OSDN Git Service

android-x86/frameworks-native.git
10 years agoUpdate GLES codegen to be in sync with frameworks/base
Ashok Bhat [Thu, 13 Feb 2014 11:08:25 +0000 (11:08 +0000)]
Update GLES codegen to be in sync with frameworks/base

Patch from synergydev with commit-id 01994c79 changed
core/jni/com_google_android_gles_jni_GLImpl.cpp in the
frameworks/base. However the corresponding code snippet
in frameworks/native was not updated. This patch addresses
the issue.

Change-Id: I89c3d88de832efc8cbbe682c8a7ff7dfbef99db3
Signed-off-by: Ashok Bhat <ashok.bhat@arm.com>
10 years agoMerge "libagl: support multilib build"
Colin Cross [Tue, 18 Feb 2014 21:29:50 +0000 (21:29 +0000)]
Merge "libagl: support multilib build"

10 years agoMerge "Use /(vendor|system)/lib64/egl for 64-bit EGL libraries"
Colin Cross [Tue, 18 Feb 2014 20:14:41 +0000 (20:14 +0000)]
Merge "Use /(vendor|system)/lib64/egl for 64-bit EGL libraries"

10 years agoUse /(vendor|system)/lib64/egl for 64-bit EGL libraries
Dan Willemsen [Sun, 16 Feb 2014 18:23:54 +0000 (10:23 -0800)]
Use /(vendor|system)/lib64/egl for 64-bit EGL libraries

Change-Id: Ie64806bfe4ba60a3167b84685cf983d4ada1042d

10 years agolibagl: support multilib build
Colin Cross [Thu, 13 Feb 2014 01:15:53 +0000 (17:15 -0800)]
libagl: support multilib build

Use LOCAL_SRC_FILES_arch and LOCAL_CFLAGS_arch to support building
for multiple architectures.

Change-Id: I829c1bea880739bf8184df4006d0a481ffccc5c2

10 years agoMerge "frameworks/native: move idmap to frameworks/base"
Dianne Hackborn [Fri, 14 Feb 2014 01:45:27 +0000 (01:45 +0000)]
Merge "frameworks/native: move idmap to frameworks/base"

10 years agoMerge "Convert all selinux_android_restorecon and _setfilecon calls to new API."
Nick Kralevich [Wed, 12 Feb 2014 23:09:44 +0000 (23:09 +0000)]
Merge "Convert all selinux_android_restorecon and _setfilecon calls to new API."

10 years agoframeworks/native: move idmap to frameworks/base
Colin Cross [Wed, 12 Feb 2014 02:04:03 +0000 (18:04 -0800)]
frameworks/native: move idmap to frameworks/base

idmap depends on libandroidfw, so it should go in frameworks/base.

Change-Id: Ieee30ba9d8170d3118acc89f0617981228a6974b

10 years agoMerge "Runtime resource overlay, iteration 2"
Dianne Hackborn [Tue, 11 Feb 2014 21:27:35 +0000 (21:27 +0000)]
Merge "Runtime resource overlay, iteration 2"

10 years agoMerge "New command line tool 'idmap'"
Dianne Hackborn [Tue, 11 Feb 2014 20:29:54 +0000 (20:29 +0000)]
Merge "New command line tool 'idmap'"

10 years agoConvert all selinux_android_restorecon and _setfilecon calls to new API.
Stephen Smalley [Fri, 7 Feb 2014 14:16:46 +0000 (09:16 -0500)]
Convert all selinux_android_restorecon and _setfilecon calls to new API.

libselinux selinux_android_restorecon API is changing to the more
general interface with flags and dropping the older variants.

Also get rid of the old, no longer used selinux_android_setfilecon API
and rename selinux_android_setfilecon2 to it as it is the only API in use.

Change-Id: I1e71ec398ccdc24cac4ec76f1b858d0f680f4925
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
10 years agoMerge "binder: fix all warnings"
Colin Cross [Thu, 6 Feb 2014 02:06:42 +0000 (02:06 +0000)]
Merge "binder: fix all warnings"

10 years agobinder: fix all warnings
Colin Cross [Thu, 6 Feb 2014 01:42:44 +0000 (17:42 -0800)]
binder: fix all warnings

Fix warnings related to casting pointers to ints, using %d or %ld to
print size_t/ssize_t, and unused parameters.

Change-Id: I7a13ba83d402952989c1f795cd9e880a95b98d9e

10 years agoMerge "Fixed apparent copy/paste mistake Bug: 11634373"
Andrew Hsieh [Wed, 5 Feb 2014 07:16:30 +0000 (07:16 +0000)]
Merge "Fixed apparent copy/paste mistake Bug: 11634373"

10 years agoFixed apparent copy/paste mistake
David Friedman [Wed, 5 Feb 2014 03:11:33 +0000 (11:11 +0800)]
Fixed apparent copy/paste mistake
Bug: 11634373

Change-Id: I9b0ac4d3b4a9a2999b2013abd39465a062ec51ea

10 years agoMerge "GLES_trace: fix compile errors on 64 bit architectures"
Siva Velusamy [Tue, 4 Feb 2014 19:14:06 +0000 (19:14 +0000)]
Merge "GLES_trace: fix compile errors on 64 bit architectures"

10 years agoMerge "Use 64 bit binder interface if TARGET_USES_64_BIT_BINDER _or_ TARGET_IS_64_BIT...
Arve Hjønnevåg [Tue, 4 Feb 2014 00:48:12 +0000 (00:48 +0000)]
Merge "Use 64 bit binder interface if TARGET_USES_64_BIT_BINDER _or_ TARGET_IS_64_BIT is set."

