OSDN Git Service

android-x86/frameworks-base.git
13 years agoDO NOT MERGE: This particular RTSP server streams MPEG4-LATM audio with extra trailin...
Andreas Huber [Wed, 26 Jan 2011 23:40:56 +0000 (15:40 -0800)]
DO NOT MERGE: This particular RTSP server streams MPEG4-LATM audio with extra trailing bytes.

And now we're just ignoring them. Yay standards.

Change-Id: Ia8c0b9161e606152fb681f0dda3ba901954dc749
related-to-bug: 3353752

13 years agoMerge "Make CHANGE_COMPONENT_ENABLED_STATE a signatureOrSystem perm." into gingerbread
Dianne Hackborn [Wed, 26 Jan 2011 21:59:42 +0000 (13:59 -0800)]
Merge "Make CHANGE_COMPONENT_ENABLED_STATE a signatureOrSystem perm." into gingerbread

13 years agoMake CHANGE_COMPONENT_ENABLED_STATE a signatureOrSystem perm.
Dianne Hackborn [Wed, 26 Jan 2011 20:40:18 +0000 (12:40 -0800)]
Make CHANGE_COMPONENT_ENABLED_STATE a signatureOrSystem perm.

Change-Id: I53a6671d5478e7dbca44b17d90f1edfa8e3420bc

13 years agodo not merge bug 3370834 Cherrypick from master
Jean-Michel Trivi [Mon, 15 Nov 2010 20:11:32 +0000 (12:11 -0800)]
do not merge bug 3370834 Cherrypick from master

Cherripick from master CL 79833, 79417, 78864, 80332, 87500

Add new audio mode and recording source for audio communications
 other than telelphony.

The audio mode MODE_IN_CALL signals the system the device a phone
 call is currently underway. There was no way for audio video
 chat or VoIP applications to signal a call is underway, but not
 using the telephony resources. This change introduces a new mode
 to address this. Changes in other parts of the system (java
 and native) are required to take this new mode into account.
The generic AudioPolicyManager is updated to not use its phone
 state variable directly, but to use two new convenience methods,
 isInCall() and isStateInCall(int) instead.

Add a recording source used to designate a recording stream for
voice communications such as VoIP.

Update the platform-independent audio policy manager to pass the
 nature of the audio recording source to the audio policy client
 interface through the AudioPolicyClientInterface::setParameters()
 method.

SIP calls should set the audio mode to MODE_IN_COMMUNICATION,
 Audio mode MODE_IN_CALL is reserved for telephony.

SIP: Enable built-in echo canceler if available.
1. Always initialize AudioRecord with VOICE_COMMUNICATION.
2. If echo canceler is available, disable our echo suppressor.

Note that this CL is intentionally not correcting the
 getAudioSourceMax() return value in MediaRecorder.java as the
 new source is hidden here.

Change-Id: Ie68cd03c50553101aa2ad838fe9459b2cf151bc8

13 years agoMerge "Make OBB results a one-way call" into gingerbread
Kenny Root [Wed, 26 Jan 2011 17:17:28 +0000 (09:17 -0800)]
Merge "Make OBB results a one-way call" into gingerbread

13 years agoMake OBB results a one-way call
Kenny Root [Wed, 26 Jan 2011 00:39:35 +0000 (16:39 -0800)]
Make OBB results a one-way call

OBB result back to an application needs to be a one-way binder call.

Bug: 3353699
Change-Id: I0e625914d18a001c2fa9d764ea6463f34cf96743

13 years agoMerge "Fix the build, proper current.xml." into gingerbread
Jeff Hamilton [Wed, 26 Jan 2011 00:11:37 +0000 (16:11 -0800)]
Merge "Fix the build, proper current.xml." into gingerbread

13 years agoFix the build, proper current.xml.
Jeff Hamilton [Wed, 26 Jan 2011 00:10:58 +0000 (18:10 -0600)]
Fix the build, proper current.xml.

Change-Id: I9b02206042c74fd5c329fe2b70ed2b6b91d41382

13 years agoMerge "fix [3361121] hang in glClear() - device unresponsive, OTA fails (DO NOT MERGE...
Mathias Agopian [Tue, 25 Jan 2011 22:43:21 +0000 (14:43 -0800)]
Merge "fix [3361121] hang in glClear() - device unresponsive, OTA fails (DO NOT MERGE)" into gingerbread

13 years agoMerge "partially fix [3306150] HTML5 video with H/W acceleration blackout (DO NOT...
Mathias Agopian [Tue, 25 Jan 2011 22:43:14 +0000 (14:43 -0800)]
Merge "partially fix [3306150] HTML5 video with H/W acceleration blackout (DO NOT MERGE)" into gingerbread

13 years agoMerge "fix a small bug that caused screenshot to show garbage in some cases (DO NOT...
Mathias Agopian [Tue, 25 Jan 2011 22:43:09 +0000 (14:43 -0800)]
Merge "fix a small bug that caused screenshot to show garbage in some cases (DO NOT MERGE)" into gingerbread

13 years agoMerge "workaround [3201922] display not on: log full of gralloc errors (DO NOT MERGE...
Mathias Agopian [Tue, 25 Jan 2011 22:43:05 +0000 (14:43 -0800)]
Merge "workaround [3201922] display not on: log full of gralloc errors (DO NOT MERGE)" into gingerbread

