This attempts to reland previously reverted CLs [1][2] due to an
unexpected regression (Bug 27824691).
The Bug 27868748 we want to address by this CL is that currently
InputConnection#finishComposingText() can be called on the root view's
Handler no matter what Handler is associated with
ControlledInputConnectionWrapper. Actually the root cause of
Bug 6789252 is the same, but there we worked around it by not calling
InputConnection#finishComposingText() in certain situations [3].
With this CL we should be able to logically revert that workaround.
This CL also removes redundant IMM#mServedInputConnection. This is safe
because the following two fields have the same lifetime.
- InputMethodManager#mServedInputConnection
- InputMethodManager#mServedInputConnectionWrapper
We do not need to maintain both of them. This also allows us to use a
strong refecente in IInputConnectionWrapper#mInputConnection instead of
a WeakReference. To understand why this is safe, we need to understand
how things previously worked, which is as follows:
1. InputMethodManager#mServedInputConnection becomes non-null.
-> IInputConnectionWrapper#mInputConnection.get() is guaranteed to
be alive.
2. InputMethodManager#mServedInputConnection becomes null or another
object.
-> IInputConnectionWrapper#mInputConnection.get() may not be alive.
Since we know exactly when InputMethodManager#mServedInputConnection is
updated, in theory we do not need to use WeakReference here, and
with this CL we do not use WeakReference anymore. Actually the initial
commit [1] accidentally removed the last strong reference to the active
InputConnection and WeakReference could be null at any time, which was
what we observed in Bug 27824691.
[1]: I1181e067aa5bedbdf0c7ec1bcec479257aea511c
afb6558c8f
[2]: Ibe94f115e607a198d12ecd3d4e4f91a7d9469c98
16e2c7b59a
[3]: I66f51da1299532793ef8fa700f35b0811670f235
4e5184f929
Bug: 27868748
Change-Id: If2a03bc84d318775fd4a197fa43acde086eda442
Adds documentation, renames Layout to WindowLayout and
splits #minimalSize to #minimalWidth and #minimalHeight.
Bug: 27528326
Change-Id: Idb440cb081a14ccdc83309284e906454633c4504
The purpose of the new StorageVolume API is to grant access to
volumes that aren't typically "visible" to a developer, so include
them in the returned results.
Also return the real mounted state instead of augmenting based on
the caller's storage permissions. Clean up API naming slightly and
return as List.
Bug: 27615770
Change-Id: Ida921a4b91e5af81e418e76f672d9108f45a9781
Now that CE data isn't available until after a user is unlocked, we
need to delay the PRE_BOOT_COMPLETED broadcasts. This is done by
adding a new RUNNING_UNLOCKING user state to the UserController
lifecycle.
We now track the last fingerprint a user was logged in under, and we
dispatch PRE_BOOT receivers when that fingerprint changes. To work
around battery pull issues, we only persist the updated fingerprint
once all PRE_BOOT receivers have finished. This is less granular
than the original solution, but it's still correct. We only consider
a user as "logged in" once it transitions into the RUNNING_UNLOCKED
state.
When starting a process, track if the user was "unlocked" when
started, so that we only spin up unaware providers in processes
started before user unlock.
Add generic IProgressListener to communicate PRE_BOOT progress and
strings up to lock screen. For now, LockSettingsService just blocks
until finished, but it could display these strings in the future.
Bug: 27220885
Change-Id: I349439776b885acd32f6a578d8951ffd95640be2
am: a04c532
* commit 'a04c532a09b8d946ebc9a086f673220059218869':
Implement transition for docking task in recents #6
Implement transition for docking task in recents #5
Implement transition for docking task in recents #4
Implement transition for docking task in recents #3
Implement transition for docking task in recents #2
Implement transition for docking task in recents #1
Show a scrim activity if task is not resizable
am: d0ee17d
* commit 'd0ee17d926cce445bcd22b15e80e725aa4639895':
DO NOT MERGE Revert "Revert "DO NOT MERGE Extend the DownloadManager.addCompletedDownload() method to support origin details.""
