
extra : rebase_source : eb78ade09ce268e886d091f6cbc38d7e5e912527
js::ObjectOps is often all null. When it's not all null, it's often duplicated
many times among classes. By pulling it out into its own struct, and using a
(possibly null) pointer in js::Class, we can save 208 KiB per process on
64-bit, and half that on 32-bit.
--HG--
extra : rebase_source : 5be8fe45f652392571b8a6d7b63777cbafba6ae4
The main change is that we move AutoJSExceptionReporter up to before the first
place where we might throw an exception, so we report those exceptions before
returning. The change to use AutoEntryScript consistently is because all of
these callsites can run JS in practice, and it seems reasonable to allow them
to.
When turned on e10s, plugin process creates from chrome process. So content
process doesn't know current sandbox level. To rewrite wmode attribute on
contnet process by sandbox level >= 2, we should store sandbox level to
nsPluginTag.
MozReview-Commit-ID: DCQ5g7uCbJF
--HG--
extra : rebase_source : 7064f469e873ea1301e28faeab0bbb27613978db
Yoshi Huang
Kyle Huey
Masayuki Nakano
David Anderson
Nicholas Nethercote
Jonathan Watt
Kyle Machulis
Arata Furukawa
Christoph Kerschbaumer
Boris Zbarsky
Mike Hommey
Dylan Roeh
Wes Kocher
sakshi
Jim Mathies
Bogdan Postelnicu
Makoto Kato
Aidin Gharibnavaz
Karl Tomlinson
Ben Kelly
Birunthan Mohanathas
Phil Ringnalda