The HeartbeatHandler for the System Server Watchdog has been running
on the wrong thread due to a race condition in initialization. It's
designed to run on ServerThread, so that it can catch lockups in the
main looper of the System Server. It has been running on
ActivityManagerThread instead, so it does not detect lockups on the
ServerThread as it should.
ActivityManagerService is calling Watchdog.getInstance() before
ServerThread calls Watchdog.getInstance().init(), so the handler is
being bound to the ActivityManagerThread instead of the ServerThread.
Explicitly bind HeartbeatHandler to ServerThread, so that the Watchdog
catches lockups on this critical thread.
Change-Id: Iccb184ac3adb817feb86ed4ee0e50e443bf74636
Don't automatically grant all normal/dangerous permissions. Instead,
check the value of requestedPermissionsRequired to see if it's required.
If the permission is not required, then only grant it if the permission
was previously granted to the application.
Change-Id: I86b1fae530c006d353f9fa22137598bc88253805
The root cause is:
There is a defect in window manager service: When a new
activity that can be ime target is added into window manager
but the Z order of input method window don't need to be
changed, then the target app token of input method window
would not be updated to new one. This defect may cause that
the layer of input method window is calculated incorrectly.
The solution:
Correct the target app token for input method window.
Change-Id: I008311e3c9b1cf5fc320b614d8675c183c506d50
Currently, NetworkManagementService only catches RemoteExceptions
when calling the BaseNetworkObserver notification methods (e.g.,
interfaceStatusChanged). However, if the observer is in the same
process, unchecked exceptions can occur as well.
When this happens, finishBroadcast does not get called, and no
further notifications can be sent, because any attempt to do so
fails with a "beginBroadcast() called while already in a
broadcast" exception.
Fix this by catching RuntimeException as well.
Bug: 8397534
Bug: 8276725
Change-Id: Icd6f32128707244978943c48a9ea3a2b952a2957
BadTokenException is a normal consequence of swapping IMEs while there
is a DO_SHOW_SOFT_INPUT message in the IIMethodWrapper queue. This
race condition cannot be avoided without an unacceptable lock down of
InputMethodManagerService.
Fixes bug 8387663.
Fixes bug 8263462.
Change-Id: I2c21573cf972145ab08e66604cdb9344139a3f31
1. Add a Nat464Xlat service that ConnectivityService can use
to start and stop clat. When clat is started, the service
waits for the clat interface to come up and then calls
ConnectivityService to add the appropriate routes.
2. Make ConnectivityService start clat when an IPv6-only mobile
interface is connected. We only support clat on mobile for
now.
3. Make tethering use the interface that has the IPv4 default
route insted of using the base interface of the
LinkProperties. This allows us to tether to a stacked
interface, which is needed for tethering with 464xlat.
Bug: 8276725
Change-Id: I24480af69ee280f504399062638af0836a56268e
Currently, ConnectivityService adds and removes routes to/from
the routing table only based on the LinkProperties's routes.
Make it update routes based on the stacked links as well.
Bug: 8276725
Change-Id: I9a2adf537af5a04de0aaab3780afbcc3bb5c6acb
This reverts commit 6f210bd0191e1936bbc1f036912c6efc4ea69475
Mako wouldn't boot for me, reverting for now.
Change-Id: Ie92d6bf77811e7257e86d65e1e15e1973c027cd7
Modify WifiService to add a controller to track the various
desired states and let the WifiStatemachine actually control
the bring up.
Bug: 8141918
Change-Id: I6e98fd5a29b43c3c50c315eff5255cd0a3eaebcd
When a new IME is attached it is not enough to remove the
WindowManager messages from the local queue, but the ones in
the previous IME queue must also be removed.
Fixes bug 8263462.
Change-Id: I9e916c6052a83dc7691bcba0b6ab8328b9b7cc36
The previous show/hide messages in the queue were still trying
to be honored even after a new ime was attached.
Fixes bug 8263462.
Change-Id: Iee60dbd1d58542f73aedeac5ccb54cddeb5d5dfe
You can now declare shared libraries in apks that are
on the system image. This is like the existing mechanism
of using raw jar files as shared libraries, but since they
are contained in an apk the library can actually be updated
from the Play Store. And this even (mostly) works.
There are some deliberate limitations on this feature. A
new shared library *must* be declared by an apk on the system
image. Installing an update to a system image apk does not
allow you to add new shared libraries; they must be defined
by everything on the base system image. This allows us to
get rid of a lot of ugly edge cases (shared libraries that were
there disappearing after an update is uninstalled for example)
and give some brakes on apps that happen to be pre-installed
on devices from being able to throw in new shared libraries
after the fact.
In working on this, I ran into a recently introduced bug where
uninstalling updated to system apps would fail. This was done
to allow for the new restricted users that don't have all
system apps, but conflicts with the existing semantics for
uninstalling system apps. To fix this I added a new uninstall
flag that lets you switch on the new mode if desired.
Also to implement the desired logic for limitations on declaring
new shared libraries in app updates, I needed to slightly tweak
the initial boot to keep the Package object for hidden system
packages associated with their PackageSetting, so we can look at
it to determine which shared libraries are allowed. I think
this is probably more right than it was before -- we already
need to parse the package anyway, so we have it, and when you
install an update to a system app we are in this same state
until you reboot anyway.
And having this fixed also allowed me to fix another bug where
we wouldn't grant a new permission to an updated app if its
system image version is updated to request the permission but
its version is still older than whatever is currently installed
as an update. So that's good.
Also add new sample code showing the implementation of an apk
shared library and a client app using it.
Change-Id: I8ccca8f3c3bffd036c5968e22bd7f8a73e69be22
modifyRoute takes both an interface name and a LinkProperties.
This is redundant because all callers get the interface name
from the LinkProperties. Make modifyRoute get the interface
name from the LinkProperties instead.
Change-Id: I41ba8e0a10241c2f1107204fcaca2be74556042b
OBB backup/ restore is no longer handled within the target app
process. This is done to avoid having to require that OBB-using
apps have full read/write permission for external storage.
The new OBB backup service is a new component running in the
same app as the already-existing shared storage backup agent.
The backup infrastructure delegates backup/restore of apps'
OBB contents to this component (because the system process
may not itself read/write external storage).
From the command line, OBB backup is enabled by using new
-obb / -noobb flags with adb backup. The default is noobb.
Finally, a couple of nit fixes:
- buffer-size mismatch between the writer and reader of chunked
file data has been corrected; now the reading side won't be
issuing an extra pipe read per chunk.
- bu now explicitly closes the transport socket fd after
adopting it. This was benign but triggered a logged
warning about leaked fds.
(Cherrypicked)
Change-Id: I471f6348abcccb7bf1e1710b7beda9f23de53e14
The Activity Manager was not properly informing the observer that
the operation had concluded (unsuccessfully).
Bug 8222595
Change-Id: I8234e32d8edf4112c8c7a5e20e341d0b41e23014
By only adding notifications to the archive when they are
removed we batch up multiple updates and only store the
final version. Some data is lost in this process, but we
save tons of memory storing otherwise redundant /
uninteresting data (e.g. each step of a download).
Change-Id: I008afefc1242bb7c433d45da2c36fcc626dd3706
Listeners should be notified for any notification if they
register for USER_ALL, or for any notification posted to
USER_ALL.
Bug: 8328357
Change-Id: Ib5024d41287090d1a390539a015d8cb4dfa854a7
PackageParser.updateApplicationInfo() has already interpreted the
various COMPONENT_ENABLED flags for us, no need to clobber them.
Bug: 8331767
Change-Id: If1363c5651a2f0326ee60e92517cfc0e6f256699