13 years agoMerge "minor clean-up. SurfaceFlinger doesn't need libpixelflinger.so. (DO NOT MERGE...
Mathias Agopian [Tue, 25 Jan 2011 22:42:59 +0000 (14:42 -0800)]
Merge "minor clean-up. SurfaceFlinger doesn't need libpixelflinger.so. (DO NOT MERGE)" into gingerbread

13 years agoMerge "Use classnames instead of ints for NFC techs." into gingerbread
Jeff Hamilton [Tue, 25 Jan 2011 22:35:46 +0000 (14:35 -0800)]
Merge "Use classnames instead of ints for NFC techs." into gingerbread

13 years agoUse classnames instead of ints for NFC techs.
Jeff Hamilton [Mon, 24 Jan 2011 20:53:00 +0000 (14:53 -0600)]
Use classnames instead of ints for NFC techs.

This makes the system more flexible and allows
adding new technology types without having to
update the API.

Change-Id: Iaee6b633965e501a70e8afc3f1d54d9d94a4d05a

13 years agofix [3361121] hang in glClear() - device unresponsive, OTA fails (DO NOT MERGE)
Mathias Agopian [Tue, 18 Jan 2011 23:51:30 +0000 (15:51 -0800)]
fix [3361121] hang in glClear() - device unresponsive, OTA fails (DO NOT MERGE)

Generally we never want to lock a buffer for write access if it is at
the "head" on the surfaceflinger side. The only exception (1) is when
the buffer is not currently in use AND there is at least one queued
buffer -- in which case, SurfaceFlinger will never use said buffer
anymore, because on the next composition around, it will be able to
retire the first queued buffer.

The logic above relies on SurfaceFlinger always retiring
and locking a buffer before composition -- unfortunately this
didn't happen during a screenshot.

This could leave us in a situation where a buffer is locked by the
application for write, and used by SurfaceFlinger for texturing,
causing a hang.

Here, we fix this issue by never assuming the exception (1), it was
intended as an optimization allowing ANativeWindow::lockBuffer() to
return sooner and was justified when most of SF composition was
done in software. The actual buffer locking is now ensured by
gralloc. We could have handled screenshots in a similar way to
a regular composition, but it could have caused glitches on screen,
essentially, taking a screenshot could cause to skip a frame.

now that we removed the notion of a "inUse" buffer in surfaceflinger
a lot of code can be simplified / removed.

noteworthy, the whole concept of "unlockClient" wrt. "compositionComplete"
is also gone.

13 years agopartially fix [3306150] HTML5 video with H/W acceleration blackout (DO NOT MERGE)
Mathias Agopian [Sat, 15 Jan 2011 01:37:42 +0000 (17:37 -0800)]
partially fix [3306150] HTML5 video with H/W acceleration blackout (DO NOT MERGE)

We used to guarantee that a layer in SurfaceFlinger would never be
destroyed before all references (to its ISurface) on the client
side would be released. At some point, this guarantee got
relaxed to allow to free gralloc resources sooner. This last
change was incorrect, because:
- in implementations with reference-counting the gralloc resources
wouldn't be released anyways, until all the mapping were gone
- in implementations without ref counting, the client side
would most likely crash or do something bad
- it also caused the SharedBufferStack slot to be reallocated
to another surface, which could be problematic if the client
continued to use the surface after the window manager destroyed it.

So, we essentially reinstate the guarantee that layers won't be
destroyed until after all references to their ISurface are
released.

NOTE: This doesn't entirely fix 3306150 because there is another
problem there where the Browser continues to use a surface after it
has been destroyed.

also improve SurfaceFlinger 'dumpsys' log

list the purgatory, which shows windows that have been closed,
but for which the client still has references.

13 years agofix a small bug that caused screenshot to show garbage in some cases (DO NOT MERGE)
Mathias Agopian [Fri, 17 Dec 2010 02:46:17 +0000 (18:46 -0800)]
fix a small bug that caused screenshot to show garbage in some cases (DO NOT MERGE)

we were not clearing the screen entirely, which caused garbage when
the screen wasn't entirely covered by windows.

Change-Id: Ia7aa13c36a8a314e0e8427d419b16b9aa2165ddf

13 years agoworkaround [3201922] display not on: log full of gralloc errors (DO NOT MERGE)
Mathias Agopian [Wed, 15 Dec 2010 22:41:59 +0000 (14:41 -0800)]
workaround [3201922] display not on: log full of gralloc errors (DO NOT MERGE)

we make sure to call compositionComplete after everytime we do
composition with the GPU (even for the screenshot case), which is
where the buffer locks are released.

