Removes > 1s of GC time and hundreds of GC during device boot.
Zygote size before: 8388608
Zygote size after: 8388608
No increase in zygote space size due to zygote compaction.
Bug: 16398684
Change-Id: I3c6f5a0b26a207ab39bb0a952400d7488e42f00c
Add a simple internal wrapper for dealing with window callback events.
This should also make working with things like action modes from
ToolbarActionBar significantly easier in later changes.
Bug 16980711
Change-Id: I2c3d7525489a21847f4af9b80f6ebf9b34b4dd58
Add a new setting to persist whether to show the IME when a hard
keyboard is connected.
Bug: 14066881
Change-Id: I2237ded850a0d4ab43ca441d0b7df13e0958e630
Previously it implied that the hardware keyboard would be disabled,
but really the toggle would just enable showing the IME even if a
hardware keyboard was present. Changed the string and swapped the
semantics to be more clear about the behavior.
Bug: 14066881
Change-Id: I9c8a7eb98b5277f1d09cc19fa7402e9b4cf51d92
There is a bug in how we deal with name overflows combined with resetting
the battery stats data. If we do a reset while a wakelock is being
actively held that has been put into the overflow bucket, then we can
end up reducing the number of known wake locks in the list so when after
that it is released we try to release it under its real name rather than
the overflow name.
This means we need to keep track of which wake locks have been placed
in the overflow bucket while they are actively being used, so we can be
sure to properly handle it as part of that bucket until it is eventually
released.
This makes things... somewhat more complicated. So now we have a class
to take care of all these details, and also use it for other places where
we have the same overflow semantics sync and job stats.
Also fix potential deadlock -- BatteryStatsHelper needs to call on to
ConnectivityManager to find out of there is telepohny, however we use
that class when doing a dump while the battery stats lock is held. To
fix this, we check the connectivity state up in the battery stats service
before acquiring the lock and propagate that information through to the
dump code.
Change-Id: Ib452206af5c36f4b0f03cc94d2845d36613d1ba5
Move location selection logic into shared PackageHelper location,
and share it between DCS and PackageInstaller. Fix bugs related to
installed footprint calculation; always count unpacked native libs.
Have PMS do its own threshold checking, since it's fine to stat
devices. PMS only ever deleted staging ASECs, so move that logic
into installer and nuke unclaimed staging ASECs. Allocate legacy
ASEC names using PackageInstaller to make sure they don't conflict
with sessions.
Start wiring up session to allocate ASEC and pass through staged
container for installation.
Fix bug to actually delete invalid cluster-style installs.
Bug: 16514385
Change-Id: I325e0c4422fc128398c921ba45fd73ecf05fc2a9
...with grantUriPermissions (no user interaction required)
Add a new path in to the activity manager to start an activity as
if it was directy started by the original calling activity. This
is specifically for the resolver activity and chooser activity to
be able to safely launch its data after serving as an intermediary.
Access to the new method is highly restricted -- it can only be
called by an activity that is declared in the framework apk itself,
and the execute-as-the-caller behavior will only happen if the
code is running under the system uid. (This means we could still
have these run in the client's process in some cases and still work
correctly.)
Note there is some commented out code here half-done about trying
to propagate security exceptions back to the original calling
activity. This would be really nice, especially now with the
chooser activity running in a system process so any errors made
by the app (bad permission grants, bad intents, etc) no longer
actually appear in the app so are essentially invisible. I'd
really like to figure out a way to propagate these exceptions back
to the app, but this is hard since the app's process may no
longer even be running at this point.
Also tweak activity manager dump output to split the recents
dump out from activities, since recents can now be super large.
Change-Id: I50410c4783faf9302c69290589a068a846e0973a
This reverts commit 87efe74e092236c372d3b6909009641123aa416a.
This should be fine now with all the dependency CLs +2-ed
Change-Id: I96ad14ad5ff81e6b5391035cb6c5a62339c6cc40
We set the system_server classpath in the environment
(like we do with BOOTCLASSPATH). After the zygote forks
the system_server, we dexopt the classpath (if needed)
and then launch the system server with the correct
PathClassLoader. This needed several small / medium
refactorings :
- The logic for connecting to installd is now in a separate
class and belongs in the system_server.
- SystemService / SystemServiceManager have now moved to
classes.jar. They are only used from there, and since they
use Class.forName, we want them to be loaded by the
system_server classloader, and not the bootclassloader.
- BootReceiver now moves to frameworks.jar, because it is
used by ActivityThread and friends.
bug: 16555230
Change-Id: Ic84f0b2baf611eeedff6d123cb7191bb0259e600
- Get rid of ActivityManager.dismissKeyguardOnNextActivity, which was
used for two different things: Dismiss keyguard from somewhere else
(not really necessary anymore), wait to actually dismiss keyguard
after the window behind is drawn. Instead, introduce
keyguardWaitingForActivityDrawn(), and change the semantics where
necessary.
- Make wallpaper_close_enter consistent with task_open_enter and the
Keyguard launch animation.
- Close the panel even on lockscreen when launching a notification.
- Block notification shade updates during the collapsing motion so
notification don't play the disappear animation immediately after
having launched a notification.
Bug: 15991916
Change-Id: I133c177b84e926c87c1a404ba93d633593fec3ab
Some devices use clatd for catching raw IPv4 traffic when running on
a pure-IPv6 carrier network. In those situations, the per-UID
stats are accounted against the clat iface, so framework users need
to combine both the "base" and "stacked" iface usage together.
This also means that policy rules (like restricting background data
or battery saver) need to apply to the stacked ifaces.
Finally, we need to massage stats data slightly:
-- Currently xt_qtaguid double-counts the clatd traffic *leaving*
the device; both against the original UID on the clat iface, and
against UID 0 on the final egress interface.
-- All clatd traffic *arriving* at the device is missing the extra
IPv6 packet header overhead when accounted against the final UID.
Bug: 12249687, 15459248, 16296564
Change-Id: I0ee59d96831f52782de7a980e4cce9b061902fff
