OSDN Git Service

android-x86/frameworks-base.git
9 years agoMerge "Framework's NetworkSecurityPolicy delegates to libcore's one."
Alex Klyubin [Wed, 25 Mar 2015 18:13:18 +0000 (18:13 +0000)]
Merge "Framework's NetworkSecurityPolicy delegates to libcore's one."

9 years agoMerge "Enforce tag types in KeymasterArgument classes"
Chad Brubaker [Wed, 25 Mar 2015 18:00:48 +0000 (18:00 +0000)]
Merge "Enforce tag types in KeymasterArgument classes"

9 years agoMerge "Fix to crash when clicking text link without view activity"
Alan Viverette [Wed, 25 Mar 2015 16:35:35 +0000 (16:35 +0000)]
Merge "Fix to crash when clicking text link without view activity"

9 years agoMerge "hwui : fix memory leak due to duplicate in shadow cache"
Chris Craik [Wed, 25 Mar 2015 16:15:11 +0000 (16:15 +0000)]
Merge "hwui : fix memory leak due to duplicate in shadow cache"

9 years agoFramework's NetworkSecurityPolicy delegates to libcore's one.
Alex Klyubin [Wed, 25 Mar 2015 16:00:37 +0000 (09:00 -0700)]
Framework's NetworkSecurityPolicy delegates to libcore's one.

Now that the libcore's NetworkSecurityPolicy abstraction is in place,
the framework version of the abstraction should delegate to the
libcore one, to avoid them getting out of sync.

Bug: 19215516
Change-Id: Ic57341d703a13e0fb100dc414958e8fd54e03816

9 years agoMerge "Address pending comments for NetworkSecurityPolicy."
Alex Klyubin [Wed, 25 Mar 2015 15:34:35 +0000 (15:34 +0000)]
Merge "Address pending comments for NetworkSecurityPolicy."

9 years agoMerge "Sort Bluetooth devices in quick settings by name"
Jason Monk [Wed, 25 Mar 2015 14:12:12 +0000 (14:12 +0000)]
Merge "Sort Bluetooth devices in quick settings by name"

9 years agoMerge "systemui: Handle case when network has been lost"
Jason Monk [Tue, 24 Mar 2015 19:57:48 +0000 (19:57 +0000)]
Merge "systemui: Handle case when network has been lost"

9 years agoAddress pending comments for NetworkSecurityPolicy.
Alex Klyubin [Mon, 23 Mar 2015 17:51:20 +0000 (10:51 -0700)]
Address pending comments for NetworkSecurityPolicy.

This is a follow-up to
https://android-review.googlesource.com/#/c/131920/, addressing some
outstanding comments.

This CL also switches the source file to the correct indentation.

Bug: 19215516
Change-Id: Ia48455ab351081e11619afcccadb8fa90340391d

9 years agoMerge "[ActivityManager] Avoid killing unrelated processes."
Olawale Ogunwale [Tue, 24 Mar 2015 16:12:22 +0000 (16:12 +0000)]
Merge "[ActivityManager] Avoid killing unrelated processes."

9 years agoMerge "Layout the window to be displayed if it would be resized"
Olawale Ogunwale [Tue, 24 Mar 2015 16:11:37 +0000 (16:11 +0000)]
Merge "Layout the window to be displayed if it would be resized"

9 years agoMerge "DimLayer : remove unnecessary surface transaction calls"
Olawale Ogunwale [Tue, 24 Mar 2015 16:10:47 +0000 (16:10 +0000)]
Merge "DimLayer : remove unnecessary surface transaction calls"

9 years agoMerge "Frameworks/base: Don't allocate another identity matrix"
Andreas Gampe [Tue, 24 Mar 2015 16:02:53 +0000 (16:02 +0000)]
Merge "Frameworks/base: Don't allocate another identity matrix"

9 years agoSort Bluetooth devices in quick settings by name
Mathias Jeppsson [Mon, 23 Mar 2015 08:00:30 +0000 (09:00 +0100)]
Sort Bluetooth devices in quick settings by name

To avoid Bluetooth devices moving randomly in list, sorting by name.

