OSDN Git Service

android-x86/system-bt.git
5 years agoUprev libchrome to r576279 (1/multiple)
Jakub Pawlowski [Mon, 23 Jul 2018 17:00:25 +0000 (10:00 -0700)]
Uprev libchrome to r576279 (1/multiple)

This patch brings the latest and greatest features of libchrome to
android. It contains ~2600 patches.
Reason for uprev: libbluetooth want to use some of the most recent
features avaliable.

Test: libchrome_test
Change-Id: I5f8d5931b404767d3b86cb3bd8df2a05dd67a768
Merged-In: I5f8d5931b404767d3b86cb3bd8df2a05dd67a768

5 years agoUse proper naming for Identity Address
Jakub Pawlowski [Fri, 12 Oct 2018 14:10:08 +0000 (16:10 +0200)]
Use proper naming for Identity Address

During Bonding, devices exchange "Identity Address Information", which
include Identity Address, and Identity Address Type.

Identity Address is either Public Address, or Static Random Address.

Thorugh stack, we improperly use "static" where we should use
"identity" to refer to this address.

Bug: 109827460
Test: compilation, just renamed variables.
Change-Id: Iccee1e8ae881e9b9f480e0bf05eea440a905109a

5 years agoMerge "Hearing Aid: Use valid COC service id"
Treehugger Robot [Fri, 12 Oct 2018 17:04:56 +0000 (17:04 +0000)]
Merge "Hearing Aid: Use valid COC service id"

5 years agoMerge "Add bluetooth_test_common to TEST_MAPPING"
Treehugger Robot [Fri, 12 Oct 2018 03:34:30 +0000 (03:34 +0000)]
Merge "Add bluetooth_test_common to TEST_MAPPING"

5 years agoTimer: Fix unstable test
Hansong Zhang [Thu, 11 Oct 2018 21:14:31 +0000 (14:14 -0700)]
Timer: Fix unstable test

* Don't use future.wait_until()
* For cancel_period_task, verify the counter won't change after
  timer is cancelled

Test: run Timer test for 5000 times
Bug: 116081383
Change-Id: Ibc99507e76180a878f417a0b4e7aba1b8adc7cb6

5 years agoHearing Aid: Use valid COC service id
Hansong Zhang [Thu, 11 Oct 2018 01:02:49 +0000 (18:02 -0700)]
Hearing Aid: Use valid COC service id

Test: disconnect hearing aid and let remote device connect to it
Bug: 116317072
Bug: 117423102
Change-Id: Iea001d287256f4d7d9e8213ac8a4f03d78f4bbc5

5 years agoMerge "Allow AVDTP SetConfig from INIT state"
Treehugger Robot [Thu, 11 Oct 2018 18:38:16 +0000 (18:38 +0000)]
Merge "Allow AVDTP SetConfig from INIT state"

5 years agoMerge "Add more Hearing Aid Profile logs"
Treehugger Robot [Thu, 11 Oct 2018 17:14:59 +0000 (17:14 +0000)]
Merge "Add more Hearing Aid Profile logs"

5 years agoAdd more Hearing Aid Profile logs
Stanley Tng [Thu, 11 Oct 2018 01:57:29 +0000 (18:57 -0700)]
Add more Hearing Aid Profile logs

Add more logs to debug the TimeCheck timeout crash issues.

Test: Manual run
Change-Id: I3a05ca975e41ab002e466150f3f3f28ba0f037df

5 years agoAllow AVDTP SetConfig from INIT state
Pavlin Radoslavov [Tue, 25 Sep 2018 01:42:53 +0000 (18:42 -0700)]
Allow AVDTP SetConfig from INIT state

This addresses AVDTP codec reconfiguration issue on the A2DP Sink side.
Apparently, the A2DP Sink rejects AVDTP Reconfig - then, after the AVDTP Close
(normal), the Sink side doesn't accept AVDTP SetConfig anymore.
If the codecs are already known to the A2DP SRC, it doesn't need the
AVDTP Discover mechanism just for the purpose of the codec reconfiguration.
Hence, we can allow SetConfig from INIT state.

