| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Currently we are still passing GtkWidget* into the renderer and
trusting the value on the way out. With this patch we switch to using
opaque values.
These opaque values are handled by a GtkNativeViewIdManger, a
singleton object which maintains the list of currently valid ids and
their current X window ids.
We don't pass the X window ids directly to the renderer because they
are a) guessable and b) possibly variable for a GtkWidget. From a
patch size point of view, the X window isn't current created at the
point where we need it so significant work would be needed to reorder
operations to fix that as well.
This patch also removes the GTK accesses from the BACKGROUND_X11
thread which were a temporary hack.
http://codereview.chromium.org/92110
BUG=9014,9869,10787
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@14405 0039d316-1c4b-4281-b951-d872f2087c98
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
(r14075 take two)
Currently we perform several X operations on the IO thread including
geometry and clipboard work. This is causing races inside Xlib and
crashing the browser.
These are the result of synchronous calls from the renderer, so we
cannot route these requests to the UI thread without risking deadlock.
Thus we introduce the UI2 thread. This thread has a second connection
to the X server and can perform X operations safely the without UI
thread.
Work remains to be done:
Since we still have the hack where we pass GtkWidget pointers into the
renderer and back, we still have to access these structures from the
IO and UI2 threads. This still needs to be fixed, but this is not the
patch for it.
Also, not all the X calls from the IO thread have been moved over in
this patch; just a few small ones.
http://codereview.chromium.org/67145
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@14206 0039d316-1c4b-4281-b951-d872f2087c98
|
| |
|
|
|
|
|
| |
Reverts r14075
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@14080 0039d316-1c4b-4281-b951-d872f2087c98
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Currently we perform several X operations on the IO thread including
geometry and clipboard work. This is causing races inside Xlib and
crashing the browser.
These are the result of synchronous calls from the renderer, so we
cannot route these requests to the UI thread without risking deadlock.
Thus we introduce the UI2 thread. This thread has a second connection
to the X server and can perform X operations safely the without UI
thread.
Work remains to be done:
Since we still have the hack where we pass GtkWidget pointers into the
renderer and back, we still have to access these structures from the
IO and UI2 threads. This still needs to be fixed, but this is not the
patch for it.
Also, not all the X calls from the IO thread have been moved over in
this patch; just a few small ones.
http://codereview.chromium.org/67145
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@14075 0039d316-1c4b-4281-b951-d872f2087c98
|
| |
|
|
|
|
|
|
|
|
|
| |
VNC servers don't support Xrender. For this use case, we implement a
slow fallback which byte-fiddles the Skia bitmaps as needed to support
32 and 24 bit visuals.
Review URL: http://codereview.chromium.org/27227
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@10523 0039d316-1c4b-4281-b951-d872f2087c98
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The NX X server doesn't support xRGB32 XRENDER picture formats. So, if
we don't find one of those formats, fall back on ARGB32, which is
fine, except that the X server will spend time dealing with an alpha
channel which is all 254 or 255. ARGB32 support is required by the
XRENDER spec.
(This doesn't fix VNC, a patch for that is forthcoming.)
Review URL: http://codereview.chromium.org/28192
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@10484 0039d316-1c4b-4281-b951-d872f2087c98
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
When we remove a drawing area from a container, GTK destroys the X
window. However, the BackingStore remembers the old X window id and
emits X requests using the wrong id.
This change makes the window id an argument to ShowRect and makes the
pixmaps from the root window instead.
Review: http://codereview.chromium.org/27169
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@10439 0039d316-1c4b-4281-b951-d872f2087c98
|
| |
|
|
|
|
|
| |
Review URL: http://codereview.chromium.org/28135
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@10378 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
This converts Linux to using server-side backing stores. Rather than
keeping a backing store in heap memory in the browser, we create a
pixmap in the X server. This means that expose messages don't require
us to transport any images to the X server, we can just direct it to
paint from the pixmap. Also, scrolling can be performed server side
also. This greatly improves performance over remote X.
Also, shared memory transport to the X server is implemented. Shared
memory segments can be created in the renderer and mapped directly
into the X server.
Review URL: http://codereview.chromium.org/27147
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@10369 0039d316-1c4b-4281-b951-d872f2087c98
|