Change-Id: I4f8e9f98fa29f9670678a3bb6051a6fcf7ae0b9d

9 years agoMerge "Don't create unnecessary RenderThread's instance when executing 'dumpsys gfxinfo'"
John Reck [Tue, 24 Mar 2015 14:44:04 +0000 (14:44 +0000)]
Merge "Don't create unnecessary RenderThread's instance when executing 'dumpsys gfxinfo'"

9 years agoDimLayer : remove unnecessary surface transaction calls
Dohyun Lee [Mon, 23 Mar 2015 00:16:34 +0000 (09:16 +0900)]
DimLayer : remove unnecessary surface transaction calls

There is the case that adjustSurface()  get called even if
the size of the surface of DimLayer is not changed actually.
Since changing the size of a surface is processed synchronously
in the SurfacFlinger, there is usually a few milliseconds delay
(up to 1 vsync interval) when we launch an application.
This patch avoids such cases.

Change-Id: Ib1f76d54f9f2364ac54b70120e4b781e8534e750
Signed-off-by: Dohyun Lee <dohyun.lee@lge.com>
9 years agoMerge "Prevent unexpected rotation while going back to keyguard"
Craig Mautner [Mon, 23 Mar 2015 21:48:55 +0000 (21:48 +0000)]
Merge "Prevent unexpected rotation while going back to keyguard"

9 years agoMerge "[RenderScript] Add create(Context, int) to be compatible with the thunker...
Miao Wang [Mon, 23 Mar 2015 21:01:58 +0000 (21:01 +0000)]
Merge "[RenderScript] Add create(Context, int) to be compatible with the thunker layer & minor tweaks."

9 years ago[RenderScript] Add create(Context, int) to be compatible with
Miao Wang [Mon, 23 Mar 2015 18:09:56 +0000 (11:09 -0700)]
[RenderScript] Add create(Context, int) to be compatible with
the thunker layer & minor tweaks.

bug:19888167
bug:19888339
Change-Id: Ib264c4ca48c990b3476456838047f73d95f6752f

9 years agoMerge "SystemUI needs the SET_WALLPAPER permission"
Michael Wright [Mon, 23 Mar 2015 15:43:23 +0000 (15:43 +0000)]
Merge "SystemUI needs the SET_WALLPAPER permission"

9 years agoSystemUI needs the SET_WALLPAPER permission
Samuel Asteberg [Wed, 25 Feb 2015 14:51:05 +0000 (15:51 +0100)]
SystemUI needs the SET_WALLPAPER permission

If changing wallpaper when low on memory, retrieving the default
wallpaper may return null, which triggers error handling in
ImageWallpaper. This error handling tries to perform
WallpaperManager.clear(), but for that it needs SET_WALLPAPER
permission, which it does not have.

For users with apps that auto change wallpaper, this issue can
be frequent in low-memory conditions.

The solution is to add the permisson.

Change-Id: I81503c1667e3952c2dd15599969f7dcc51623e5b

9 years agoPrevent unexpected rotation while going back to keyguard
tiger_huang [Thu, 26 Feb 2015 06:45:32 +0000 (14:45 +0800)]
Prevent unexpected rotation while going back to keyguard

The original logic would let the app hidden by keyguard be able to
decide the orientation. While going back from a show-when-locked app
to keyguard, there would be a short time that keyguard is unable to
decide the orientation, which causes WMS uses the wrong orientation
from the wrong app.

https://code.google.com/p/android/issues/detail?id=155640

Change-Id: Ibc17bfe4603f68b241dc7380459ec9de42a3e259

9 years agoEnforce tag types in KeymasterArgument classes
Chad Brubaker [Sun, 22 Mar 2015 11:56:46 +0000 (04:56 -0700)]
Enforce tag types in KeymasterArgument classes

Change-Id: Ib1aca884a0b42c3e245db65d48c945b34935072d

9 years agoMerge "Move arrow_pointer hotspot to better match actual tip"
Alan Viverette [Sat, 21 Mar 2015 20:06:56 +0000 (20:06 +0000)]
Merge "Move arrow_pointer hotspot to better match actual tip"

9 years agoMerge "Specify public exponent for RSA"
Chad Brubaker [Sat, 21 Mar 2015 19:18:23 +0000 (19:18 +0000)]
Merge "Specify public exponent for RSA"

9 years agoSpecify public exponent for RSA
Chad Brubaker [Fri, 20 Mar 2015 21:03:49 +0000 (14:03 -0700)]
Specify public exponent for RSA

Keymaster no longer adds a default value, so these tests were failing.