10 years agoUse 64 bit binder interface if TARGET_USES_64_BIT_BINDER _or_ TARGET_IS_64_BIT is...
Arve Hjønnevåg [Mon, 3 Feb 2014 22:46:30 +0000 (14:46 -0800)]
Use 64 bit binder interface if TARGET_USES_64_BIT_BINDER _or_ TARGET_IS_64_BIT is set.

Change-Id: I46a29939fa29ec9421ab6dff3285502f69a31745

10 years agoMerge changes Ib0e5a037,I1bd7c38e,Icfc67c2a,I96c64312,I59528054, ...
Arve Hjønnevåg [Mon, 3 Feb 2014 21:12:57 +0000 (21:12 +0000)]
Merge changes Ib0e5a037,I1bd7c38e,Icfc67c2a,I96c64312,I59528054, ...

* changes:
  ServiceManager: Implement PING_TRANSACTION
  ServiceManager: Use 32/64 bit types from new binder header
  Binder: Use 64 bit pointers in 32 processes if selected by the target
  Add BINDER_IPC_32BIT to CFLAGS unless TARGET_USES_64_BIT_BINDER is true
  Binder: Make binder portable
  ServiceManager: Fix the binder interface
  ServiceManager: Store handles in uint32_t instead of void *
  ServiceManager: Generic Fixes
  ServiceManager: Add extra error handling
  ServiceManager: Fix Android.mk
  ServiceManager: Make use of kernel exported structures

10 years agoRuntime resource overlay, iteration 2
Mårten Kongstad [Fri, 31 Jan 2014 13:42:59 +0000 (14:42 +0100)]
Runtime resource overlay, iteration 2

Support any number of overlay packages. Support any target package.

UPDATED PACKAGE MATCHING
------------------------
In Runtime resource overlay, iteration 1, only a single overlay package
was considered. Package matching was based on file paths:
/vendor/overlay/system/framework-res.apk corresponded to
/system/framework-res.apk. Introduce a more flexible matching scheme
where any package is an overlay package if its manifest includes

    <overlay targetPackage="com.target.package"/>

For security reasons, an overlay package must fulfill certain criteria
to take effect: see below.

THE IDMAP TOOL AND IDMAP FILES
------------------------------
Idmap files are created by the 'idmap' binary; idmap files must be
present when loading packages. For the Android system, Zygote calls
'idmap' as part of the resource pre-loading. For application packages,
'idmap' is invoked via 'installd' during package installation (similar
to 'dexopt').

UPDATED FLOW
------------
The following is an outline of the start-up sequences for the Android
system and Android apps. Steps marked with '+' are introduced by this
commit.

Zygote initialization
   Initial AssetManager object created
+    idmap --scan creates idmaps for overlays targeting 'android', \
           stores list of overlays in /data/resource-cache/overlays.list
   AssetManager caches framework-res.apk
+  AssetManager caches overlay packages listed in overlays.list

Android boot
   New AssetManager's ResTable acquired
     AssetManager re-uses cached framework-res.apk
+    AssetManager re-uses cached 'android' overlays (if any)

App boot
   ActivityThread prepares AssetManager to load app.apk
+  ActivityThread prepares AssetManager to load app overlays (if any)
   New AssetManager's ResTable acquired as per Android boot

SECURITY
--------
Overlay packages are required to be pre-loaded (in /vendor/overlay).
These packages are trusted by definition. A future iteration of runtime
resource overlay may add support for downloaded overlays, which would
likely require target and overlay signatures match for the overlay to
be trusted.

LOOKUP PRIORITY
---------------
During resource lookup, packages are sequentially queried to provide a
best match, given the constraints of the current configuration. If any
package provide a better match than what has been found so far, it
replaces the previous match. The target package is always queried last.

When loading a package with more than one overlay, the order in which
the overlays are added become significant if several packages overlay
the same resource.

Had downloaded overlays been supported, the install time could have been
used to determine the load order. Regardless, for pre-installed
overlays, the install time is randomly determined by the order in which
the Package Manager locates the packages during initial boot. To support
a well-defined order, pre-installed overlay packages are expected to
define an additional 'priority' attribute in their <overlay> tags:

    <overlay targetPackage="com.target.package" priority="1234"/>

Pre-installed overlays are loaded in order of their priority attributes,
sorted in ascending order.

Assigning the same priority to several overlays targeting the same base
package leads to undefined behaviour. It is the responsibility of the
vendor to avoid this.

The following example shows the ResTable and PackageGroups after loading
an application and two overlays. The resource lookup framework will
query the packages in the order C, B, A.

        +------+------+-     -+------+------+
        | 0x01 |      |  ...  |      | 0x7f |
        +------+------+-     -+------+------+
            |                           |
        "android"                Target package A
                                        |
                       Pre-installed overlay B (priority 1)
                                        |
                       Pre-installed overlay C (priority 2)

Change-Id: If49c963149369b1957f7d2303b3dd27f669ed24e

10 years agoNew command line tool 'idmap'
Mårten Kongstad [Fri, 31 Jan 2014 13:00:54 +0000 (14:00 +0100)]
New command line tool 'idmap'

Introduce a new tool 'idmap' to handle generation and verification of
idmap files. The tool is modelled on 'dexopt', and is intended to be
used similarly, notably by 'installd'.
See cmds/idmap/idmap.cpp for further documentation on 'idmap'.