Change-Id: I450430d1e4d1ee9ce1023970642378c42cdcfa4c

13 years agominor clean-up. SurfaceFlinger doesn't need libpixelflinger.so. (DO NOT MERGE)
Mathias Agopian [Tue, 14 Dec 2010 00:47:31 +0000 (16:47 -0800)]
minor clean-up. SurfaceFlinger doesn't need libpixelflinger.so. (DO NOT MERGE)

Change-Id: I1ddbbbec4fa5b2521ef3787bc28efe1bc90d0060

13 years agoMerge "Increase the number of useful bits in the spectrum." into gingerbread
Marco Nelissen [Tue, 25 Jan 2011 18:06:16 +0000 (10:06 -0800)]
Merge "Increase the number of useful bits in the spectrum." into gingerbread

13 years agoMake Mifare Classic increment/decrement operands little endian
Nick Pelly [Tue, 25 Jan 2011 15:45:07 +0000 (07:45 -0800)]
Make Mifare Classic increment/decrement operands little endian

Also make sure they are non-negative.

This is not documented in Mifare Classic spec, but based on findings from NXP:
- Operand should be stored in little-endian format in the transceive buffer
- Tag ignores the sign bit on the operand, its effectively 31-bit unsigned
- Overflow and underflow generates an error.

Change-Id: Id3389b3894ded732c4b00d564ca53f5df651359e

13 years agoMerge "Get mute state from active call." into gingerbread
Hung-ying Tyan [Tue, 25 Jan 2011 05:13:08 +0000 (21:13 -0800)]
Merge "Get mute state from active call." into gingerbread

13 years agoMerge changes Ib70e0cf2,I0691cd70 into gingerbread
Hung-ying Tyan [Tue, 25 Jan 2011 03:41:42 +0000 (19:41 -0800)]
Merge changes Ib70e0cf2,I0691cd70 into gingerbread

* changes:
  SipService: registers broadcast receivers on demand.
  SipService: release wake lock for cancelled tasks.

13 years agoFix SENSOR_LANDSCAPE and SENSOR_PORTRAIT (DO NOT MERGE)
Jeff Brown [Sun, 23 Jan 2011 21:05:44 +0000 (13:05 -0800)]
Fix SENSOR_LANDSCAPE and SENSOR_PORTRAIT (DO NOT MERGE)

Fixed a bug in the handling of SENSOR_LANDSCAPE and SENSOR_PORTAIT
on devices that have a natural landscape orientation.  The old code
was disabling 180 degree orientation detection when it shouldn't have.

Bug: 3381359
Change-Id: I19bd2519e7f69a24835840fbf542aa7349981029

13 years agoDO NOT MERGE: Support non-multiple-of-16 dimensions in MPEG4/H.263 software decoder
Andreas Huber [Mon, 24 Jan 2011 21:36:30 +0000 (13:36 -0800)]
DO NOT MERGE: Support non-multiple-of-16 dimensions in MPEG4/H.263 software decoder

Change-Id: I6c27c00a6e13cb3635e61089c0db0989f5810c37
related-to-bug: 3384367

13 years agoUpdated checkbox assets
Justin Ho [Mon, 24 Jan 2011 21:05:06 +0000 (13:05 -0800)]
Updated checkbox assets

Change-Id: I93c979816bc3310c510bf8942c273ce8c411150d

13 years agoFix trace mismatch.
Arnaud Ferir [Mon, 24 Jan 2011 10:13:58 +0000 (11:13 +0100)]
Fix trace mismatch.

Change-Id: Ie6e9a2baef63d2b5621593203742c2ffeaafc4ba

13 years agoHandle mock tags in android.nfc.Tag conversion to Parcel.
Sylvain Fonteneau [Mon, 24 Jan 2011 09:23:43 +0000 (10:23 +0100)]
Handle mock tags in android.nfc.Tag conversion to Parcel.

When generating a mock tag (after a NDEF exchange over LLCP), one of
the internal fields is set to null. This was causing NullPointerException
when being converted to a Parcel.

This is fixed by not including this field in the Parcel for mock tags.

Change-Id: I000e2faa54d71fd755ba7993e1e258743aad98fb

13 years agoMerge "Enable recovery in RIL wakelock release check." into gingerbread
John Wang [Mon, 24 Jan 2011 17:14:14 +0000 (09:14 -0800)]
Merge "Enable recovery in RIL wakelock release check." into gingerbread

13 years agoIncrease the number of useful bits in the spectrum.
Marco Nelissen [Wed, 19 Jan 2011 00:44:28 +0000 (16:44 -0800)]
Increase the number of useful bits in the spectrum.

Change-Id: I3efc2ac8db689285566df35cb6f594aab41141db

13 years agoMerge "Add operands to mifare classic increment, decrement." into gingerbread
Nick Pelly [Mon, 24 Jan 2011 06:12:25 +0000 (22:12 -0800)]
Merge "Add operands to mifare classic increment, decrement." into gingerbread

13 years agoAdd operands to mifare classic increment, decrement.
Nick Pelly [Mon, 24 Jan 2011 06:11:35 +0000 (22:11 -0800)]
Add operands to mifare classic increment, decrement.

Change-Id: Ib35f615142bda48a5e33888a09ebae2880624788

13 years agoMerge "Make MifareClassic methods more consistent." into gingerbread
Nick Pelly [Mon, 24 Jan 2011 05:36:14 +0000 (21:36 -0800)]
Merge "Make MifareClassic methods more consistent." into gingerbread

13 years agoMake MifareClassic methods more consistent.
Nick Pelly [Sat, 22 Jan 2011 06:11:29 +0000 (22:11 -0800)]
Make MifareClassic methods more consistent.