Bug: 115894702
Test: Manual with switching the AAC Sample Rate in the Develop Options
Change-Id: I3b1603dc6ef344eea50383df2eddf309dec80a28

5 years agoLE Enhanced Connection Complete: Check address types
Jakub Pawlowski [Tue, 14 Aug 2018 10:24:33 +0000 (12:24 +0200)]
LE Enhanced Connection Complete: Check address types

Only pass identity addresses to btm_identity_addr_to_random_pseudo.
Only pass random addresses into btm_ble_resolve_random_addr.

Test: bond twice to device using non-resolvable address
Change-Id: I5931e5bc89449ea4ae9b873aedfd7bbcdeef2e01

5 years agoMerge "Add DataElementReader for SDP"
Ajay Panicker [Tue, 9 Oct 2018 18:53:02 +0000 (18:53 +0000)]
Merge "Add DataElementReader for SDP"

5 years agoFix Timer unittest
Hansong Zhang [Sun, 7 Oct 2018 20:31:01 +0000 (13:31 -0700)]
Fix Timer unittest

* Never set promise more than once
* To verify running task cannot be cancelled, make sure Cancel() is
  called after callback starts
* Remove a stress test case. Sometimes there is unexpected event causing
  deviation.

Test: run Timer test for 5000 times
Bug: 116081383
Change-Id: If99e4958289e9b6aaabc51b15ce4ac035a3ff1e9

5 years agoMerge "Replace A2DP source osi/alarm with common/Timer"
Treehugger Robot [Mon, 8 Oct 2018 23:38:02 +0000 (23:38 +0000)]
Merge "Replace A2DP source osi/alarm with common/Timer"

5 years agoMerge "Move btif_state_machine to common/StateMachine"
Treehugger Robot [Mon, 8 Oct 2018 22:34:10 +0000 (22:34 +0000)]
Merge "Move btif_state_machine to common/StateMachine"

5 years agoMerge "Fix race condition around Le Start Encryption handling."
Treehugger Robot [Mon, 8 Oct 2018 22:17:28 +0000 (22:17 +0000)]
Merge "Fix race condition around Le Start Encryption handling."

5 years agoReplace A2DP source osi/alarm with common/Timer
Hansong Zhang [Thu, 23 Aug 2018 22:17:10 +0000 (15:17 -0700)]
Replace A2DP source osi/alarm with common/Timer

Test: A2DP stream music
Change-Id: I4f7325b3eb3be07eca6c599c7f6f5e037b8e7a64

5 years agoFix race condition around Le Start Encryption handling.
Jakub Pawlowski [Mon, 1 Oct 2018 11:25:11 +0000 (13:25 +0200)]
Fix race condition around Le Start Encryption handling.

When sending Le Start Encryption, it's possible that at the same exact
moment the connection was disconnected, and there is an incoming
"Disconnection Complete Event". The controller would send error 0x02
"unknown connection identifier" in such case. This can happen
sporadically when reconnecting device multiple times.

Currently, we don't handle this error code. This lead to the timeout
when waiting for "Encryption Complete", and unbonding of the device.

From now on, if 0x02 error code is returned, we'll cancel current SMP
operation, which is the "LE Start Encryption". This will stop the SMP
timer, and make sure the bond to remote device is not lost.

Test: Manual test reconnecting with HID device multiple time in a row,
causing the link drop right after connection, by removing battery, or
closing the shield box.
Bug: 113652889
Change-Id: I2ff9c13dbc8e7b71505908996e26b89fa1ea6a42

5 years agoAdd DataElementReader for SDP
Ajay Panicker [Fri, 14 Sep 2018 22:10:13 +0000 (15:10 -0700)]
Add DataElementReader for SDP

DataElementReader extracts DataElement's from packets which are the
basic container for all data in the SDP protocol.

Bug: 114751344
Test: run host test bluetooth_test_sdp
Change-Id: If01f22d81af1b80e7c9ac8993ee161eba779bfa3

5 years agoMerge "Stop A2DP stream when shutting down A2DP source."
Treehugger Robot [Mon, 8 Oct 2018 11:06:32 +0000 (11:06 +0000)]
Merge "Stop A2DP stream when shutting down A2DP source."

