On some devices, the USB "set configuration" command is propogated to the framework
after the "start accessory" event is received. However, on other devices like
the 2011 ADK board, no "set configuration" command is sent at all until we have
reenumerated in acccessory mode. To fix the original problem without breaking other
devices, we can simply remove assumptions about if or when "set configuration"
will be received. Now we simply remain switch USB configuration to accessory mode
when we receive the "start accessory" command, and remain there until the existing
10 second timeout expires.
Bug: 13393825
Change-Id: I4c9e3423185bd7252a907e4568d9e3ff06044b7d
This reverts commit fbd5521fb5e94fad066e21b9f91f4782afe71660.
This change broke support for the 2011 ADK board, which never sends a "set configuration"
command before "accessory start". So we revert this change and will replace it with a better fix.
Bug: 13535051
Bug: 13393825
Change-Id: Icd870d7ff6daff1567e04d93907f70f5d7e37884
In certain situations it was possible for a task to move to the top
in activity manager but not in window manager. This resulted in
the task appearing behind the launcher icons.
Fixes bug 13410184.
Change-Id: If0582b395e126a8aff70a0e4c64b731083c6ae8a
This CL adds a system service handling HDMI-CEC protocol. The service
is equipped with the capability sending/receiving HDMI-CEC messages
Not all the messages are in place. Currently it has messages to support
a few features only, as follows:
- One touch play
- System information
- Routing control (partially - active source status maintenance only)
- Device OSD transfer
- Power status
It will be extended to cover the wider usages in the follow up CLs.
The CEC standard version referenced in the implementation is 1.3a.
Change-Id: Ifed0b02f52ebf098eddb3bd0987efbf353b7e8fe
This fixes a bug where an allocated DeathRecipient in WindowManagerService
was holding a reference to keyguard binder interface after a call to
linkToDeath() without a matchin unlinkToDeath().
It was causing the keyguard side of the binder interface to stick around,
which in tern prevented the keyguard side from releasing its references.
The solution is to ensure matching linkToDeath()/unlinkToDeath() calls.
Fixes bug 11982048
Change-Id: I6959816b819ba953512c53675162195cbf1e0653
Add a new hidden method to recovery system to trigger a factory reset
followed by a shutdown, rather than a reboot back to the regular
system. Use this flag when the MASTER_CLEAR intent is broadcast with
a boolean extra "shutdown" set to true.
Change-Id: I2ceb25b715d9f0ef492a75b3d287d1e17cec89ef
This is a start and two tests succeed:
Tested expired AT&T SIM and waiting 15min for alarm to fire.
Tested a provisioned Verizon SIM and works normally.
I've NOT tested AT&T where I've properly completed the provisioning.
I've NOT tested T-Mobile SIM either provisioned or not-provisioned.
I've NOT tested provisioning over WiFi.
I've NOT tested that WiFi <-> Mobile works
I've NOT tested voice calls, SMS, MMS
...
The current bug is below, but it is poorly named either it should be
renamed or a new bug created.
Bug: 13190133
Change-Id: I0a09f642614cd27a8655e9dae764b8999ce485b8
Adapted from change I44ce250e, "Maintain binding to crashed
listeners from NotificationManagerService.":
> This fixes the logic on the death handlers for notification
> listeners, and doesn't unbind from the listener services so
> that the system will bring them back up again.
Bug: 12587702
Change-Id: I41d0c6733046ee9760bcfae4407e754592b805ea
A VPN can once again bring up a new tun interface while the old tun is
running. Once the new tun is set up the routing rules will be removed from the
old tun. It is up to the application to drain the old tun of traffic and
close it.
If the new tun fails to come up the old tun will remain untouched and
can still be used.
To prevent leakage the new rules are added before the old tun is
shutdown. Netd/Dns has been changed to allow multiple rules to exist
at once with the most recently added rule taking priority.
Bug: 12134439
Change-Id: I7e00c7c68cc339d81f09b3d2a33276ffc76985f5
With netd allowing overlapping rules for uid range rules the interface
name is needed to make sure only the correct rule is removed.
Bug: 12134439
Change-Id: I94f77f154f49ca8d5f6cf49683a4473cc92c3eb7
When Intent.FLAG_ACTIVITY_REORDER_TO_FRONT was set the TaskRecord
member frontOfTask was being set true incorrectly for the top
activity. It should only be true for the bottom activity. This fix
ensures that frontOfTask is always set correctly for all activities by
consoldating it into one method.
Fixes bug 12171535.
Change-Id: If982dad3c81b2b816adc5d89e7e0496923098a70
For managing the ui mode, consider the device to be a television if it
has either the FEATURE_TELEVISION or FEATURE_LEANBACK system features.
When leanback is available for non-television devices, we need to
reconsider the use of these features for setting the television ui mode.
Change-Id: I7b82372ce19067e010be9b42111816bab186f95b
The value for the TCP initial receive window comes from,
in order,
kernel
/proc/sys/net/ipv4/tcp_default_init_rwnd
init.rc (via properties)
net.tcp.default_init_rwnd
properties
net.tcp.default_init_rwnd
gservices
Settings.Global.TCP_DEFAULT_INIT_RWND
Bug: 12020135
Change-Id: I0e271be19472900fa9f3bab037d53383ec014a9e
This is a cherry-pick of https://googleplex-android-review.git.corp.google.com/#/c/399886/
Instead of storing a kb layout per device descriptor (which is expected
to be unique), store it for each vendor/product. This way we can keep
a consistent layout between identical but physically different keyboards.
There are some corner cases this is expected to fail on, namely devices
that incorrectly have the same vendor/product id. Devices that don't
define a vendor/product id will continue to use the descriptor to store
layout files.
Change-Id: I1f2508561992080459310d5a644dad65a9c24f1a
The activity manager needs to set launchedFromPackage to be that of
the previous package in the case where flow has been redirected
through an intermediate activity.
Change-Id: I678fc2e7d984991ac715251a784ba7d7ccbf9fca
Under certain circumstances, the power manager might continue to
hold the display wakelock long after the display had been turned
off due to the mDisplayReady flag having an incorrect value.
1. An inverted conditional caused DisplayPowerState to incorrectly
signal the screen on ready state.
2. The DisplayPowerController failed to clear the block screen on
flag in the case where the screen was turned off before it became
unblocked from turning on. This could happen when the display was
rapidly turned on-off-on-off.
Bug: 13248135
Change-Id: I8faa3034695c83c8cd35613d81acccf40d22128d
Under certain circumstances, the power manager might continue to
hold the display wakelock long after the display had been turned
off due to the mDisplayReady flag having an incorrect value.
1. An inverted conditional caused DisplayPowerState to incorrectly
signal the screen on ready state.
2. The DisplayPowerController failed to clear the block screen on
flag in the case where the screen was turned off before it became
unblocked from turning on. This could happen when the display was
rapidly turned on-off-on-off.
Bug: 13248135
Change-Id: I8faa3034695c83c8cd35613d81acccf40d22128d
The window manager and view hierarchy currently disable all drawing
when PowerManager.isScreenOn() returns false so no drawing occurs
while dozing. This will be fixed in a future patch to take the
display blanking state into account correctly.
This patch is a workaround to unblock development in the meantime.
Bug: 13133142
Change-Id: I2dc0b422c77285e657d73adca2606aa68264d987
We'd otherwise leave the data dirs & native libraries
lying around. This will leave the app permanently broken
because the next install of the app will fail with
INSTALL_FAILED_UID_CHANGED.
Also remove an unnecessary instance variable.
Cherry-pick from master
Bug 13416059
Change-Id: I1e644aab74d5ea519231800915b39c2f55d043ae