Change-Id: I9c5a8d2552534db1d7fa77965b0d675acc3ccc0a

9 years agoMerge "Add android.security.NetworkSecurityPolicy."
Alex Klyubin [Fri, 20 Mar 2015 17:56:25 +0000 (17:56 +0000)]
Merge "Add android.security.NetworkSecurityPolicy."

9 years agoDon't create unnecessary RenderThread's instance when executing 'dumpsys gfxinfo'
youngmin0822.lee [Fri, 20 Mar 2015 12:22:32 +0000 (21:22 +0900)]
Don't create unnecessary RenderThread's instance when executing 'dumpsys gfxinfo'

To obtain the gfxinfo for each process, the static method of RenderProxy is used, which is named outputLogBuffer().
In there,
1. RenderTask is created for getting DisplayList Commands in RenderNode.
2. staticPostAndWait() is called
3. RenderThread's instance is created by 'RenderThread::getInstance()' in staticPostAndWait()

In case of the service, they don't use HW Acceleration, so don't need RenderThread.
But, by the process of No.3, RenderThread is created for all process.
As we know, RenderThread never be destroyed while the process is alive.
This patch checks RenderThread instance before the creation of RenderTask.
And, there is no one, just return to prevent the unnecessay creation of it.

Change-Id: I4fe29d83c9ced3e8b67177c0874c5d8ee62e1870

9 years agoMerge "New API methods for Script Group"
Yang Ni [Thu, 19 Mar 2015 23:21:27 +0000 (23:21 +0000)]
Merge "New API methods for Script Group"

9 years agoNew API methods for Script Group
Yang Ni [Wed, 11 Mar 2015 23:25:37 +0000 (16:25 -0700)]
New API methods for Script Group

Use a single vararg Object array for both kernel/invoke arguments
and global bindings. The assumption is that global bindings are in
the end after the kernel/invoke arguments.
This simplifies the api to add a kernel or invoke to a script group.

Change-Id: I7efb1035c5967e2cadd15e6fe27f20c80ba197a2

9 years agoMerge "Added copyright notice and formatted to convention"
Yang Ni [Thu, 19 Mar 2015 21:28:13 +0000 (21:28 +0000)]
Merge "Added copyright notice and formatted to convention"

9 years agoMerge "[RenderScript] Remove duplicate include in jni"
Miao Wang [Thu, 19 Mar 2015 20:35:55 +0000 (20:35 +0000)]
Merge "[RenderScript] Remove duplicate include in jni"

9 years agoMerge "Make it harder to leak contexts"
Jason Sams [Thu, 19 Mar 2015 18:38:14 +0000 (18:38 +0000)]
Merge "Make it harder to leak contexts"

9 years agoMerge "Remove unused printapk.cpp."
Elliott Hughes [Thu, 19 Mar 2015 18:06:17 +0000 (18:06 +0000)]
Merge "Remove unused printapk.cpp."

9 years agoAdd android.security.NetworkSecurityPolicy.
Alex Klyubin [Thu, 12 Feb 2015 19:43:09 +0000 (11:43 -0800)]
Add android.security.NetworkSecurityPolicy.

The initial purpose of the NetworkSecurityPolicy class is to provide a
way for network libraries to check whether cleartext network traffic
(e.g., HTTP, WebSockets, XMPP, IMAP, SMTP) should be blocked from this
process.

The policy is set declaratively by the app developer in the app's
manifest and can be queried from ApplicationInfo.flags. Unfortunately,
several network stacks (bundled and unbundled) do not have a reference
to ApplicationInfo or Context.

Alternatives:
* Keep this API hidden (and thus potentially move it from framework to
  libcore), thus precluding unbundled HTTP stacks from using the API.
* Introduce a new java.lang.System property instead of this API.
  However, such properties are a mess and not as powerful/extensible
  as a public class.

Bug: 19215516
Change-Id: If22056a74d257bf1d805ebb4fc284240b3d338f1

9 years agoMerge "Revert "Change jit properties to be debug properties""
Mathieu Chartier [Thu, 19 Mar 2015 16:44:58 +0000 (16:44 +0000)]
Merge "Revert "Change jit properties to be debug properties""

9 years agoMerge "Add app-level usesCleartextTraffic manifest attribute."
Alex Klyubin [Thu, 19 Mar 2015 16:10:20 +0000 (16:10 +0000)]
Merge "Add app-level usesCleartextTraffic manifest attribute."