5 years agoAdd bluetooth_test_common to TEST_MAPPING
Hansong Zhang [Fri, 5 Oct 2018 23:21:09 +0000 (16:21 -0700)]
Add bluetooth_test_common to TEST_MAPPING

Bug: 116081383
Test: unit test
Change-Id: I8231b8f168ab1d86009b7be2cf0b6044f5c8d0cf

5 years agoMove btif_state_machine to common/StateMachine
Hansong Zhang [Thu, 4 Oct 2018 22:04:18 +0000 (15:04 -0700)]
Move btif_state_machine to common/StateMachine

In Bluetooth native infrastructure refactor plan, we need a state
machine class for entire stack

Test: manual
Change-Id: I0e6671e42570482d9a9466b117054b3aadb3e9cc

5 years agoStop A2DP stream when shutting down A2DP source.
weichinweng [Wed, 3 Oct 2018 07:08:54 +0000 (15:08 +0800)]
Stop A2DP stream when shutting down A2DP source.

When shutting down a2dp source,
A2DP source set StateShuttingDown state too fast cause the A2DP doesn’t stop stream.
Will try to stop A2DP when doing A2DP source end session.

Bug: 112279820
Test: Build Pass, BT off/on when play A2DP
Change-Id: I6bfb5c235da1ff4a7e6814bae3a98ae7dd1788aa

5 years agoMerge "MessageLoopThread: Release lock during blocked StartUp/ShutDown"
Treehugger Robot [Thu, 4 Oct 2018 06:10:47 +0000 (06:10 +0000)]
Merge "MessageLoopThread: Release lock during blocked StartUp/ShutDown"

5 years agoMessageLoopThread: Release lock during blocked StartUp/ShutDown
Hansong Zhang [Sat, 29 Sep 2018 21:23:31 +0000 (14:23 -0700)]
MessageLoopThread: Release lock during blocked StartUp/ShutDown

During StartUp/ShutDown(), if another running task in message loop needs
api_lock, then it will deadlock. Release api_lock in StartUp and
ShutDown when current thread is blocked by another task

Bug: 116081383
Test: run unit test
Change-Id: Ic34406d0cfc6364ac965c4f1aa917a5ccaf6b373

5 years agoTimer: Set data member before scheduling task
Hansong Zhang [Fri, 28 Sep 2018 19:10:15 +0000 (12:10 -0700)]
Timer: Set data member before scheduling task

Also increase tolerance for unit tests

Bug: 116081383
Test: Run unit test
Change-Id: I5531c2259cbbc58b6ee7527813e8d5c4164e94c5

5 years agoHearing Aid: use correct Data Length and MediaType
Stanley Tng [Tue, 25 Sep 2018 18:15:24 +0000 (11:15 -0700)]
Hearing Aid: use correct Data Length and MediaType

Adjusted the requested data length to the BT Controller from 168 to 167.
Also, change the default media type during playback to unknown.

Test: Manual testing with Hearing Aids and verifying with OTA logs
Change-Id: Ic4e934729a98bef5c10ebca64093d8a192e79a01

5 years agoMerge "Hearing Aid: Clean up GATT on disconnect"
Treehugger Robot [Fri, 28 Sep 2018 19:08:21 +0000 (19:08 +0000)]
Merge "Hearing Aid: Clean up GATT on disconnect"

5 years agoMerge "Rename functions related to BLE scan cache"
Treehugger Robot [Fri, 28 Sep 2018 14:22:08 +0000 (14:22 +0000)]
Merge "Rename functions related to BLE scan cache"

5 years agoRename functions related to BLE scan cache
Jakub Pawlowski [Wed, 8 Aug 2018 13:19:58 +0000 (15:19 +0200)]
Rename functions related to BLE scan cache

They are not related to GATT, update naming.

Test: compilation
Bug: 67058417
Change-Id: Ia84ed98e02fbc04ecfde979f860b8766b0018218

5 years agoHearing Aid: Clean up GATT on disconnect
Hansong Zhang [Thu, 27 Sep 2018 05:02:38 +0000 (22:02 -0700)]
Hearing Aid: Clean up GATT on disconnect

When user requests Disconnect(), after disconnecting GATT,
OnGattDisconnected() never cleans up GATT queue, because the
corresponding device is already removed from HearingDevices. We need to
clean up the queue before connecting the same device again. Without the
fix, sometimes ACL is connected, but the host failed to read GATT
characteristic to open COC. Also clean up connection update status
during disconnect