Note: this commit is interdependent on a commit in project build/ to add
'idmap' to PRODUCT_PACKAGES.

Note: the changes to androidfw are only stubs. The actual implementation
will be provided in Runtime resource overlay, iteration 2.

Change-Id: I7131b74ece1e46c8a9c0a31d103e686aa07da2bb

10 years agoServiceManager: Implement PING_TRANSACTION
Arve Hjønnevåg [Wed, 29 Jan 2014 05:35:03 +0000 (21:35 -0800)]
ServiceManager: Implement PING_TRANSACTION

Stop printing "invalid id " to stderr every time a process tries to
connect to the servicemanager.

Change-Id: Ib0e5a0375bfa2dec2c2f9cd668bd5dda46ed6588

10 years agoServiceManager: Use 32/64 bit types from new binder header
Arve Hjønnevåg [Wed, 29 Jan 2014 05:25:12 +0000 (21:25 -0800)]
ServiceManager: Use 32/64 bit types from new binder header

Change-Id: I1bd7c38ed9f43125cf9c63aa533434ee7ca06f80

10 years agoBinder: Use 64 bit pointers in 32 processes if selected by the target
Arve Hjønnevåg [Wed, 29 Jan 2014 04:12:59 +0000 (20:12 -0800)]
Binder: Use 64 bit pointers in 32 processes if selected by the target

Uses new kernel header where void * has been replaced by binder_uintptr_t

Change-Id: Icfc67c2a279269f700343bd9246fd7cb94efe2c1

10 years agoAdd BINDER_IPC_32BIT to CFLAGS unless TARGET_USES_64_BIT_BINDER is true
Arve Hjønnevåg [Wed, 29 Jan 2014 03:57:02 +0000 (19:57 -0800)]
Add BINDER_IPC_32BIT to CFLAGS unless TARGET_USES_64_BIT_BINDER is true

Change-Id: I96c643123b0314c361b7f48a18d5c22c660d4ff5

10 years agoBinder: Make binder portable
Serban Constantinescu [Tue, 5 Nov 2013 16:53:55 +0000 (16:53 +0000)]
Binder: Make binder portable

Changes include
- Binder attempts to cast pointers to a int datatype
  which is not sufficient on a 64-bit platform.

- This patch introduces new read/write functions into
  Parcel that allow pointers to be written using the
  uintptr_t datatype for compile-time data type size
  selection.

-  Change access specifier for the methods above.

-  Binder uses the 64bit android_atomic_release_cas64
   (aka cmpxchg)

Change-Id: I595280541e0ba1d19c94b2ca2127bf9d96efabf1
Signed-off-by: Matthew Leach <matthew.leach@arm.com>
Signed-off-by: Serban Constantinescu <serban.constantinescu@arm.com>
10 years agoServiceManager: Fix the binder interface
Serban Constantinescu [Thu, 19 Dec 2013 09:11:41 +0000 (09:11 +0000)]
ServiceManager: Fix the binder interface

This patch adds support for binder transactions on 64bit systems
without breaking the existing 32bit ABI. It has been tested on
the Android emulator and ARMv8 Model.

Most of the changes in this patch just follow the binder ABI.

Change-Id: I8c37b847ea65008d56554d34d4696fe3d22f7533
Signed-off-by: Serban Constantinescu <serban.constantinescu@arm.com>
10 years agoServiceManager: Store handles in uint32_t instead of void *
Serban Constantinescu [Thu, 30 Jan 2014 14:07:34 +0000 (14:07 +0000)]
ServiceManager: Store handles in uint32_t instead of void *

This patch corrects the types used for storing handles.

Change-Id: If9c10782345f1de9e12b4b3fd6be9e02e6b568cd
Signed-off-by: Serban Constantinescu <serban.constantinescu@arm.com>
10 years agoServiceManager: Generic Fixes
Serban Constantinescu [Fri, 10 Jan 2014 15:48:29 +0000 (15:48 +0000)]
ServiceManager: Generic Fixes

This patch fixes some of the ServiceManager issues. The following patches
of the series add fixes to the ABI.

Change-Id: Ib479234c8704e12592f1b149ddec67881bc50230
Signed-off-by: Serban Constantinescu <serban.constantinescu@arm.com>
10 years agoServiceManager: Add extra error handling
Serban Constantinescu [Thu, 30 Jan 2014 15:16:45 +0000 (15:16 +0000)]
ServiceManager: Add extra error handling

This patch extends the error handling. It also adds a check for a matching
binder version - kernel/userspace.

Change-Id: I43a262934b38c5711536aaa42754fed1ef04b39e
Signed-off-by: Serban Constantinescu <serban.constantinescu@arm.com>
10 years agoServiceManager: Fix Android.mk
Serban Constantinescu [Thu, 30 Jan 2014 14:58:32 +0000 (14:58 +0000)]
ServiceManager: Fix Android.mk

This patch fixes Android.mk and enables building bctest as an optional
module without any extra hacks.

Change-Id: Icaf8bf9452776db2ea4a2ba75f3abf05b4e2cdab
Signed-off-by: Serban Constantinescu <serban.constantinescu@arm.com>
10 years agoServiceManager: Make use of kernel exported structures
Serban Constantinescu [Fri, 10 Jan 2014 13:56:27 +0000 (13:56 +0000)]
ServiceManager: Make use of kernel exported structures

This patch switches ServiceManager to use the structures exported in the
kernel headers rather then redefining its own.

struct binder_txn is replaced with struct binder_transaction_data and struct
binder_object with struct flat_binder_object, both defined in the binder driver
header <linux/binder.h>.