9 years agoRemove unused printapk.cpp.
Elliott Hughes [Thu, 19 Mar 2015 16:03:41 +0000 (09:03 -0700)]
Remove unused printapk.cpp.

This is one of the few remaining users of libzipfile, but doesn't get built.
Not worth porting to libziparchive.

Change-Id: Ica466b22e3a7ebaf123727c0d703af39cc89cf9c

9 years ago[RenderScript] Remove duplicate include in jni
Miao Wang [Thu, 19 Mar 2015 02:16:47 +0000 (19:16 -0700)]
[RenderScript] Remove duplicate include in jni

Change-Id: Ib6019ab22a83d952dce3fd6f5351120d94b49d58

9 years agoRevert "Change jit properties to be debug properties"
Mathieu Chartier [Thu, 19 Mar 2015 01:54:36 +0000 (18:54 -0700)]
Revert "Change jit properties to be debug properties"

Bug: 19735273

This reverts commit 1ad10ef1e4b83ef9e845761a82e202bb57e2379d.

Change-Id: Iac930925ab6d5cd5947fa4077c8d93e2984927d5

9 years agoMerge "Rename KM_TAG_USER_AUTH_ID to KM_TAG_USER_AUTH_TYPE."
Shawn Willden [Thu, 19 Mar 2015 00:30:30 +0000 (00:30 +0000)]
Merge "Rename KM_TAG_USER_AUTH_ID to KM_TAG_USER_AUTH_TYPE."

9 years agoRename KM_TAG_USER_AUTH_ID to KM_TAG_USER_AUTH_TYPE.
Shawn Willden [Thu, 19 Mar 2015 00:28:39 +0000 (18:28 -0600)]
Rename KM_TAG_USER_AUTH_ID to KM_TAG_USER_AUTH_TYPE.

Also change to an ENUM, to match changes in keymaster_defs.h

Change-Id: Ieda3402563b0b9e4e8b100eb69a94651e51c6347

9 years agoMerge changes Id9a22ad3,I629ab2c4
Shawn Willden [Thu, 19 Mar 2015 00:27:37 +0000 (00:27 +0000)]
Merge changes Id9a22ad3,I629ab2c4

* changes:
  Add KM_TAG_AUTH_TOKEN.
  Add support for KM_TAG_USER_SECURE_ID.

9 years agoMake it harder to leak contexts
Jason Sams [Thu, 19 Mar 2015 00:04:18 +0000 (17:04 -0700)]
Make it harder to leak contexts

Change-Id: Ied60b6428bedb200d7b5b627e8196fbf2b4151e1

9 years agoMerge "Add JIT late override property"
Mathieu Chartier [Wed, 18 Mar 2015 22:56:54 +0000 (22:56 +0000)]
Merge "Add JIT late override property"

9 years agoMerge "Frameworks/base: Fix potential NPE in InputMethod"
Andreas Gampe [Wed, 18 Mar 2015 22:19:58 +0000 (22:19 +0000)]
Merge "Frameworks/base: Fix potential NPE in InputMethod"

9 years agoMerge "Frameworks/base: Remove duplicate check in Mesh"
Andreas Gampe [Wed, 18 Mar 2015 22:19:26 +0000 (22:19 +0000)]
Merge "Frameworks/base: Remove duplicate check in Mesh"

9 years agoMerge "Frameworks/base: Fix visibility flag in Editor"
Andreas Gampe [Wed, 18 Mar 2015 22:19:07 +0000 (22:19 +0000)]
Merge "Frameworks/base: Fix visibility flag in Editor"

9 years agoAdd JIT late override property
Mathieu Chartier [Tue, 17 Mar 2015 16:51:36 +0000 (09:51 -0700)]
Add JIT late override property

Required for JIT debug property since user builds can't restart
the shell.

Bug: 19735273

Change-Id: I1983852e80010c344b45e0bb459de47c56adfce8

9 years agoMerge "Frameworks/base: Fix potential NPE in Script"
Andreas Gampe [Wed, 18 Mar 2015 22:18:25 +0000 (22:18 +0000)]
Merge "Frameworks/base: Fix potential NPE in Script"

9 years agoMerge "logd: optimize statistics"
Mark Salyzyn [Wed, 18 Mar 2015 20:13:34 +0000 (20:13 +0000)]
Merge "logd: optimize statistics"

9 years agoMerge "Avoid duplicate surface creation."
Jason Sams [Wed, 18 Mar 2015 19:44:36 +0000 (19:44 +0000)]
Merge "Avoid duplicate surface creation."