Bug: 115365334
Test: connect to both devices -> disconnect -> connect, and repeat
Change-Id: I158b19b3d1f2cb2123ec554eb1ffe8c842df998e

5 years agoAdd TEST_MAPPING to system/bt
Jack He [Wed, 29 Aug 2018 01:28:55 +0000 (18:28 -0700)]
Add TEST_MAPPING to system/bt

* Add "postsubmit" test group to TEST_MAPPING in system/bt
* Developer can run tests mannually in system/bt via:
  system/bt$ atest :postsubmit
* Remove legacy configuration files spread across system/bt
* Created unified configuration file system/bt/AndroidTestTemplate.xml
  and add it to the default build config so that developers don't have
  to configure it

Bug: 113359343
Test: atest :postsubmit
Change-Id: I0fdd55b47ef176b5c4ce3d2cd217ffa67cf9bc23

5 years agoMerge "GATTC: Use hex for conn_id in log"
Treehugger Robot [Thu, 27 Sep 2018 22:16:39 +0000 (22:16 +0000)]
Merge "GATTC: Use hex for conn_id in log"

5 years agoMerge "HID: Free timer after removing device"
Treehugger Robot [Thu, 27 Sep 2018 20:47:50 +0000 (20:47 +0000)]
Merge "HID: Free timer after removing device"

5 years agoGATTC: Use hex for conn_id in log
Hansong Zhang [Thu, 27 Sep 2018 00:42:40 +0000 (17:42 -0700)]
GATTC: Use hex for conn_id in log

For host conn_id, use hex format

Test: manual
Change-Id: I7d7e690a1dbe0665b827c59d104540eafa45d426

5 years agoHID: Free timer after removing device
Hemant Gupta [Thu, 27 Sep 2018 16:05:46 +0000 (21:35 +0530)]
HID: Free timer after removing device

Timer conn.process_repage_timer is being freed in HID_HostDeregister
which causes use after free condition in HID_HostRemoveDev->
HID_HostCloseDev.

Fix:
Call alarm_free after HID_HostRemoveDev in HID_HostDeregister

Test:
Code analysis

Bug: 116774368
Change-Id: Iddecd126716a5660f4c37ddc917bacee84342297

5 years agoMove Bluetooth to using C++17
Ajay Panicker [Mon, 17 Sep 2018 18:35:24 +0000 (11:35 -0700)]
Move Bluetooth to using C++17

C++ 17 has a bunch of features that will help us refactor the Bluetooth
stack to be much more safe and stable.

Bug: 115895445
Test: Compile and testtracker/201115
Change-Id: Ia1e8e179fa1df116969606d119cb8abe11f44761

5 years agoMerge "Add noexcept to move constructors and assignment operators."
Chih-hung Hsieh [Thu, 27 Sep 2018 04:33:34 +0000 (04:33 +0000)]
Merge "Add noexcept to move constructors and assignment operators."

5 years agoMerge "Use maximum SSR timeout value when A2DP is streaming"
Treehugger Robot [Thu, 27 Sep 2018 01:58:13 +0000 (01:58 +0000)]
Merge "Use maximum SSR timeout value when A2DP is streaming"

5 years agoMerge "Remove obsoleted Coverity-related comment"
Treehugger Robot [Wed, 26 Sep 2018 23:18:32 +0000 (23:18 +0000)]
Merge "Remove obsoleted Coverity-related comment"

5 years agoUse maximum SSR timeout value when A2DP is streaming
Ugo Yu [Thu, 13 Sep 2018 12:12:36 +0000 (20:12 +0800)]
Use maximum SSR timeout value when A2DP is streaming

* Add a new SSR entry BTA_DM_PM_SSR4 with maximum
  remote/local timeout value.
* Use BTA_DM_PM_SSR4 when A2DP is streaming to prevent
  phone enter sniff subrating mode.

