Bug #7385090
This change gets rid of two silly asumptions:
- That a layer needs to be cleared with opaque black (it shouldn't,
it's already cleared to transparent and the view will cover it up
with its own background)
- The the clip should be dirty at the beginning of a frame only
when the render target is opaque
Change-Id: I415b6d3cab196057fb0281419a53fef601a44e28
The native side of the tracing code latches a copy of the tags from a
system property on first use. The Java-side tracing code latches a
copy of the native's copy during class init. The tracing code is
preloaded by the zygote, which means we get the flags during zygote
init and don't update them when we launch a new app.
This changes the Java sources to also defer initialization until
first use, so that newly-launched apps will use the current value
of the system property.
Bug 7323431
Change-Id: I7db048ec54345ae9565088a35c2e2b4c82f993fd
Define external storage as being isolated for each user, with the
caveat for shared OBB files. Also mention external storage
permissions.
Bug: 7384941
Change-Id: Ib156b5874c5587749d7bc066791e9e88d931b174
This patch fixes bug 7346656. In this particular case, the text line in
the EditText was split into multiple spans, with the boundary between
the "r" and "," in "r,". These were being drawn as two separate runs,
but measured as a single run, leading to inconsistent measurements
because this is a kern pair in Roboto.
The fix is to eliminate the special-case code for measuring. This will
actually improve efficiency, as the value computed in one pass is now
more likely to be reused in another.
Change-Id: I04142a0ec98f280fc1027c7cbdbf903e3096f8e4
Bug: 7368245
Log a warning if the system process calls unqualified sendBroadcast()
and other calls.
As a result of the logging above, found a few more method calls such as
bindService() that would benefit from being more explicit to avoid
future confusion and reduce the log warnings.
Change-Id: I17f15c8be9adf7becd456d6abbab606f19befdbf
This was a regression introduced by a previous CL.
- we actually need to let the layout params resolution goes all the time
as we dont have any "isResolved" state and a public API for doing its reset thru ViewGroup.
The current implementation is simple and works even if it is doing sometimes a bit more
than it should really do. This is a well conscious tradeoff.
Change-Id: I5c4d532331b3970dfe88f016bc305cbc4a0d83f1
This change makes use of the new 'secure' argument to the
ISurfaceComposer::createDisplay method. In this change both the overlay and
wifi displays are hard-coded to be non-secure displays.
Bug: 7368436
Change-Id: Ib65312f2adab5104d8deefbfc32af9dc106a9129
Make USER_REMOVED an ordered broadcast and send it before the user's
state is completely removed from the system. This gives services the
opportunity to clean up their state, while still having access to the
user's directory and UserInfo object (such as serial number).
Tell SyncManager to skip over dying/partially created users.
Improve UserManager tests, waiting for users to be removed fully.
Bug: 7382252
Change-Id: I93cfb39c9efe6f15087bf83c569a2d154ef27168
In particular, make it quite clear that the set returned by
getStringSet() must not be modfied by the app, period; and
add a similar caution about the map returned via getAll().
Also, fix a bug that could lead to unexpected data being committed
if the set instance passed to putStringSet() was mutated by the
app after that call (including mutations after commit() was
invoked).
Bug 7312641
Change-Id: If9a1be1b0669ac879ff7a7dc67a8805548ea10cc
The on-disk format of VpnProfile only stores the username/password
when saveLogin is set, which was dropping them across IPC boundaries.
Bug: 7378327
Change-Id: I30c16efb358ab8f8d3d9229bbf9811146fe034ce
So that:
- the values can be shared (to fix assoc bug)
- the values can be customized in product overlays
Bug:7373284
Change-Id: I37f037082523a3d975f6014f36afa28c60117372
- set the popup layout direction depending on the anchor view's layout direction
(if not defined before)
- make first pass of ViewRootImpl.performTraversals() and configuration update
not override any layout direction that could have been set before
Change-Id: I8e86ca805f0caf52c058d06aa7861df56fb862e6
Try to get installd to free up cache before giving up when there is too
little space free.
Bug: 7232123
Change-Id: Ie3c8ca8dfc190abbb9a29a7baee31f32e9de7d69
Bug #7326824
When a layer is taken out of the cache and initialized it gets cleared
to prepare it for future rendering. This triggers the following sequence
of operations:
glBindFramebuffer(layer.fbo)
attach texture storage to FBO
glClear()
glBindFramebuffer(defaultFbo)
The clear forces a resolve on tilers which stalls the CPU for a little
while, thus producing jank during animations. This change moves the
clear to the next frame when we know we will have to execute a resolve
anyway.
Change-Id: Ic1939c25df20ed65a4c48dc81ee549b2cd8b6ec3
... and check for null returns. This prevents DisplayContent objects
from containing null Display references.
Bug: 7368565 fixed.
Change-Id: I830fb4c1349204c366193657a95a92c48ccee66c
- resolve layout params in ViewGroup when layout direction is changed
- layout param resolution is checking the previous layout direction to
check if we need to resolve
Change-Id: I70af2ad2b4ec83c2ec6c93b3ff445852500d1687
