When memory gets low on a device, activities flush everything they can.
Hardware-accelerated activites, such as Launcher, flush GL resources and destroy
the GL context. However, some resources were still hanging around, due to deferred
destruction policies (we don't delete layers until the DisplayLists they are in
are finalized, to ensure we don't deref deleted objects). This meant that we were
referring to obsolete GL data in these objects. in particular, it meant that we might
come around later, after a new GL context was created, and delete a texture object
that was incorrect. We use the layer's "texture id" to refer to the texture underlying the
layer. But if there's a new GL context, then this texture ID is no longer valid, and
we may be deleting the texture that a different object (layer, icon, whatever) is referring
to, because the driver may return that same ID under the new GL context.
The fix is to more aggressively delete things that we know will not be used again
when the GL context is destroyed. In particular, we delete all resources being used
by all DisplayLists at GL context destruction time.
Issue #7195815 Textures corruption on all devices, in many apps
Change-Id: I52d2d208173690dbb794a83402d38f14ea4c6c22
...#testScreenLayout failures on JO
This doesn't actually fix it; I have concluded that the test is broken
(the platform is correctly reporting that this is a NOT LONG device
because in portrait once you account for the status bar and system
bar our size is 880dp high and 600dp wide, which is not enough for us
to be in the LONG config).
However while working on this I noticed that the code for computing
the configuration of the external display was wrong. I have fixed
that by putting this code for computing these parts of the configuration
in a common place that both the window manager and external display
code can use.
Change-Id: Ic6a84b955e9ec345a87f725203a29e4712dac0ad
- do not need those variables. Use what we already have in View.
- reset padding to initial values before changing it.
Change-Id: Ib396b3dca6e98a94d83a538a9b594b5eb426c453
When computing the adjusted view bounds, don't constrain the
dimensions by the original estimate if the opposite dimension has a
fixed size. This can result in the view never getting properly
enlarged.
Also fix a long-standing bug in MeasureSpec.makeMeasureSpec where
oversized or negative values could result in broken packed values.
Bug 7240251
Change-Id: I359d108ff52b6f3b5c4bf393d2271d28999c0127
Issue #7211769: Crash dialog from background user has non-working "report"
The report button now launches the issue reporter for the correct user.
Also for crashes on background users, either disable the report button,
or simply don't show the dialog depending on the build config.
Issue #7244492: Bugreport button in Quick Settings doesn't actually do anything
Now they do.
Issue #7226656: second user seeing primary user's apps
I haven't had any success at reproducing this. I have tried to tighten up
the path where we create the user to ensure nothing could cause the
user's applications to be accessed before the user it fully created and thus
make them installed... but I can't convince myself that is the actual problem.
Also tightened up the user switch code to use forground broadcasts for all
of the updates about the switch (since this is really a foreground operation),
added a facility to have BOOT_COMPELTED broadcasts not get launched for
secondary users and use that on a few key system receivers, fixed some debug
output.
Change-Id: Iadf8f8e4878a86def2e495e9d0dc40c4fb347021
When apps request layouts while scrolling is in progress, Gallery
children can get offset in strange ways. Compensate for this by
tracking the last known offset and applying it during layout.
Bug 7245853
Change-Id: I9d746ae6bb06918e2d920c58052f72e749a7ffbf