Bug: 113496412
Test: Manually
Change-Id: I194ef9fdb4dd38eab67ff20e1bb9cab47edbfe8f

5 years agoAdd new hidden API to return SDP record's MapSupportedFeatures field.
Vasu Nori [Sat, 18 Aug 2018 00:28:11 +0000 (17:28 -0700)]
Add new hidden API to return SDP record's MapSupportedFeatures field.
That field is documented in Bluetooth MAP 1.4 spec, page 114.
Spec is here: http://shortn/_zfoIrfbIzR

Bug: 111614861
Test: tested w/ KitchenSink App
Change-Id: I89903c386d97934ec7b9c80333ecc6a6d560cbec

5 years agoAdd noexcept to move constructors and assignment operators.
Chih-Hung Hsieh [Tue, 25 Sep 2018 21:29:59 +0000 (14:29 -0700)]
Add noexcept to move constructors and assignment operators.

Bug: 116614593
Test: build with WITH_TIDY=1
Change-Id: I4adaf89db5001e9b627beac8160f62b50513d18f

5 years agoRemove obsoleted Coverity-related comment
Pavlin Radoslavov [Tue, 25 Sep 2018 18:18:34 +0000 (11:18 -0700)]
Remove obsoleted Coverity-related comment

Bug: 27078725
Test: Code compilation
Change-Id: I9398f0354e54cc6a754ed46b05b9f7cef09add8f

5 years agoMerge changes I737a864e,Idc41b2a5
Treehugger Robot [Tue, 25 Sep 2018 04:41:08 +0000 (04:41 +0000)]
Merge changes I737a864e,Idc41b2a5

* changes:
  Timer: Don't run new task at old timepoint
  Timer: Separate code path for single and periodic

5 years agoTimer: Don't run new task at old timepoint
Hansong Zhang [Thu, 20 Sep 2018 17:15:12 +0000 (10:15 -0700)]
Timer: Don't run new task at old timepoint

* Fix a race condition that the task is just finished but IsScheduled
  is true
* Fix a bug where cancelling the old task will make new task run at
  the previous task's scheduled timepoint
* Protect racing read/write to message_loop_thread_ in RunTask() and
  Cancel() by storing a local reference to message_loop_thread_ in
  RunTask()

Bug: 116081383
Test: Run bluetooth_test_common.TimerTest
      reschedule_task_dont_invoke_new_task_early
      schedule_task
      for multiple times
Change-Id: I737a864e1b061adbcd41245711e0d3e8adf448d4

5 years agoMerge "Send volume just once when starting the stream."
Treehugger Robot [Mon, 24 Sep 2018 20:33:51 +0000 (20:33 +0000)]
Merge "Send volume just once when starting the stream."

5 years agoSend volume just once when starting the stream.
Jakub Pawlowski [Mon, 24 Sep 2018 14:43:59 +0000 (16:43 +0200)]
Send volume just once when starting the stream.

The volume is in the Start command already, no need to send it in a
separate command.

Bug: 116145308
Test: play music, pause, adjust volume, start playing again, verify
      playing was re-started with proper volume level.
Change-Id: I2a900b03a35670de4a265184f333517db17497d8

5 years agoTimer: Separate code path for single and periodic
Hansong Zhang [Thu, 20 Sep 2018 17:15:12 +0000 (10:15 -0700)]
Timer: Separate code path for single and periodic

* Add helper RunSingleTask() and RunPeriodicTask() to handle two cases
  separately
* In RunTask(), check that it must run on specified message loop thread

Bug: 116081383
Test: Run bluetooth_test_common
Change-Id: Idc41b2a509f43dbe946b2e26f2afd775726514aa

5 years agoMerge "Add net_test_avrcp to general-tests suite"
Treehugger Robot [Wed, 19 Sep 2018 00:12:50 +0000 (00:12 +0000)]
Merge "Add net_test_avrcp to general-tests suite"

5 years agoFix clang-tidy performance warnings in system/bt.
Chih-Hung Hsieh [Mon, 17 Sep 2018 22:16:18 +0000 (15:16 -0700)]
Fix clang-tidy performance warnings in system/bt.

* Use more efficient overloaded string methods.

Bug: 30411878
Test: build with WITH_TIDY=1
Change-Id: Iff759bf68ede3bb6bdd8e332b3490128eee3b5fe

5 years agoSuppress implicit-fallthrough warnings.
Chih-Hung Hsieh [Wed, 12 Sep 2018 22:20:43 +0000 (15:20 -0700)]
Suppress implicit-fallthrough warnings.

Add FALLTHROUGH_INTENDED for clang compiler.