9 years agoFrameworks/base: Fix potential NPE in Script
Andreas Gampe [Wed, 18 Mar 2015 03:05:46 +0000 (20:05 -0700)]
Frameworks/base: Fix potential NPE in Script

Bug: 19797138
Change-Id: I5cd9394bd2c0f7c044c1f2f24665384fa797cd97

9 years agoAdd app-level usesCleartextTraffic manifest attribute.
Alex Klyubin [Wed, 18 Mar 2015 17:05:45 +0000 (10:05 -0700)]
Add app-level usesCleartextTraffic manifest attribute.

The attribute declares whether the app intends to use cleartext
network traffic (e.g., HTTP, WebSockets, XMPP, SMTP, IMAP -- without
TLS or STARTTLS). The default value is true. If set to false, the app
declares that it does not intend to use cleartext network traffic. In
this case the app requests the platform, tooling, and third-party
libraries to prevent it from using cleartext traffic. The danger of
cleartext network traffic is that its confidentiality, authenticity,
and integrity are not guaranteed.

This feature is designed to help apps which care about security of
data exchanged over the network. These apps can accidentally
regress/downgrade to using cleartext network communications. This
typically happens when the server the app communicates with all of a
sudden tells it to use cleartext communications (e.g, HTTP URL
instead of an HTTPS URL) or when one of the components of the app gets
updated and regresses to cleartext communications without the
developer noticing.

In general, the prevention measures are on best effort basis. It's
impossible to automatically prevent all instances of cleartext
traffic. For example, an app bent on bypassing restrictions could
perform low-level network I/O with unusual TCP packet fragmentation,
or could use a custom application-level protocol.

The expectation is that most apps use libraries for network
communications and these libraries over time will start to honor this
flag, thus increasing the protections offered by it.

Bug: 19215516
Change-Id: I8700d51ddbc5d528faa4b6a5fa5bc9551ad02d13

9 years agoFrameworks/base: Fix potential NPE in InputMethod
Andreas Gampe [Wed, 18 Mar 2015 04:07:21 +0000 (21:07 -0700)]
Frameworks/base: Fix potential NPE in InputMethod

Don't read the size of an unchecked list.

Bug: 19797138
Change-Id: I9d8c087aff7bc9cc1e8aae9a0b489e23b5442765

9 years agoFrameworks/base: Remove duplicate check in Mesh
Andreas Gampe [Wed, 18 Mar 2015 02:10:14 +0000 (19:10 -0700)]
Frameworks/base: Remove duplicate check in Mesh

Bug: 19797138
Change-Id: I0b11c4ff63a8031d5e58a06ac13f91ae0bbac5dc

9 years agoAvoid duplicate surface creation.
Jason Sams [Tue, 17 Mar 2015 23:36:55 +0000 (16:36 -0700)]
Avoid duplicate surface creation.

Change-Id: I43104c8b48dd26681735940e6b2e1ba902af2020

9 years agoFrameworks/base: Fix visibility flag in Editor
Andreas Gampe [Tue, 17 Mar 2015 23:08:43 +0000 (16:08 -0700)]
Frameworks/base: Fix visibility flag in Editor

Fix double check.

Bug: 19797138
Change-Id: I95e694f384f1f25d6cf3b6a1669052940385e41d

9 years agoMerge "Frameworks/base: Make IDENTITY_MATRIX final"
Andreas Gampe [Tue, 17 Mar 2015 21:33:08 +0000 (21:33 +0000)]
Merge "Frameworks/base: Make IDENTITY_MATRIX final"

9 years agoFrameworks/base: Make IDENTITY_MATRIX final
Andreas Gampe [Mon, 16 Mar 2015 01:55:33 +0000 (18:55 -0700)]
Frameworks/base: Make IDENTITY_MATRIX final

Bug: 19797138
Change-Id: I127f24b7060a0c4dab401ce8e3057d362c6d6b06

9 years agoMerge "Frameworks/base: Fix format string in Geofence"
Andreas Gampe [Tue, 17 Mar 2015 21:30:49 +0000 (21:30 +0000)]
Merge "Frameworks/base: Fix format string in Geofence"

9 years agoFrameworks/base: Fix format string in Geofence
Andreas Gampe [Sun, 15 Mar 2015 21:49:15 +0000 (14:49 -0700)]
Frameworks/base: Fix format string in Geofence

%p is not a valid conversion in format strings. It is also superfluous,
as it is already known that location is null.