Change-Id: I3b3e97918173ea35a289e184774ae06193192da3
Signed-off-by: Serban Constantinescu <serban.constantinescu@arm.com>
10 years agoMerge "frameworks/native: Rename persist.sys.dalvik.vm.lib to allow new default"
Brian Carlstrom [Thu, 30 Jan 2014 21:50:56 +0000 (21:50 +0000)]
Merge "frameworks/native: Rename persist.sys.dalvik.vm.lib to allow new default"

10 years agoframeworks/native: Rename persist.sys.dalvik.vm.lib to allow new default
Brian Carlstrom [Thu, 30 Jan 2014 21:14:01 +0000 (13:14 -0800)]
frameworks/native: Rename persist.sys.dalvik.vm.lib to allow new default

Bug: 12798969
Change-Id: I6b40317eceb3d89b0acff88238a9a9ab423c4d78

10 years agoGLES_trace: fix compile errors on 64 bit architectures
Siva Velusamy [Tue, 28 Jan 2014 22:08:46 +0000 (14:08 -0800)]
GLES_trace: fix compile errors on 64 bit architectures

GLES_trace passes some pointers back to the host. Encode them
using int64's rather than int32's

This fix requires an update to the host tool to look at pointers
as 64 bit, and if not found, then fall back to 32 bit integers.

Change-Id: I11ff2ca51290f05913c1b9143cecee1caf015543

10 years agoMerge "SurfaceFlinger: mFbProducerSlot initailization"
Jesse Hall [Tue, 28 Jan 2014 18:17:46 +0000 (18:17 +0000)]
Merge "SurfaceFlinger: mFbProducerSlot initailization"

10 years agoSurfaceFlinger: mFbProducerSlot initailization
mayank parshar [Thu, 23 Jan 2014 09:38:31 +0000 (15:08 +0530)]
SurfaceFlinger: mFbProducerSlot initailization

b/12487813

SurfaceFlinger crash is observed during simulation of
Secondary display

Note: change 14e8b01a761180250671f3d6494532d8bd04fa8e
removed the initialization leading to the crash when
simulating secondary display.  Restore the initialization
to solve the problem.

Change-Id: Iae5845fb82735e01de5cc0dc582d13c27e3c614f
Signed-off-by: mayank parshar <mayankp@broadcom.com>
Signed-off-by: Pierre Couillaud <pierre@broadcom.com>
10 years agoMerge "Make libEGL build cleanly under 64-bit"
Colin Cross [Tue, 28 Jan 2014 02:22:11 +0000 (02:22 +0000)]
Merge "Make libEGL build cleanly under 64-bit"

10 years agoMake libEGL build cleanly under 64-bit
Dan Stoza [Fri, 17 Jan 2014 19:38:34 +0000 (11:38 -0800)]
Make libEGL build cleanly under 64-bit

Change-Id: I255dfb70b166b7469d59352b1acfc7aececa07de

10 years agoMerge "frameworks/native: convert LOCAL_MODULE_PATH to LOCAL_MODULE_RELATIVE_PATH"
Colin Cross [Tue, 28 Jan 2014 00:48:31 +0000 (00:48 +0000)]
Merge "frameworks/native: convert LOCAL_MODULE_PATH to LOCAL_MODULE_RELATIVE_PATH"

10 years agoframeworks/native: convert LOCAL_MODULE_PATH to LOCAL_MODULE_RELATIVE_PATH
Colin Cross [Sat, 25 Jan 2014 05:00:32 +0000 (21:00 -0800)]
frameworks/native: convert LOCAL_MODULE_PATH to LOCAL_MODULE_RELATIVE_PATH

LOCAL_MODULE_PATH doesn't work for multiarch builds, replace it
with LOCAL_MODULE_RELATIVE_PATH.

Change-Id: I4e4ceec61d026bbe74ba604554c06104bde42e5e

10 years agoMerge "SurfaceFlinger: mFbFence initialization"
Jesse Hall [Sun, 26 Jan 2014 08:15:15 +0000 (08:15 +0000)]
Merge "SurfaceFlinger: mFbFence initialization"

10 years agoSurfaceFlinger: mFbFence initialization
mayank parshar [Fri, 10 Jan 2014 04:57:45 +0000 (10:27 +0530)]
SurfaceFlinger: mFbFence initialization

MobC00383030
b/12487813

SurfaceFlinger crash is observed while connecting
to Wi-Fi display.

Note: change 14e8b01a761180250671f3d6494532d8bd04fa8e
removed the initialization leading to the crash when
running through the HWC composition path.  Restore the
initialization to solve the problem.

