...in cases involving uninstalled apps, or apps whose install state
varies across different users.
Bug 17398315
Change-Id: I7297d82f8bf5d49c50a7fd53d795a706bf2d2313
NativeDaemonEvent.unescapeArgs() was improperly skipping the terminating
quote in cases like "\\", where the char preceding the quote is a
backslash, but the backslash itself is escaped, so the quote is indeed
an unescaped terminator.
unescapeArgs() doesn't unescape "\xxx" decimal escapes used by mDNS, so
fix NsdService to do that sort of unescaping explicitly (which is only
applicable when it receives a "fullname", in SERVICE_RESOLVED).
Bug: 16983542
Bug: 16986203
Change-Id: Idfa79749336c68424d961bc414f984c525b7e5e6
This way the audio system can route the adjustments correctly when a bt
headset is connected.
bug:17281977
Change-Id: Ic41d038e47179baffc86e539562da285446148c7
For the system restore use-case, an installer may need to enqueue
their sessions quickly before badging details, like icons, have been
downloaded. This change relaxes to allow an installer to update
their session badging after the session has been created. Notify
observers when badging changes.
Rename callback registration methods to match style guide. Relax
constraint that observers are home app. Fix bug around internal
progress reporting.
Bug: 17376797, 17389236, 17334199
Change-Id: I5fb88508baea2f08e89a1504fcf5ef972afad4a7
We need to allow for event dispatching in non-interactive states so
that we can enable a richer set of interactions when a device is
dozing (i.e. is in a low power state with an Always-on-Display).
Bug: 17167296
Change-Id: I8ae0f544a8106cb91ff38c2309b8b57cbe2f2c72
The remote print spooler can only be accessed off the main thread
by design as calls into it may block for a short amoun of time since
the frist call into the spooler may have to wait for the system to
bind to the spooler service. A recent change introduced a regression
where the user state changes are now dispatched on the main thread.
This change schedules the user change handling to a background handler
thread.
bug:17396682
Change-Id: I059be8f31ba41122cb2967d8afadd19e6b5a08e5
The existing recursion in resumeTopActivityLocked() finally caused
a problem by attempting to pause the same activity twice. By
preventing the recursion with this change we no longer face that
problem.
Fixes bug 17260463.
Change-Id: I0954614fc1a7dc0eeeef9335c1b973a42a7fc345
This change fixes the following cases.
- <Report Power Status>(0x90)
- When TV receives <Report Power Status> as a reply of <Give Device Power Status>, TV does not send <Feature Abort>. But if a device sends <Report Power Status> actively, TV sends <Feature Abort>.
- <Set System Audio Mode>(0x72)
- <System Audio Mode Status>(0x7E) Directly address message is also defined.
- <Record Status>(0x0A) (We will support One-touch record function at least in Japan)
- <Timer Status>(0x35) (We will support Timer programming function at least in Japan)
- No response for <Record TV Screen>(0x0F)
Bug: 17382769
Change-Id: I8ae355337757710d54b788edfdf37293f96cfa97
Here is a list of changes
1. Change volume control into event base not level base
2. Hide volume ui if volume change is triggered by CEC
3. Report volume change triggered by CEC in OSD message
4. Revamp HdmiLogger so that normal class uses static method only.
5. Apply format message to HdmiLogger's helper methods.
Bug: 17367215
Change-Id: I9f3cd41f7c66f76919059b463df956ed5176b054
- Hide APIs not approved by API council (b/17189780)
- Change configOverrides in MMS APIs from ContentValues to Bundle to be
consistent (b/17390017)
- Change SmsManager.MMS_EXTRA_DATA to EXTRA_MMS_DATA per API council
review (b/17390977)
b/17189780
b/17390017
b/17390977
Change-Id: If5a43e9ab9e42b6328f1581b39051ddf1a58f0dd
Use enforceCallingOrSelfPermission() for throwing a security exception
rather than creating the security exception ourselves. This has the
advantage that the exception is in a well known format expected by
the CTS test android.security.cts.ServicePermissionsTest#testDumpProtected
Bug: 17165920
Change-Id: I7d3c2d2a0852a490fe1d553cd81d1efc79d2fec3
The call to RecoverySystem.rebootWipeUserData() was made while
holding the lock to DevicePolicyManagerService. But it blocks
waiting for system_process' main thread to receive the ordered
broadcast complete callback. It won't receive that callback
because Keyguard is running on the main thread and is concurrently
blocked on DevicePolicyManagerService.
By moving the call to rebootWipeUserData() out of the synchronized
block the deadlock is eliminated.
Fixes bug 16870054.
Change-Id: I3eb587211e5484859cc9dab7e80e5a1f6c85225d