Remove method overloading for combinations of sector+block addressing.
Instead provide methods that more closly match the raw commands, and more
efficient helpers to convert between blocks and sectors.

o fix off-by-one bug in getBlockCountInSector()
o add BLOCK_SIZE
o remove DESFIRE not operating in classic emulation (SAK 0x20)
o hide isEmulated(), there is no use case, and the info is available elsewhere
o getTotalBlockCount() -> getBlockCount()
o getBlockCount(int) -> getBlockCountInSector(int)
o introduce blockToSector() and sectorToBlock()
o remove authenticateBlock()
    make it really clear that authentication is per sector, and reduce function
    explosion. blockToSector() allows you to use authenticateSector...
o explicit authenticateSectorWithKeyA() / authenticateSectorWithKeyB()
    get rid of magic boolean
o remove all (int sector, int block) parameters
    always address by absolute block. this makes the API crystal clear, and
    helps reduce function explosion
o validation of all sector and block indices
o dont & 0xff when converting to byte - its redundant
o Remove TYPE_OTHER. Mifare Classic types are well-known and stable.

Change-Id: I3c9f8254ff307f31b388b3d7592c862d5de6afa5

13 years agoMerge "Allow null filters for foreground dispatch to act as catch all." into gingerbread
Jeff Hamilton [Mon, 24 Jan 2011 03:17:22 +0000 (19:17 -0800)]
Merge "Allow null filters for foreground dispatch to act as catch all." into gingerbread

13 years agoFix 3379239: Catch exceptions from workaround for OOM.
Jim Miller [Sun, 23 Jan 2011 21:40:31 +0000 (13:40 -0800)]
Fix 3379239: Catch exceptions from workaround for OOM.

This catches possible exceptions caused by attempting to
unregister a receiver that may have already been unregistered.
This is related to the OOM issue workaround in Change I99a7e0c3.

Change-Id: Ib69a8acc6b12659125c958a03efac0a8a68a3859

13 years agoAllow null filters for foreground dispatch to act as catch all.
Jeff Hamilton [Sun, 23 Jan 2011 18:14:42 +0000 (12:14 -0600)]
Allow null filters for foreground dispatch to act as catch all.

Change-Id: Idf3d2e344a6e87035ccb5b533ed9e81fec596132

13 years agoMerge "Do Not Merge Connect other profiles when priority is auto connect." into ginge...
Jaikumar Ganesh [Sat, 22 Jan 2011 05:25:01 +0000 (21:25 -0800)]
Merge "Do Not Merge Connect other profiles when priority is auto connect." into gingerbread

13 years agoDo Not Merge Connect other profiles when priority is auto connect.
Jaikumar Ganesh [Sat, 22 Jan 2011 02:58:17 +0000 (18:58 -0800)]
Do Not Merge Connect other profiles when priority is auto connect.

ACL auto connection was removed so other profiles
won't auto connect when priority was auto connect.

Reduced connect other profiles time delay.

Bug: 3281016
Change-Id: Ic56ca6e95f618f02bc316d1f80014e58c4b6c3bc

13 years agoEnable recovery in RIL wakelock release check.
John Wang [Sat, 22 Jan 2011 00:18:26 +0000 (16:18 -0800)]
Enable recovery in RIL wakelock release check.

Wakelock will get released while
1) no request pending to be sent out, in which mRequestMessagesPending increases
before calling EVENT_SEND and decreases while handling EVENT_SEND.

and

2) no waiting requests sent to RIL but no replied, in which mRequestMessagesWaiting
increases while sending request and decreases while handling response.

Both will be cleared while WAKE_LOCK_TIMEOUT occurs to recovery from out of sync situation.

bug: 33694273370827
Change-Id: Ib2fc54db3b155bd3fb1296ad83720b7836708caf

13 years agoMerge "DO NOT MERGE - Create db path if it doesn't exit" into gingerbread
John Reck [Fri, 21 Jan 2011 22:03:30 +0000 (14:03 -0800)]
Merge "DO NOT MERGE - Create db path if it doesn't exit" into gingerbread

13 years agoDO NOT MERGE - Create db path if it doesn't exit
John Reck [Tue, 18 Jan 2011 23:42:49 +0000 (15:42 -0800)]
DO NOT MERGE - Create db path if it doesn't exit

 Bug: 3241808
 Make sure the directory the caller is trying to open the icon db
 in actually exists and create it if it doesn't.

Change-Id: Icc0a67b0ceba03a746ac2cd3b84fc34c2c8b4fe2

13 years agoMerge "API changes for NFC." into gingerbread
Jeff Hamilton [Fri, 21 Jan 2011 19:34:20 +0000 (11:34 -0800)]
Merge "API changes for NFC." into gingerbread

13 years agoAPI changes for NFC.
Jeff Hamilton [Fri, 21 Jan 2011 07:13:06 +0000 (01:13 -0600)]
API changes for NFC.