Change-Id: I581defc7135ac512080c0da06a62b1dae7d218c4
Signed-off-by: mayank parshar <mayankp@broadcom.com>
Signed-off-by: Pierre Couillaud <pierre@broadcom.com>
10 years agoMerge "libGLES_android: fix 64-bit compile errors"
Colin Cross [Sat, 25 Jan 2014 00:38:15 +0000 (00:38 +0000)]
Merge "libGLES_android: fix 64-bit compile errors"

10 years agolibGLES_android: fix 64-bit compile errors
Colin Cross [Fri, 24 Jan 2014 22:35:39 +0000 (14:35 -0800)]
libGLES_android: fix 64-bit compile errors

Fix size of vertex_t on 64-bit architectures
Fix __get_tls casts, the TLS area is a void**
Cast through uintptr_t to store integers in a pointer
Use %zu to print size_t

Change-Id: I91079ec76025237e6d2081784c5348518b0ce122

10 years agoMerge "Implement per-display EGLConfig and configless EGLContext"
Jesse Hall [Fri, 3 Jan 2014 18:25:07 +0000 (18:25 +0000)]
Merge "Implement per-display EGLConfig and configless EGLContext"

10 years agoMerge "Move EGLConfig selection to RenderEngine"
Jesse Hall [Fri, 3 Jan 2014 18:24:30 +0000 (18:24 +0000)]
Merge "Move EGLConfig selection to RenderEngine"

10 years agoImplement per-display EGLConfig and configless EGLContext
Jesse Hall [Tue, 24 Dec 2013 05:02:15 +0000 (21:02 -0800)]
Implement per-display EGLConfig and configless EGLContext

Bug: 12230666
Change-Id: Icca608b108cbdcab9cf01a9236d8cdbda000a836
Signed-off-by: Jesse Hall <jessehall@google.com>
10 years agoMove EGLConfig selection to RenderEngine
Jesse Hall [Tue, 24 Dec 2013 04:44:38 +0000 (20:44 -0800)]
Move EGLConfig selection to RenderEngine

Bug: 12230666
Change-Id: I8d1111a7e0fd9d9e2525e6a80da8ce46d7dd085d
Signed-off-by: Jesse Hall <jessehall@google.com>
10 years agoMerge "Track Looper decoupling from ALooper"
Brian Carlstrom [Thu, 19 Dec 2013 07:13:03 +0000 (07:13 +0000)]
Merge "Track Looper decoupling from ALooper"

10 years agoMerge "restorecon /data/media when re-creating it for multi-user upgrade."
Nick Kralevich [Tue, 17 Dec 2013 23:02:36 +0000 (23:02 +0000)]
Merge "restorecon /data/media when re-creating it for multi-user upgrade."

10 years agorestorecon /data/media when re-creating it for multi-user upgrade.
Stephen Smalley [Tue, 17 Dec 2013 21:04:20 +0000 (16:04 -0500)]
restorecon /data/media when re-creating it for multi-user upgrade.

Change-Id: I112c61863f2104d1962697c54cff25106e9b48a0
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
10 years agoTrack Looper decoupling from ALooper
Brian Carlstrom [Fri, 13 Dec 2013 07:13:18 +0000 (23:13 -0800)]
Track Looper decoupling from ALooper

Change-Id: I612f0de7002e24948a4cbbea7b416b4fbb3929aa

10 years agoMerge commit '1ee45c2d2d816ee912624d81ff1b0aaa98e8fa74' into HEAD
The Android Open Source Project [Thu, 5 Dec 2013 20:38:18 +0000 (12:38 -0800)]
Merge commit '1ee45c2d2d816ee912624d81ff1b0aaa98e8fa74' into HEAD

10 years agoMerge "DispSync: remove delay when enabling vsync events"
Jesse Hall [Tue, 3 Dec 2013 05:12:31 +0000 (05:12 +0000)]
Merge "DispSync: remove delay when enabling vsync events"

10 years agoDispSync: remove delay when enabling vsync events
Jamie Gennis [Tue, 29 Oct 2013 20:36:12 +0000 (13:36 -0700)]
DispSync: remove delay when enabling vsync events

This change fixes a bug that caused an extra frame of latency when enabling
vsync event callbacks in DispSync.  The bug was related to the logic that
prevents the two events from firing with very little time between them due to
updates to the vsync model.

Bug: 11479720
Change-Id: Ie7eaff9e92ffb7b7b6cb4d3d4402c96cbd29af7e

10 years agoMerge "Handle policy reloads within installd rather than restarting it."
Nick Kralevich [Wed, 27 Nov 2013 21:09:12 +0000 (21:09 +0000)]
Merge "Handle policy reloads within installd rather than restarting it."

10 years agoam 66ad04bc: Merge "Remove incompatible ashmem.h"
Dima Zavin [Sat, 23 Nov 2013 04:45:07 +0000 (20:45 -0800)]
am 66ad04bc: Merge "Remove incompatible ashmem.h"

* commit '66ad04bc039c73763c55c8d3aad282378e05d257':
  Remove incompatible ashmem.h

10 years agoMerge "Remove incompatible ashmem.h"
Dima Zavin [Sat, 23 Nov 2013 04:41:17 +0000 (04:41 +0000)]
Merge "Remove incompatible ashmem.h"

10 years agoMerge commit 'afd0debe4bdf47dc0f968282ca1261842bb65d60' into HEAD
The Android Open Source Project [Fri, 22 Nov 2013 19:20:20 +0000 (11:20 -0800)]
Merge commit 'afd0debe4bdf47dc0f968282ca1261842bb65d60' into HEAD

Change-Id: Ic9aed2c57bdfeb250ce2dd7f0e9289ba5b6e936a

10 years agoMerge "Remove references to obsolete pmem functionality."
Elliott Hughes [Thu, 21 Nov 2013 20:42:32 +0000 (20:42 +0000)]
Merge "Remove references to obsolete pmem functionality."

10 years agoRemove references to obsolete pmem functionality.
Elliott Hughes [Thu, 21 Nov 2013 20:22:39 +0000 (12:22 -0800)]
Remove references to obsolete pmem functionality.