* changes:
Implement transition for docking task in recents #6
Implement transition for docking task in recents #5
Implement transition for docking task in recents #4
Implement transition for docking task in recents #3
Implement transition for docking task in recents #2
Implement transition for docking task in recents #1
Show a scrim activity if task is not resizable
- Use a future to provide the app thumbnail so the app can restart
in parallel when recents draws the bitmap (extremely expensive).
- Don't call startRecents from AM when recents is already running - this
messes up the transition information.
- Make sure to put the task into resizing mode if it needs to be restored
from the disk.
- Some minor fixes for the transition animation spec.
- Add NO_MOVE_ANIMATION to recents flags to prevent wallpaper
flickering.
Bug: 27607141
Change-Id: I7d0c75b88775ab467927b8cf94303ddb60222e7f
This pruns all the stored trusted issuers so that changes to the system
or user CA store are detected. Currently this is only exposed as a
TestApi, but it can be hooked up to the trusted storage change event
in a future commit.
Bug: 27526668
Change-Id: Ic426254babab9a3177c968bc05b45e95eaac1fdd
- When the docking transition is happening, defer updating
the bounds of the home stack until the transition is done.
This is to preserve the scrim which is drawn in the recents
activity.
- Use the PROLONG_AT_START infrastructure to hide the task
in recents when starting the app transition.
- When recents finally get resized at the end of the transition,
reset it's draw state so we don't move the old surface around,
and the new surface gets drawn at the new position, to avoid
flickering.
- Remove hack around not layouting docked divider if it's not
visible, it's not needed anymore and resulted in a wrong
initial position.
- Fix animation selection for docked stack divider.
- Make sure win.moved() always gets called.
Bug: 27607141
Change-Id: I76c35f09461f044a90e2c88335008284b5839cc0
Add a callback to TaskStackChangeListener which gets fired when the system
might need to inform the user that a specific app might not work in
multi-window.
Use that callback in SysUI to show a translucent activity which scrims the
activity behind to inform that it might not be resizable.
Debounce the information to once per multi-window session, to not make it
annoying.
Introduce launchTaskId to start an activity in an existing task, and protect
that with START_TASKS_FROM_RECENTS permission.
Bug: 27327287
Bug: 27431869
Change-Id: I89e8d653872ab01ba3c1e252b426e5481da0e6ca
Netd provides 2 bandwidth control rules to restrict which uids can use
metered networks:
- bw_penalty_box is a blacklist-based firewall chain used to determine
which uids do not have access to metered interfaces.
- bw_happy_box is whitelist-based firewall chain used to determine which
uids have access to metered interfaces.
Currently, both NetworkManagerService (NMS) and
NetworkPolicyManagerService (NPMS) uses just the bw_penalty_box rule,
which makes turning Data Saver mode on / off too slow (since NPMS needs
to build the bw_penalty_box on demand); this CL adds support for both
rules on NMS, although NPMS doesn't take advantage of it yet (it will be
refactored in a separate CL).
BUG: 27127112
BUG: 26685616
Change-Id: Ib954574f7c86269fc9b4cf8ce4ba72ba5878c23d
There's a few advantages to having ApfFilter in IpManager:
1. If things go wrong, crashing a particular transport is less bad then
crashing ConnectivityService. We also don't want to use
ConnectivityService as a dumping ground for transport-specific logic.
2. This makes implementing WifiManager.MulticastLock a lot simpler and
safer because enabling/disabling it doesn't have to go through the
NetworkAgent, which could risk various races (e.g. installing a filter
into the wrong WiFi network).
3. IpManager is the ultimate source for LinkProperties for a particular
transport and since ApfFilter uses the LinkProperties it's better to
have it closely paired with the IpManager. Likewise, ApfFilter needs
to know the APF capabilities of the transport, so having it in
the transport avoids having to parcel this information through the
NetworkAgent.
Bug: 26238573
Change-Id: I99b85f2b64972f0e7572170ec5d1926081aa3429