Bug: 3366009
Change-Id: Ia28e03e1501421906b640b78d4c9a59a20e9c668

13 years agoMerge "Bug 3365937 notify TTS initialization listener of binding errors" into gingerbread
Jean-Michel Trivi [Fri, 21 Jan 2011 17:09:30 +0000 (09:09 -0800)]
Merge "Bug 3365937 notify TTS initialization listener of binding errors" into gingerbread

13 years agoMerge "do not merge - Fix issue 3371096." into gingerbread
Eric Laurent [Fri, 21 Jan 2011 01:43:04 +0000 (17:43 -0800)]
Merge "do not merge - Fix issue 3371096." into gingerbread

13 years agoBug 3365937 notify TTS initialization listener of binding errors
Jean-Michel Trivi [Fri, 21 Jan 2011 01:25:48 +0000 (17:25 -0800)]
Bug 3365937 notify TTS initialization listener of binding errors

The creation of a TextToSpeech object causes the TTS service to
 start, and the initialization listener to be called when the
 service is connected. But the listener is never called when
 the service binding failed to be notified of this error.
The fix consists in checking the result of the bind to service
 operation, and notify the listener in case of an error.

More log was added in case speak() and synthesizeToFile() are
 called but the service is not known to have started.

Change-Id: I7dcc1fa44be31fee3177ec6215fca3306377b934

13 years agoFix issue #3373438: Update GBMR versions
Dianne Hackborn [Thu, 20 Jan 2011 20:16:43 +0000 (12:16 -0800)]
Fix issue #3373438: Update GBMR versions

Change-Id: Ic4b4f1b0534fd2fbd6322eae01a4cb9d5a67cc6f

13 years agodo not merge - Fix issue 3371096.
Eric Laurent [Thu, 20 Jan 2011 02:36:13 +0000 (18:36 -0800)]
do not merge - Fix issue 3371096.

EffectModule::process() was copying effect chain input buffer to
output buffer if no effect was active instead of accumulating it.

Change-Id: If4ca75601ea69a088d0f71d88aec53e90a1dec89

13 years agoMerge "DO NOT MERGE: Prefill the cache before trying to instantiate the media extract...
Dave Sparks [Thu, 20 Jan 2011 03:23:32 +0000 (19:23 -0800)]
Merge "DO NOT MERGE: Prefill the cache before trying to instantiate the media extractor." into gingerbread

13 years agoMerge "fix [3369027] Sensor.TYPE_ROTATION_VECTOR is unstable and returns NaNs when...
Mathias Agopian [Thu, 20 Jan 2011 01:32:21 +0000 (17:32 -0800)]
Merge "fix [3369027] Sensor.TYPE_ROTATION_VECTOR is unstable and returns NaNs when running slowly" into gingerbread

13 years agoMerge "Never drag scroll views with no children. (DO NOT MERGE)" into gingerbread
Jeff Brown [Thu, 20 Jan 2011 01:03:03 +0000 (17:03 -0800)]
Merge "Never drag scroll views with no children. (DO NOT MERGE)" into gingerbread

13 years agoDO NOT MERGE: Prefill the cache before trying to instantiate the media extractor.
Andreas Huber [Wed, 19 Jan 2011 23:07:19 +0000 (15:07 -0800)]
DO NOT MERGE: Prefill the cache before trying to instantiate the media extractor.

The latter is an operation that otherwise could block on the datasource
for a significant amount of time. During that time we'd be unable to
abort the preparation phase without this prefill.

Change-Id: I3bc889b264f599bfd5c2bbdf48b88ccb55d86172
related-to-bug: 3362836

13 years agofix [3369027] Sensor.TYPE_ROTATION_VECTOR is unstable and returns NaNs when running...
Mathias Agopian [Wed, 19 Jan 2011 22:26:25 +0000 (14:26 -0800)]
fix [3369027] Sensor.TYPE_ROTATION_VECTOR is unstable and returns NaNs when running slowly

The cut-off frequency of the lowpass filter was too high
for the sampling rate used by DELAY_NORMAL.

Now we use the same filters used for the gravity vector
(cascaded biquad at 1.5 Hz)

Change-Id: Iac290a716cc47a78337a8f0e45b103e49b4d9d78

13 years agoMerge "Use EGL_NATIVE_VISUAL_ID to select EGLConfig" into gingerbread
Mathias Agopian [Wed, 19 Jan 2011 19:31:07 +0000 (11:31 -0800)]
Merge "Use EGL_NATIVE_VISUAL_ID to select EGLConfig" into gingerbread

13 years agoNever drag scroll views with no children. (DO NOT MERGE)
Jeff Brown [Wed, 19 Jan 2011 02:22:48 +0000 (18:22 -0800)]
Never drag scroll views with no children. (DO NOT MERGE)

Fixes a regression from Froyo.  Previously, when a scroll view
had no children, its onTouchEvent would return because the scroll
view would only start dragging if the user touched one of its children.
In Gingerbread, the user can drag from anywhere within the scroll
view, not just by touching a child.  However, it makes no sense
to drag a scroll view that has no children so an empty scroll view
should just ignore touches like any other empty view group would.

This change fixes applications that for some reason or other happen
to have empty scroll views in strange places.