Change-Id: I5d3befd075e51614e6801115388a0a9f32fbb6ea

10 years agoam 14e8b01a: Don\'t change the framebuffer target until we render a new one
Jesse Hall [Fri, 8 Nov 2013 20:58:20 +0000 (12:58 -0800)]
am 14e8b01a: Don\'t change the framebuffer target until we render a new one

* commit '14e8b01a761180250671f3d6494532d8bd04fa8e':
  Don't change the framebuffer target until we render a new one

10 years agoDon't change the framebuffer target until we render a new one
Jesse Hall [Thu, 7 Nov 2013 20:28:26 +0000 (12:28 -0800)]
Don't change the framebuffer target until we render a new one

Continuing to send the last-rendered framebuffer to HWC on subsequent
frames allows the HWC to read partially-composed regions that haven't
changed, instead of re-composing from scratch.

Bug: 11573910
Change-Id: I8829877d2a06001f1e1b3f168cbba71c7b217b2d

10 years agoam 356c2386: Merge "Provide virtual display output buffer to HWC in prepare" into...
Jesse Hall [Thu, 7 Nov 2013 04:48:51 +0000 (20:48 -0800)]
am 356c2386: Merge "Provide virtual display output buffer to HWC in prepare" into klp-dev

* commit '356c23865d588659f1eea55bbe7e2b16b9cff08e':
  Provide virtual display output buffer to HWC in prepare

10 years agoMerge "Provide virtual display output buffer to HWC in prepare" into klp-dev
Jesse Hall [Thu, 7 Nov 2013 04:43:19 +0000 (04:43 +0000)]
Merge "Provide virtual display output buffer to HWC in prepare" into klp-dev

10 years agoam 49318631: Merge "Don\'t use implementation-defined format with CPU consumers"...
Jesse Hall [Wed, 6 Nov 2013 15:38:36 +0000 (07:38 -0800)]
am 49318631: Merge "Don\'t use implementation-defined format with CPU consumers" into klp-dev

* commit '493186312a9a8c0215da0d6d3a9b0369491c0796':
  Don't use implementation-defined format with CPU consumers

10 years agoMerge "Don't use implementation-defined format with CPU consumers" into klp-dev
Jesse Hall [Wed, 6 Nov 2013 15:35:08 +0000 (15:35 +0000)]
Merge "Don't use implementation-defined format with CPU consumers" into klp-dev

10 years agoProvide virtual display output buffer to HWC in prepare
Jesse Hall [Wed, 6 Nov 2013 00:12:31 +0000 (16:12 -0800)]
Provide virtual display output buffer to HWC in prepare

We were already making sure the HWComposer class had the handle before
prepare, but it wasn't passing the handle along to HWC as intended.

Partial fix for bug: 11430248
Change-Id: I25f672c4fdfaa6a81fe0acb24d9ad05153ee17dc

10 years agoam ee874caa: Merge "DispSync: remove delay when enabling vsync events" into klp-dev
Jamie Gennis [Tue, 5 Nov 2013 05:55:09 +0000 (21:55 -0800)]
am ee874caa: Merge "DispSync: remove delay when enabling vsync events" into klp-dev

* commit 'ee874caa81ad48d6d0ee34509d3050ca9c046676':
  DispSync: remove delay when enabling vsync events

10 years agoDon't use implementation-defined format with CPU consumers
Jesse Hall [Tue, 5 Nov 2013 00:43:03 +0000 (16:43 -0800)]
Don't use implementation-defined format with CPU consumers

If the virtual display surface is being consumed by the CPU, it can't
be allowed with HAL_PIXEL_FORMAT_IMPLEMENTATION_DEFINED since there is
no way for the CPU consumer to find out what format gralloc chose. So
for CPU-consumer surfaces, just use the BufferQueue's default format,
which can be set by the consumer.

A better but more invasive change would be to let the consumer require
a certain format (or set of formats?), and disallow the producer from
requesting a different format.

Bug: 11479817
Change-Id: I5b20ee6ac1146550e8799b806e14661d279670c0

10 years agoMerge "DispSync: remove delay when enabling vsync events" into klp-dev
Jamie Gennis [Mon, 4 Nov 2013 22:47:31 +0000 (22:47 +0000)]
Merge "DispSync: remove delay when enabling vsync events" into klp-dev

10 years agoDispSync: remove delay when enabling vsync events
Jamie Gennis [Tue, 29 Oct 2013 20:36:12 +0000 (13:36 -0700)]
DispSync: remove delay when enabling vsync events

This change fixes a bug that caused an extra frame of latency when enabling
vsync event callbacks in DispSync.  The bug was related to the logic that
prevents the two events from firing with very little time between them due to
updates to the vsync model.

Bug: 11479720
Change-Id: Ie7eaff9e92ffb7b7b6cb4d3d4402c96cbd29af7e

10 years agoresolved conflicts for merge of 51b27e82 to klp-dev-plus-aosp
Jesse Hall [Thu, 31 Oct 2013 06:21:21 +0000 (23:21 -0700)]
resolved conflicts for merge of 51b27e82 to klp-dev-plus-aosp

File was moved to system/core.git. Change will be applied there.