Bug: 19797138
Change-Id: I5784e28b05b4ca9aac57e0fc9da4a7f01d9b3247

9 years agoMerge "Frameworks/base: Fix format string in Camera"
Andreas Gampe [Tue, 17 Mar 2015 20:52:20 +0000 (20:52 +0000)]
Merge "Frameworks/base: Fix format string in Camera"

9 years agoFrameworks/base: Fix format string in Camera
Andreas Gampe [Sun, 15 Mar 2015 21:43:31 +0000 (14:43 -0700)]
Frameworks/base: Fix format string in Camera

One cannot print a boolean with %d. That will result in an exception.

Bug: 19797138
Change-Id: I86c42ea834cebebaecff8463637cc9de14d1fc88

9 years agoMerge "Frameworks/base: Fix precedence bug"
Andreas Gampe [Tue, 17 Mar 2015 20:36:21 +0000 (20:36 +0000)]
Merge "Frameworks/base: Fix precedence bug"

9 years agoFrameworks/base: Fix precedence bug
Andreas Gampe [Sun, 15 Mar 2015 20:59:50 +0000 (13:59 -0700)]
Frameworks/base: Fix precedence bug

Explicit cast has higher precedence than shift.

Bug: 19797138
Change-Id: Ifcf569bf774fbf65ee50c078f736ad167bcc6b8c

9 years agoMerge "Frameworks/base: Use || instead of |"
Andreas Gampe [Tue, 17 Mar 2015 20:28:52 +0000 (20:28 +0000)]
Merge "Frameworks/base: Use || instead of |"

9 years agoFrameworks/base: Use || instead of |
Andreas Gampe [Mon, 16 Mar 2015 03:32:22 +0000 (20:32 -0700)]
Frameworks/base: Use || instead of |

Nothing wrong with | in this case, but || is canonical.

Bug: 19797138
Change-Id: I5f145736a5470f7cde06efce9a217d86eda2135f

9 years agoMerge "Frameworks/base: Fix trivial equals implementation"
Andreas Gampe [Tue, 17 Mar 2015 20:25:20 +0000 (20:25 +0000)]
Merge "Frameworks/base: Fix trivial equals implementation"

9 years agoFrameworks/base: Fix trivial equals implementation
Andreas Gampe [Sun, 15 Mar 2015 22:21:25 +0000 (15:21 -0700)]
Frameworks/base: Fix trivial equals implementation

The comparator's equal implementation doesn't satisfy the constraints
of an equals method, namely being reflexive. Use the standard Object
implementation instead.

Bug: 19797138
Change-Id: I74f888e99533e1945aab7ab10fe8ee3ded6388f4

9 years ago[ActivityManager] Avoid killing unrelated processes.
riddle_hsu [Tue, 17 Mar 2015 10:22:51 +0000 (18:22 +0800)]
[ActivityManager] Avoid killing unrelated processes.

Kernel will reuse process id, when wrap-around happens,
there may be just after a pid is freed, a new thread/process
uses the pid immediately.

For examples, there may be a process with pid 1234 killed
by lowmemorykiller or itself (app.killed=false).
Before the death recipient enters AMS, a new thread/process
is started with pid 1234, then when appDiedLocked executes,
this new pid 1234 will be killed again.

It is especially easy happens to zygote:
During zygote starting process, it will stop 5 daemons threads.
And restart the 5 threads after fork is called, so it has
larger pid range to hit the resue case.
https://code.google.com/p/android/issues/detail?id=160661

Solution:
If the dead event is from binder, it is not necessary
to call killProcessQuiet again, because it should really
be dead.
Keep no checking for killProcessGroup because zygote does
not have record under /acct/, and also the parameter uid
has restriction that will not kill another application
which reuses the pid.

Change-Id: Iec4a4884ae641c4d036f4d024ce463f7a351a17b

9 years agohwui : fix memory leak due to duplicate in shadow cache
Mykola Kondratenko [Thu, 12 Mar 2015 14:20:38 +0000 (15:20 +0100)]
hwui : fix memory leak due to duplicate in shadow cache

New ShadowTask with the ShadowDescription key that already
exists in the shadow LruCache will leak as it is not being
added.