Bug: 112564944
Test: build with global -Wimplicit-fallthrough.
Change-Id: Iedf7e46445101a5e3358473e1200212d67cf58d1

5 years agoMerge "AVRCP Controller media session queue"
Joseph Pirozzo [Fri, 14 Sep 2018 22:48:55 +0000 (22:48 +0000)]
Merge "AVRCP Controller media session queue"

5 years agoAdd net_test_avrcp to general-tests suite
Dan Shi [Fri, 14 Sep 2018 17:38:26 +0000 (10:38 -0700)]
Add net_test_avrcp to general-tests suite

This is the first step for it to run in lab. With this, the test binary
will be included in general-tests.zip (test_suite* target). From there,
we can kick off the test in the lab using TF, with the right options.

Bug: 111921389
Test: m -j net_test_avrcp
./host/linux-x86/testcases/net_test_avrcp/x86_64/net_test_avrcp

Change-Id: I4de9409b4a9d4c066a9024a33d6a980ecb451dc1

5 years agoAdd Timer as an alternative to osi alarm
Hansong Zhang [Tue, 14 Aug 2018 21:29:23 +0000 (14:29 -0700)]
Add Timer as an alternative to osi alarm

* Add a private method MessageLoopThread.DoInThreadDelayed to post a
  delayed task in message loop, as an alternative approach to osi alarm
  clock
* Add a unit test for MessageLoopThread to check ShutDown() waits until
  current task finishes
* Add Timer using MessageLoopThread.DoInThreadDelayed
* Timer provides similar API as osi alarm, and uses same OS clock (boot
  timer) as alarm
* Add benchmark and unit tests to ensure the performance is comparable
  to the existing osi alarm

Test: Run unit test and benchmark test
      ./test/run_unit_tests.sh bluetooth_test_common
      ./test/run_benchmarks.sh bluetooth_benchmark_timer_performance
      --benchmark_repetitions=10 --benchmark_report_aggregates_only=true
Bug: 110303473
Change-Id: I6f2e7ae2f80f9889fc5fe3c8cd6b9b2670938b46

5 years agoMerge "Add support for A2DP LDAC Sink function"
Treehugger Robot [Thu, 13 Sep 2018 01:30:08 +0000 (01:30 +0000)]
Merge "Add support for A2DP LDAC Sink function"

5 years agoMerge "Add extractBE to the Bluetooth Packets Iterators"
Ajay Panicker [Wed, 12 Sep 2018 22:27:42 +0000 (22:27 +0000)]
Merge "Add extractBE to the Bluetooth Packets Iterators"

5 years agoAdd support for A2DP LDAC Sink function
Chisato Kenmochi [Fri, 10 Aug 2018 11:44:05 +0000 (20:44 +0900)]
Add support for A2DP LDAC Sink function

Note: The LDAC decoder source code is needed to run with this patch.

Bug: 112839060
Test: Manual
Change-Id: Idce0070dcce4d6b83cfd08213b846d957d694d26
Signed-off-by: Chisato Kenmochi <Chisato.Kenmochi@sony.com>
5 years agoAdd a check whether an A2DP codec is supported before including it
Pavlin Radoslavov [Tue, 11 Sep 2018 23:47:47 +0000 (16:47 -0700)]
Add a check whether an A2DP codec is supported before including it

Some of the A2DP SRC or SNK codecs might not be supported if the
corresponding shared library is not included.
Don't include such codecs when configuring and registering
the stream end points on the local device.

Bug: 112839060
Test: Manual - remove A2DP codec shared libraries and check the
      returned AVDTP Capabilities
Change-Id: Ia8fbdac915b208f9950ef7c9f50e3e0dc87e9111

5 years agoAdd extractBE to the Bluetooth Packets Iterators
Ajay Panicker [Mon, 10 Sep 2018 19:01:50 +0000 (12:01 -0700)]
Add extractBE to the Bluetooth Packets Iterators

ExtractBE allows data to be extracted from Bluetooth packets in a big
endian format.