Change-Id: Ic2027b7f3f1f1fb4b896a219a8c1b4ab545d9723

10 years agoMerge "fix corruption in Vector<> when malloc falied"
Jesse Hall [Thu, 31 Oct 2013 04:02:24 +0000 (04:02 +0000)]
Merge "fix corruption in Vector<> when malloc falied"

10 years agoam 40da5283: Merge "EGL: rename CallStack::dump into CallStack::log" into klp-dev
Igor Murashkin [Wed, 30 Oct 2013 21:16:19 +0000 (14:16 -0700)]
am 40da5283: Merge "EGL: rename CallStack::dump into CallStack::log" into klp-dev

* commit '40da5283ebc6b5cf1e3820740dc274c47cc55f6d':
  EGL: rename CallStack::dump into CallStack::log

10 years agoMerge "EGL: rename CallStack::dump into CallStack::log" into klp-dev
Igor Murashkin [Wed, 30 Oct 2013 21:09:39 +0000 (21:09 +0000)]
Merge "EGL: rename CallStack::dump into CallStack::log" into klp-dev

10 years agoam 46417b84: Merge "Bug fixes for SensorService" into klp-dev
Aravind Akella [Sat, 26 Oct 2013 01:10:55 +0000 (18:10 -0700)]
am 46417b84: Merge "Bug fixes for SensorService" into klp-dev

* commit '46417b845a257d6725adf048a3cfad6306218e9b':
  Bug fixes for SensorService

10 years agoEGL: rename CallStack::dump into CallStack::log
Igor Murashkin [Sat, 26 Oct 2013 00:13:57 +0000 (17:13 -0700)]
EGL: rename CallStack::dump into CallStack::log

Bug: 11324229
Change-Id: Ibac1be182c336e7aa5ac098f08d18b3c188964b4

10 years agoMerge "Bug fixes for SensorService" into klp-dev
Aravind Akella [Fri, 25 Oct 2013 23:16:59 +0000 (23:16 +0000)]
Merge "Bug fixes for SensorService" into klp-dev

10 years agoam 708cc794: Native counterpart of new power manager method
Marco Nelissen [Fri, 25 Oct 2013 22:01:29 +0000 (15:01 -0700)]
am 708cc794: Native counterpart of new power manager method

* commit '708cc794f6ccd4d31e0a00a23b289272c665adc1':
  Native counterpart of new power manager method

10 years agoBug fixes for SensorService
Aravind Akella [Tue, 15 Oct 2013 22:43:10 +0000 (15:43 -0700)]
Bug fixes for SensorService

i) Emulate Flush for AOSP Fusion Sesnsors on newer HALs that support batching.
ii) Early return if there are no events for the current SensorEventConnection.

Bug: 1132570711376538

Change-Id: Idb856302463649a99d3d5d0c965bb06ba06b8e1d

10 years agoNative counterpart of new power manager method
Marco Nelissen [Wed, 16 Oct 2013 17:57:51 +0000 (10:57 -0700)]
Native counterpart of new power manager method

Cherrypicked from master.

b/9464621

Change-Id: I54c0ed50a57ac870da91db8589a73a2c792dbd8d

10 years agoam c354effe: Fix handling of undefined TARGET_FORCE_HWC_FOR_VIRTUAL_DISPLAYS
Jesse Hall [Fri, 25 Oct 2013 17:52:37 +0000 (10:52 -0700)]
am c354effe: Fix handling of undefined TARGET_FORCE_HWC_FOR_VIRTUAL_DISPLAYS

* commit 'c354effe97be26638618e3bc0da81a3f8b384b20':
  Fix handling of undefined TARGET_FORCE_HWC_FOR_VIRTUAL_DISPLAYS

10 years agoFix handling of undefined TARGET_FORCE_HWC_FOR_VIRTUAL_DISPLAYS
Jesse Hall [Fri, 25 Oct 2013 17:44:41 +0000 (10:44 -0700)]
Fix handling of undefined TARGET_FORCE_HWC_FOR_VIRTUAL_DISPLAYS

related to b/8316155 - Implement HWComposer 1.3 w/ virtual display support
Change-Id: I2dc8d65a74174aa833cc491264f97f1f2df49307

10 years agoam 93fcb823: Merge "Add a boardconfig to force HWC composition of virtual displays...
Jesse Hall [Wed, 23 Oct 2013 22:17:29 +0000 (15:17 -0700)]
am 93fcb823: Merge "Add a boardconfig to force HWC composition of virtual displays" into klp-dev

* commit '93fcb82361a1cd5ceb138f8331ad9cf11fc77c95':
  Add a boardconfig to force HWC composition of virtual displays

10 years agoam c5807f90: Merge "Use implementation-defined format when HWC writes to output buffe...
Jesse Hall [Wed, 23 Oct 2013 22:17:29 +0000 (15:17 -0700)]
am c5807f90: Merge "Use implementation-defined format when HWC writes to output buffer" into klp-dev

* commit 'c5807f90f8e05227e6396f30b3ed91fc01392d9a':
  Use implementation-defined format when HWC writes to output buffer

10 years agoMerge "Add a boardconfig to force HWC composition of virtual displays" into klp-dev
Jesse Hall [Wed, 23 Oct 2013 22:12:54 +0000 (22:12 +0000)]
Merge "Add a boardconfig to force HWC composition of virtual displays" into klp-dev

10 years agoMerge "Use implementation-defined format when HWC writes to output buffer" into klp-dev
Jesse Hall [Wed, 23 Oct 2013 22:12:37 +0000 (22:12 +0000)]
Merge "Use implementation-defined format when HWC writes to output buffer" into klp-dev

10 years agofix corruption in Vector<> when malloc falied
Shuo Gao [Thu, 17 Oct 2013 03:36:11 +0000 (11:36 +0800)]
fix corruption in Vector<> when malloc falied

1. When alloc or realloc failed in the function SharedBuffer::editResize,
it would return a NULL pointer, then mStorage would update to be 1 by
SharedBuffer::data() if no pointer check here, which is an obviously
wrong address, and would cause corruption when used it e.g. in capacity().