Bug: 3246230
Change-Id: I76c6136d3cd74968983033014d60deec1718ce30

13 years agoMerge "DO NOT MERGE: Apparently our native TCP sockets do not return an error from...
Andreas Huber [Wed, 19 Jan 2011 16:11:39 +0000 (08:11 -0800)]
Merge "DO NOT MERGE: Apparently our native TCP sockets do not return an error from blocking "connect"" into gingerbread

13 years agoMerge "Change signature of MifareClassic.authenticate*() to throw IOException." into...
Nick Pelly [Wed, 19 Jan 2011 02:53:42 +0000 (18:53 -0800)]
Merge "Change signature of MifareClassic.authenticate*() to throw IOException." into gingerbread

13 years agoMerge "Support for TagLostException (API)." into gingerbread
Nick Pelly [Wed, 19 Jan 2011 02:53:40 +0000 (18:53 -0800)]
Merge "Support for TagLostException (API)." into gingerbread

13 years agoChange signature of MifareClassic.authenticate*() to throw IOException.
Nick Pelly [Wed, 19 Jan 2011 02:52:37 +0000 (18:52 -0800)]
Change signature of MifareClassic.authenticate*() to throw IOException.

Update API for this and the previous commit.

Change-Id: I19275a5d00081bd0f00afea59a2b673344a32954

13 years agoMerge "Watchdog can get deadlocked on activity manager" into gingerbread
Brad Fitzpatrick [Wed, 19 Jan 2011 02:52:04 +0000 (18:52 -0800)]
Merge "Watchdog can get deadlocked on activity manager" into gingerbread

13 years agoSupport for TagLostException (API).
Martijn Coenen [Tue, 18 Jan 2011 23:57:17 +0000 (00:57 +0100)]
Support for TagLostException (API).

Change-Id: Idf2c3253f7bc5f371133c4f18bfc2afde69c3010

13 years agoMerge "Fix javadoc which broke build." into gingerbread
Nick Pelly [Wed, 19 Jan 2011 01:07:56 +0000 (17:07 -0800)]
Merge "Fix javadoc which broke build." into gingerbread

13 years agoFix javadoc which broke build.
Nick Pelly [Wed, 19 Jan 2011 01:07:05 +0000 (17:07 -0800)]
Fix javadoc which broke build.

Change-Id: Ie3cac9bec04e80defae44ba02086ad0cdfd301a9

13 years agoWatchdog can get deadlocked on activity manager
Brad Fitzpatrick [Fri, 14 Jan 2011 19:28:22 +0000 (11:28 -0800)]
Watchdog can get deadlocked on activity manager

Bug: 3351719
Change-Id: Idf3907709340062849a09dd094137e86256ed739

13 years agoMerge "Fix omission in NOTICE file." into gingerbread
Eric Laurent [Wed, 19 Jan 2011 00:42:20 +0000 (16:42 -0800)]
Merge "Fix omission in NOTICE file." into gingerbread

13 years agoMerge "hange the default IsoDep timeout to 300ms (docs update)." into gingerbread
Nick Pelly [Wed, 19 Jan 2011 00:11:14 +0000 (16:11 -0800)]
Merge "hange the default IsoDep timeout to 300ms (docs update)." into gingerbread

13 years agoMerge "fix [3355413] Gyro calls onAccuracyChanged and reports accuracy as UNRELIABLE...
Mathias Agopian [Wed, 19 Jan 2011 00:10:08 +0000 (16:10 -0800)]
Merge "fix [3355413] Gyro calls onAccuracyChanged and reports accuracy as UNRELIABLE" into gingerbread

13 years agoDO NOT MERGE: Apparently our native TCP sockets do not return an error from blocking...
Andreas Huber [Tue, 18 Jan 2011 23:26:40 +0000 (15:26 -0800)]
DO NOT MERGE: Apparently our native TCP sockets do not return an error from blocking "connect"

if the network interface is shutdown while connecting.

Change-Id: I168c6026de24812efa9b7e607a9eb83efded8c1f
related-to-bug: 3362836

13 years agoMerge "Set the default state of NOTIFICATIONS_USE_RING_VOLUME to true." into gingerbread
Amith Yamasani [Tue, 18 Jan 2011 23:41:26 +0000 (15:41 -0800)]
Merge "Set the default state of NOTIFICATIONS_USE_RING_VOLUME to true." into gingerbread

13 years agohange the default IsoDep timeout to 300ms (docs update).
Martijn Coenen [Tue, 18 Jan 2011 20:22:08 +0000 (21:22 +0100)]
hange the default IsoDep timeout to 300ms (docs update).

Change-Id: I4fe5a6bf9fd42a28c9619e6e0bf11885bcc43f91

13 years agoFix omission in NOTICE file.
Eric Laurent [Tue, 18 Jan 2011 23:25:32 +0000 (15:25 -0800)]
Fix omission in NOTICE file.

Added copyright notice for audio effects source files under
media/libeffects/lvm.

Change-Id: I8e142bacb7e4b1faaeafe88e907b73f21fced1c3

13 years agoMerge "Do not add unsolicited events to queue" into gingerbread
Irfan Sheriff [Tue, 18 Jan 2011 22:40:18 +0000 (14:40 -0800)]
Merge "Do not add unsolicited events to queue" into gingerbread

13 years agoMerge "Filter virtual keys after touches. (DO NOT MERGE)" into gingerbread
Jeff Brown [Tue, 18 Jan 2011 22:31:31 +0000 (14:31 -0800)]
Merge "Filter virtual keys after touches.  (DO NOT MERGE)" into gingerbread

13 years agoSet the default state of NOTIFICATIONS_USE_RING_VOLUME to true.
Amith Yamasani [Tue, 18 Jan 2011 21:13:47 +0000 (13:13 -0800)]
Set the default state of NOTIFICATIONS_USE_RING_VOLUME to true.