Bug: 114751344
Test: run net_test_btpackets
Change-Id: Ib01d726893e106d34c11048acc09fee201d81c09

5 years agoMerge "Fix spacing in hearing aid dumpsys"
Treehugger Robot [Mon, 10 Sep 2018 23:40:29 +0000 (23:40 +0000)]
Merge "Fix spacing in hearing aid dumpsys"

5 years agoDisable absolute volume control for deepblue2 speaker
Pavlin Radoslavov [Sun, 9 Sep 2018 18:06:21 +0000 (11:06 -0700)]
Disable absolute volume control for deepblue2 speaker

The speaker does not accept the volume levels from the Phone.
It responds back with rounded down volume levels.

Bug: b/37834035
Test: Manual
Change-Id: Ie5575233c3cf31c087addd3cef1c12ad36b820b0

5 years agoFix spacing in hearing aid dumpsys
Hansong Zhang [Fri, 7 Sep 2018 16:39:22 +0000 (09:39 -0700)]
Fix spacing in hearing aid dumpsys

Bug: 114242876
Test: See dumpsys output for bluetooth_manager
Change-Id: Ifb0884ba4e90e2b925bafc52a15d3375c9aec288

5 years agoAllocate lcb when connection complete if can't find one
Chienyuan [Wed, 29 Aug 2018 05:38:19 +0000 (13:38 +0800)]
Allocate lcb when connection complete if can't find one

When BT stack sends create connection command and receive connection
request at the same time, connection request will reuse the link control
block created by create connection command. But the link control block
will be released before connection complete due to controller busy.

Bug:112818458
Test: manual - turn on BT and turn on Headset within little time gap
Change-Id: Ie30c655b933250f1709ba10c578ea5f30ea13139

5 years agoImplement A2DP source and AVRCP target binder servers
Bailey Forrest [Wed, 22 Aug 2018 00:20:29 +0000 (17:20 -0700)]
Implement A2DP source and AVRCP target binder servers

Bug: 64231556
Test: On device.
Change-Id: I57136d5550e8e52626a1e7eb26b7d8c02cd94eb5

5 years agoMerge "[bluetoothtb] Fix bug processing BT_PROPERTY_UUIDS"
Treehugger Robot [Wed, 29 Aug 2018 00:32:15 +0000 (00:32 +0000)]
Merge "[bluetoothtb] Fix bug processing BT_PROPERTY_UUIDS"

5 years agoAVRCP Controller media session queue
Joseph Pirozzo [Tue, 28 Aug 2018 22:33:47 +0000 (15:33 -0700)]
AVRCP Controller media session queue

Add support for the Now Playing Content available through AVRCP to
receive proper notifications when it changes and expose it via media
session rather than a node in the browse tree.

Bug: 78598689
Test: Stream Music, change songs and playlist, verify media session
queue.

Change-Id: I46c32d05c5a674388c9e59830ec0596432cd78ef

5 years agoFix out-of-bounds reading in btm_ble_vendor_capability_vsc_cmpl_cback
Yamei Du [Mon, 11 Sep 2017 11:10:47 +0000 (19:10 +0800)]
Fix out-of-bounds reading in btm_ble_vendor_capability_vsc_cmpl_cback

8 bits parameters get 16 bits from buffer in btm_ble_vendor_
capability_vsc_cmpl_cback, which would cause out-of-bounds reading.
Correct STREAM_TO_UINT16 to STREAM_TO_UINT8

Test: manual
Bug: 112167084
Change-Id: Id2b71a8f74eee5d93f287d4705aadab449c55fce

5 years agoDon't add items to a response if a previous item fails to be added
Ajay Panicker [Wed, 22 Aug 2018 23:24:03 +0000 (16:24 -0700)]
Don't add items to a response if a previous item fails to be added

Bug: 112164711
Test: Run native test net_test_avrcp
Change-Id: Iace287bb716b28eda62a8a0500deb99cc46ae485

5 years agoMerge "Set the AVRCP Browsing MTU for a device after browse open"
Treehugger Robot [Tue, 28 Aug 2018 18:38:17 +0000 (18:38 +0000)]
Merge "Set the AVRCP Browsing MTU for a device after browse open"

5 years agoSet the AVRCP Browsing MTU for a device after browse open
Ajay Panicker [Fri, 24 Aug 2018 17:32:07 +0000 (10:32 -0700)]
Set the AVRCP Browsing MTU for a device after browse open

Trying to get the Browsing MTU before the Browsing channel is opened
returns the minimum MTU. This leads to smaller browsing messages than
expected.