Fix adds check for the existing key that is common in the hwui
code but missing for the TessellationCache::precacheShadow
function.

Change-Id: I37fd5ec82f8b8da5d1ec0f2ab9fd04c5f8534367

9 years agoMerge "Change jit properties to be debug properties"
Mathieu Chartier [Tue, 17 Mar 2015 02:04:33 +0000 (02:04 +0000)]
Merge "Change jit properties to be debug properties"

9 years agoMerge "Frameworks/base: Fix a comparison"
Andreas Gampe [Tue, 17 Mar 2015 00:12:24 +0000 (00:12 +0000)]
Merge "Frameworks/base: Fix a comparison"

9 years agoChange jit properties to be debug properties
Mathieu Chartier [Mon, 16 Mar 2015 23:10:52 +0000 (16:10 -0700)]
Change jit properties to be debug properties

Bug: 19735273
Change-Id: I266cf447a7af0595366bf89109d2823094f5cf41

9 years agoAdded copyright notice and formatted to convention
Yang Ni [Mon, 16 Mar 2015 22:53:18 +0000 (15:53 -0700)]
Added copyright notice and formatted to convention

Pure format change. No other changes at all.

Change-Id: I74a53aa67ef0960f70becaf288b85be40066a0ae

9 years agoMerge "Frameworks/base: Force long computation"
Andreas Gampe [Mon, 16 Mar 2015 20:36:04 +0000 (20:36 +0000)]
Merge "Frameworks/base: Force long computation"

9 years agoMerge "Frameworks/base: Check before foreach in Script"
Andreas Gampe [Mon, 16 Mar 2015 19:17:44 +0000 (19:17 +0000)]
Merge "Frameworks/base: Check before foreach in Script"

9 years agoMerge "Frameworks/base: Use equals for Integer comparison"
Andreas Gampe [Mon, 16 Mar 2015 19:10:39 +0000 (19:10 +0000)]
Merge "Frameworks/base: Use equals for Integer comparison"

9 years agoMerge "Revert "Frameworks/base: Fix a constructor""
Andreas Gampe [Mon, 16 Mar 2015 17:51:14 +0000 (17:51 +0000)]
Merge "Revert "Frameworks/base: Fix a constructor""

9 years agoRevert "Frameworks/base: Fix a constructor"
Andreas Gampe [Mon, 16 Mar 2015 17:51:01 +0000 (17:51 +0000)]
Revert "Frameworks/base: Fix a constructor"

The constructor was public API, doh. Gotta do this differently.

This reverts commit 33c5b2a62f3e62382c41e24c6b527119978816a0.

Change-Id: Iadca87fe6a8866a8bd9d6f2a91578ec0d4c44691

9 years agoMerge "Frameworks/base: Fix null-pointer check"
Andreas Gampe [Mon, 16 Mar 2015 16:59:13 +0000 (16:59 +0000)]
Merge "Frameworks/base: Fix null-pointer check"

9 years agoMerge "Frameworks/base: Fix a constructor"
Andreas Gampe [Mon, 16 Mar 2015 16:51:00 +0000 (16:51 +0000)]
Merge "Frameworks/base: Fix a constructor"

9 years agoMerge "Frameworks/base: Fix a hashCode implementation"
Robert Greenwalt [Mon, 16 Mar 2015 16:17:50 +0000 (16:17 +0000)]
Merge "Frameworks/base: Fix a hashCode implementation"

9 years agoMerge "Frameworks/base: Fix always-false equals"
Andreas Gampe [Mon, 16 Mar 2015 15:28:52 +0000 (15:28 +0000)]
Merge "Frameworks/base: Fix always-false equals"

9 years agoMerge "Frameworks/base: Fix missing cast"
John Reck [Mon, 16 Mar 2015 14:56:51 +0000 (14:56 +0000)]
Merge "Frameworks/base: Fix missing cast"

9 years agoFrameworks/base: Force long computation
Andreas Gampe [Mon, 16 Mar 2015 03:17:07 +0000 (20:17 -0700)]
Frameworks/base: Force long computation

Ensure that an int-based computation is carried out as long.

Change-Id: I23b10a95600674e8a5a65c0ea349afdc6aa152ae

9 years agoFrameworks/base: Don't allocate another identity matrix
Andreas Gampe [Mon, 16 Mar 2015 02:03:29 +0000 (19:03 -0700)]
Frameworks/base: Don't allocate another identity matrix

There is already an identity matrix in Matrix. Don't allocate another
one for VectorDrawable.