So add the pointer check here for the return value of SharedBuffer::editResize,
if it's NULL do not use it to update mStorage, to avoid the value of mStorage
polluted.

2. when alloc or realloc falied in _grow & _shrink function, mStorage keep
the original value, so mCount should not be updated here.

Otherwise, mStorage might be 0 but mCount>0, so a corruption would happend
when it try to delete items from the Vector since mCount>0.

Change-Id: I7c3814e843c459834ca5eed392e8d63d1cb7d2d8
Signed-off-by: Shuo Gao <shuo.gao@intel.com>
Signed-off-by: Jian Luo <jian.luo@intel.com>
Signed-off-by: Bruce Beare <bruce.j.beare@intel.com>
Signed-off-by: Jack Ren <jack.ren@intel.com>
Author-tracking-BZ: 139626

10 years agoam 3223217a: Only consume touches up until the frame time
Michael Wright [Tue, 22 Oct 2013 00:23:55 +0000 (17:23 -0700)]
am 3223217a: Only consume touches up until the frame time

* commit '3223217aed6dcfa55b5b952cd6be71f70e41ba63':
  Only consume touches up until the frame time

10 years agoOnly consume touches up until the frame time
Michael Wright [Mon, 21 Oct 2013 19:05:22 +0000 (12:05 -0700)]
Only consume touches up until the frame time

When resampling is disabled, it's currently possible to consume
touches after the current frame time. This breaks some guarantees and
could cause unexpected behaviors.

Change-Id: I99908a2cac2df9f795dd4a07188b4451213cf3e4

10 years agoam 85075869: Wait for buffers to drain
Andy McFadden [Fri, 18 Oct 2013 23:57:01 +0000 (16:57 -0700)]
am 85075869: Wait for buffers to drain

* commit '8507586903fa803abf535853a169913f2cf2e555':
  Wait for buffers to drain

10 years agoWait for buffers to drain
Andy McFadden [Fri, 18 Oct 2013 22:50:24 +0000 (15:50 -0700)]
Wait for buffers to drain

When a BufferQueue producer disconnects and reconnects, we retain
the previously-queued buffers but empty the slots.  This allows
the number of queued buffers to grow without limit.  The low-memory
killer does not approve.

Bug 11069934

Change-Id: Ia2eaa954c7a3904b54209a3701dba01689e204d8

10 years agoAdd a boardconfig to force HWC composition of virtual displays
Naseer Ahmed [Fri, 4 Oct 2013 20:15:22 +0000 (16:15 -0400)]
Add a boardconfig to force HWC composition of virtual displays

When this boardconfig is defined, even when all virtual display
composition is done by GLES, the HWC will be forced to copy from the
GLES framebuffer to the output buffer. On some hardware this allows
HWC to do format conversions that would otherwise have to be done by
the consumer, with worse power and/or performance.

Bug: 8316155
Change-Id: If980ecc589f138cef063eafa757f7f748196713e

10 years agoUse implementation-defined format when HWC writes to output buffer
Jesse Hall [Fri, 27 Sep 2013 16:05:09 +0000 (09:05 -0700)]
Use implementation-defined format when HWC writes to output buffer

When GLES isn't writing to the output buffer directly, request an
implementation-defined format with minimal usage flags, leaving the
format choice up to gralloc. On some hardware this allows HWC to do
format conversions during composition that would otherwise need to be
done (with worse power and/or performance) by the consumer.

Bug: 8316155
Change-Id: Iee6ee8404282036f9fd1833067cfe11dbadbf0bf

10 years agoam dacd4159: merge in klp-release (no-op)
Ed Heyl [Thu, 17 Oct 2013 00:56:33 +0000 (17:56 -0700)]
am dacd4159: merge in klp-release (no-op)

* commit 'dacd4159631865da1a02203fe5fa7c5f1339cda9':
  Change API from flush(handle) to flush(). Call flush on all active sensors in the given SensorEventConnection.
  Eliminate latency when resampling is disabled
  Fix blank / partial screenshots
  Set the outbuf acquire fence after we actually have it.
  only clear FB when asked for the opaque layer
  Treat composition frames with no layers as using GLES composition
  Fix two EGLConfig selection bugs
  fix crashers with wifi/virtual displays

10 years agomerge in klp-release (no-op)
Ed Heyl [Thu, 17 Oct 2013 00:43:06 +0000 (17:43 -0700)]
merge in klp-release (no-op)

10 years agomerge in klp-release history after reset to klp-dev
The Android Automerger [Wed, 16 Oct 2013 04:14:09 +0000 (21:14 -0700)]
merge in klp-release history after reset to klp-dev

10 years agoam 0a645cc5: SurfaceFlinger: give SF its own vsync phase
Jamie Gennis [Tue, 15 Oct 2013 23:17:18 +0000 (16:17 -0700)]
am 0a645cc5: SurfaceFlinger: give SF its own vsync phase

* commit '0a645cc5a935e67a8d1effc7679a838160b971d8':
  SurfaceFlinger: give SF its own vsync phase