Bug: 3319615

This doesn't affect upgrades, only new installs.

Change-Id: I86fa77b36867649becce8fc470bf9a559fe48131

13 years agoCheck NDEF before calling makeReadOnly().
Martijn Coenen [Fri, 14 Jan 2011 15:26:45 +0000 (16:26 +0100)]
Check NDEF before calling makeReadOnly().

Change-Id: Iadbaf5c29821eff78ad487b248af529febcdc4ce

13 years agoImplement IsoDep timeout handling (API).
Martijn Coenen [Wed, 12 Jan 2011 20:17:43 +0000 (21:17 +0100)]
Implement IsoDep timeout handling (API).

Added a method for setting the timeout on IsoDep transactions.

Change-Id: Ie627e7a826556e46295fefe69b9be83ebf911d93

13 years agoMerge "Fix the ISurfaceComposer onTransact switch." into gingerbread
Mathias Agopian [Tue, 18 Jan 2011 19:20:04 +0000 (11:20 -0800)]
Merge "Fix the ISurfaceComposer onTransact switch." into gingerbread

13 years agoMerge "Javadoc updates for NFC." into gingerbread
Jeff Hamilton [Tue, 18 Jan 2011 19:01:55 +0000 (11:01 -0800)]
Merge "Javadoc updates for NFC." into gingerbread

13 years agoMerge "New APIs for NDEF Push Protocol." into gingerbread
Jeff Hamilton [Tue, 18 Jan 2011 19:01:05 +0000 (11:01 -0800)]
Merge "New APIs for NDEF Push Protocol." into gingerbread

13 years agoJavadoc updates for NFC.
Jeff Hamilton [Mon, 17 Jan 2011 19:05:03 +0000 (11:05 -0800)]
Javadoc updates for NFC.

Change-Id: Ibd91829979576297599fbcc9eb8054924af1d527

13 years agoRemove API that snuck into MockPackageManager
Kenny Root [Mon, 17 Jan 2011 20:40:57 +0000 (12:40 -0800)]
Remove API that snuck into MockPackageManager

We didn't release the get/setPackageObbPath API, but it snuck into
MockPackageManager. However, it is not usable as a developer anyway.
Removing it so it doesn't stay in the API forever.

Change-Id: Ia569754691dc3a9f2672c1f25e4e08d780db6028

13 years agoRevert "Add the {get,set}PackageObbPaths calls to API"
Kenny Root [Mon, 17 Jan 2011 01:25:18 +0000 (17:25 -0800)]
Revert "Add the {get,set}PackageObbPaths calls to API"

This reverts commit 6c4d904851772313930f800ac7c323cf90c709bb.

Going with a different tactic that doesn't dump stuff on
PackageManagerService.

Bug: 3214719
Change-Id: I0bbeccf3c21d264deda4256eb53713d2c98284f4

13 years agoNew APIs for NDEF Push Protocol.
Jeff Hamilton [Mon, 17 Jan 2011 15:59:03 +0000 (07:59 -0800)]
New APIs for NDEF Push Protocol.

The NPP is only usable by the foregorund
activity to prevent dispatching confusion
on the far end.

Change-Id: I08475a52083fd7f81b79b7fe2faf4e126121a809

13 years agofix [3355413] Gyro calls onAccuracyChanged and reports accuracy as UNRELIABLE
Mathias Agopian [Mon, 17 Jan 2011 03:11:20 +0000 (19:11 -0800)]
fix [3355413] Gyro calls onAccuracyChanged and reports accuracy as UNRELIABLE

- make sure to call onAccuracyChanged only when it actually changes
- only call onAccuracyChanged on Magnetic and Orientation sensors, which
  are the only ones to support it.
- for other sensors, call onAccuracyChanged only once with the
  SENSOR_STATUS_ACCURACY_HIGH value

Change-Id: Ic861e3d48d3b98dab74ffebe2746576f5b69f371

13 years agoUse EGL_NATIVE_VISUAL_ID to select EGLConfig
Mathias Agopian [Mon, 17 Jan 2011 01:57:20 +0000 (17:57 -0800)]
Use EGL_NATIVE_VISUAL_ID to select EGLConfig

EGLUtils::selectConfigForPixelFormat() now uses EGL_NATIVE_VISUAL_ID
to select a config with the proper format. this is more robust and
future proof.

Change-Id: Ib85e0974160945d838956b50a3cac4de78618957

13 years agoFix the ISurfaceComposer onTransact switch.
Jamie Gennis [Mon, 20 Dec 2010 19:05:18 +0000 (11:05 -0800)]
Fix the ISurfaceComposer onTransact switch.

Two of the cases in the ISurfaceComposer onTransact switch statement
were missing 'break' statements at the end, and would fall through to
the next case block. This change adds those break statements.