Change-Id: I51735f262d6680e043b0009707ec42acb2d0d1ad

9 years agoFrameworks/base: Fix a comparison
Andreas Gampe [Mon, 16 Mar 2015 01:04:41 +0000 (18:04 -0700)]
Frameworks/base: Fix a comparison

Change-Id: I80d62869920e77110c95f20369ec2631c75f6ed4

9 years agoFrameworks/base: Check before foreach in Script
Andreas Gampe [Sun, 15 Mar 2015 22:57:30 +0000 (15:57 -0700)]
Frameworks/base: Check before foreach in Script

According to the if below, ains == null is potentially valid. But
the foreach loop would throw a NullPointerException.

Change-Id: I4460fb1357eaa3abfe0ab9a21effb608f474ab51

9 years agoFrameworks/base: Use equals for Integer comparison
Andreas Gampe [Sun, 15 Mar 2015 22:39:21 +0000 (15:39 -0700)]
Frameworks/base: Use equals for Integer comparison

Integer == is dangerous, as equal objects may not be identical
objects. In fact, MediaFormat.setInteger was creating a new object
every time.

Change MediaFormat.setInteger and setLong to use valueOf, which
may reuse returned objects.

Change-Id: Iedcc6003adbf05c0c870aa4b3ada7f181a5b870e

9 years agoFrameworks/base: Fix null-pointer check
Andreas Gampe [Sun, 15 Mar 2015 22:07:19 +0000 (15:07 -0700)]
Frameworks/base: Fix null-pointer check

Change-Id: I715a21c313e909ae654e0c1aa67bdf7bcd89de76

9 years agoFrameworks/base: Fix a hashCode implementation
Andreas Gampe [Sun, 15 Mar 2015 21:38:59 +0000 (14:38 -0700)]
Frameworks/base: Fix a hashCode implementation

Equals uses Arrays.equals. That means two responses are equal if
the content of the data arrays is equal. By convention, the hash
code of those objects should agree. In that case one cannot use
hashCode on the array (which is the identity hash code).

Change-Id: Icce8e2e71e9142421f5dac8a0ee8a211623fb704

9 years agoFrameworks/base: Fix a constructor
Andreas Gampe [Sun, 15 Mar 2015 21:29:16 +0000 (14:29 -0700)]
Frameworks/base: Fix a constructor

This was meant to be a constructor, according to the comment.

Change-Id: Ief49011b392e58b37d9acb4a3f754f1828b256af

9 years agoFrameworks/base: Fix always-false equals
Andreas Gampe [Sun, 15 Mar 2015 21:19:43 +0000 (14:19 -0700)]
Frameworks/base: Fix always-false equals

Rect != Insets.

Change-Id: I3d4ff890608e446b51f09a1b633af742f0c069d4

9 years agoFrameworks/base: Fix missing cast
Andreas Gampe [Sun, 15 Mar 2015 21:10:23 +0000 (14:10 -0700)]
Frameworks/base: Fix missing cast

Without a cast, the division is integer division.

Change-Id: I050e53778de8b1591a0be16ebbee8eed70eb1528

9 years agoAdd KM_TAG_AUTH_TOKEN.
Shawn Willden [Sat, 14 Mar 2015 16:51:56 +0000 (10:51 -0600)]
Add KM_TAG_AUTH_TOKEN.

Change-Id: Id9a22ad32137f3e0380c2812f790bbecab511d11

9 years agoAdd support for KM_TAG_USER_SECURE_ID.
Shawn Willden [Sat, 14 Mar 2015 14:35:57 +0000 (08:35 -0600)]
Add support for KM_TAG_USER_SECURE_ID.

Change-Id: I629ab2c47ee6d42de20a963ef283e330364c8ee7

9 years agoMerge "Switch libjnigraphics to sysv-only hash style"
Dmitriy Ivanov [Fri, 13 Mar 2015 21:35:03 +0000 (21:35 +0000)]
Merge "Switch libjnigraphics to sysv-only hash style"

9 years agoSwitch libjnigraphics to sysv-only hash style
Dmitriy Ivanov [Fri, 13 Mar 2015 19:49:47 +0000 (12:49 -0700)]
Switch libjnigraphics to sysv-only hash style

Bug: 19059885
Change-Id: I3474974983931c67dce30dfc581a8d28d8b62e6a