Change-Id: I6dcc84263d3ea03d94612c667103283846b3dee1

13 years agoMerge "Flush Binder commands in ndk looper." into gingerbread
Jeff Brown [Mon, 17 Jan 2011 00:15:32 +0000 (16:15 -0800)]
Merge "Flush Binder commands in ndk looper." into gingerbread

13 years agoMerge "Fix 3106227: use WeakReferences for receivers in DigitalClock class" into...
Jim Miller [Mon, 17 Jan 2011 00:14:52 +0000 (16:14 -0800)]
Merge "Fix 3106227: use WeakReferences for receivers in DigitalClock class" into gingerbread

13 years agoDo not add unsolicited events to queue
Irfan Sheriff [Sun, 16 Jan 2011 22:31:55 +0000 (14:31 -0800)]
Do not add unsolicited events to queue

Adding unsolicited events to response queue
results in doCommand() returning the wrong
result.

Pulling this change from master.

Bug: 3258529
Change-Id: I2a4b0bd8bb30b31682d7d61ee94a0d246bf10de2

13 years agoMerge "Make sure service handles are live before passing to other objects." into...
Nick Pelly [Sat, 15 Jan 2011 02:18:25 +0000 (18:18 -0800)]
Merge "Make sure service handles are live before passing to other objects." into gingerbread

13 years agoFix 3106227: use WeakReferences for receivers in DigitalClock class
Jim Miller [Fri, 14 Jan 2011 01:56:35 +0000 (17:56 -0800)]
Fix 3106227: use WeakReferences for receivers in DigitalClock class

This works around a bug in the framework where LockScreen wouldn't
get GC'd under certain circumstances which would lead to an OOM
crash.  It now uses WeakReferences for observers inside the
DigitalClock container class and unregisters them if the containing
DigitalClock goes away.

Also removed mLive variable which was unused and could potentially
leak the receivers.

Left mAttached for debugging so we can use it to determine if the
calls to onAttachToWindow() and onDetachFromWindow() are grossly
unbalanced which may be the root cause of the original problem.

Have cleanUp() explicitly clear unused references to make
tracing through hprof references easier.

Change-Id: I99a7e0c356001b05eab5aa729564553666febfea

13 years agoMerge "Squash commits of the following patches, cherry-picked from other branch ...
James Dong [Sat, 15 Jan 2011 00:27:48 +0000 (16:27 -0800)]
Merge "Squash commits of the following patches, cherry-picked from other branch - do not merge." into gingerbread

13 years agoMerge "Do Not Merge: Expose insecure rfcomm Bluetooth API." into gingerbread
Jaikumar Ganesh [Sat, 15 Jan 2011 00:11:41 +0000 (16:11 -0800)]
Merge "Do Not Merge: Expose insecure rfcomm Bluetooth API." into gingerbread

13 years agoMerge "Added check to make orientation calculations more robust" into gingerbread
Dianne Hackborn [Fri, 14 Jan 2011 23:36:26 +0000 (15:36 -0800)]
Merge "Added check to make orientation calculations more robust" into gingerbread

13 years agoMerge "Make the Phone options dialog use current language" into gingerbread
Dianne Hackborn [Fri, 14 Jan 2011 23:33:44 +0000 (15:33 -0800)]
Merge "Make the Phone options dialog use current language" into gingerbread

13 years agoDo Not Merge: Expose insecure rfcomm Bluetooth API.
Jaikumar Ganesh [Fri, 14 Jan 2011 06:50:51 +0000 (22:50 -0800)]
Do Not Merge: Expose insecure rfcomm Bluetooth API.

This complements the secure rfcomm API.
The link key is unauthenticated and is subject to MITM attacks.
The link key may be encrypted depending on the type of Bluetooth device.
This helps apps which don't need the extra security or
have their own security layer built on top of the rfcomm link.

Bug: 3352266

Change-Id: I633fd0372e5e23288d6fec950dd1abc2896031f1

13 years agoMake the Phone options dialog use current language
Anders Hammar1 [Fri, 14 Jan 2011 09:00:39 +0000 (10:00 +0100)]
Make the Phone options dialog use current language

The title of the Phone options dialog is displayed using wrong
translation if the user changes the current language setting. Moving
the setTitle call to prepareDialog to ensure that the title gets
updated before the dialog is shown.

Change-Id: I03ff59c7f4ff711a06b05de7cca94fa928cf67ef

13 years agoAdded check to make orientation calculations more robust
Niclas Kellgren [Fri, 14 Jan 2011 08:11:35 +0000 (09:11 +0100)]
Added check to make orientation calculations more robust

Added check avoid a division by zero resulting in NaN which in turn
makes checkFullyTilted to ignore high tilt angles from then on.
If (x, y, z) == (0, 0, 0) then there is no tilt or rotation and
this vector must be ignored. This check is extended to ignore all
small acceleration values where noise can be of big influence.

Low or zero readings can happen when space travelling free falling,
but more commonly when shaking or getting bad readings from the sensor.
The accelerometer is turned off when not used and polling it too soon
after it is turned on may result in (0, 0, 0).

Change-Id: I19aec653abb8ab6f7126778035c8c96449f1326f