Bug: 113160447
Test: Connect to device and see that Avrcp Browse packets are the
correct size

Change-Id: Iefda855d0a777439a8a6829619bc4c5543b62eaf

5 years agoMake thread names more meaningful
Jack He [Mon, 27 Aug 2018 20:55:14 +0000 (13:55 -0700)]
Make thread names more meaningful

* Rename threads:
  - module_wrapper -> bt_module_lifecycle_thread
  - btif_a2dp_sink_worker_thread -> bt_a2dp_sink_worker_thread
  - btif_a2dp_source_thread -> bt_a2dp_source_worker_thread
  - bt_jni_workqueue -> bt_jni_thread
  - stack_manager -> bt_stack_manager_thread
  - hci_thread -> bt_hci_thread
  - bt_workqueue -> bt_startup_thread
  - btu_message_loop -> bt_main_thread
* Rename functions:
  - do_in_bta_thread -> do_in_main_thread
  - get_message_loop -> get_main_message_loop
  - post_to_hci_message_loop -> post_to_main_message_loop
* Refactored bta_sys_sendmsg

Bug: 110303473
Test: make, unit test, stream music
Change-Id: I3051f6ea2d33db37af8eebafba7db4d1e113f611

5 years agoCommon: Replace ExecutionBarrier with std::promise and std::future
Jack He [Sat, 25 Aug 2018 00:14:10 +0000 (17:14 -0700)]
Common: Replace ExecutionBarrier with std::promise and std::future

* std::promise and std::future are able to achieve the same
  functionalities of ExecutionBarrier with extra flexibility
* Replace "_barrier" with "_promise" in system/bt/common

Bug: 110303473
Fixes: 112159657
Test: mm -j40, unit test
Change-Id: I2a420bbf16bf92e4b3dd256d9f23480fc2be7be1

5 years agoRun the wakelock JNI callbacks on the JNI thread
Pavlin Radoslavov [Sat, 25 Aug 2018 01:45:16 +0000 (18:45 -0700)]
Run the wakelock JNI callbacks on the JNI thread

All JNI callbacks must be on the JNI thread, otherwise this could
impact the threads' operation.

Bug: 111277984
Test: A2DP streaming
Change-Id: I94fa4500aea62a574c80749e904434e46eb4e942

5 years agoMerge "Move metrics and time library to libbt-common"
Treehugger Robot [Mon, 27 Aug 2018 19:03:21 +0000 (19:03 +0000)]
Merge "Move metrics and time library to libbt-common"

5 years agoMove metrics and time library to libbt-common
Jack He [Wed, 15 Aug 2018 19:38:37 +0000 (12:38 -0700)]
Move metrics and time library to libbt-common

* libbt-osi is being deprecated and libbt-common will replace it
* Move recently implemented metrics library to libbt-common
* Move time library from libbt-osi to libbt-common and renamed as time_util
  to avoid potention name collision with system time libraries
* Need to keep time library for now because we need to use
  CLOCK_BOOTTIME that include system suspension time for A2DP encoding
* Use libchrome logging methods in metrics library
* Put both metrics and time_util into bluetooth::common namespace
* Return uint64_t for time_get_os_boottime_ms and update various
  receiver types
* Remove tBTM_CB.max_collision_delay and use the constant
  BTM_SEC_MAX_COLLISION_DELAY instead

Bug: 111568640
Test: mm -j40, unit test, stream music
Change-Id: I8c384a810d592bb6b9eb322134e947d066489ba4

5 years agoFix V535 CWE-691: inner/outer for loop sharing variables
Ruchi Kandoi [Fri, 17 Aug 2018 20:30:06 +0000 (13:30 -0700)]
Fix V535 CWE-691: inner/outer for loop sharing variables

Test: compiles
Bug: 112146072
Change-Id: Ie9e10d727edc652865458298cb59a24ac5732c98

5 years ago[bluetoothtb] Fix bug processing BT_PROPERTY_UUIDS
Bailey Forrest [Sat, 25 Aug 2018 00:04:00 +0000 (17:04 -0700)]
[bluetoothtb] Fix bug processing BT_PROPERTY_UUIDS

BT_PROPERTY_UUIDS returns an array of UUID objects

Bug: None
Test: Manual
Change-Id: I9ace472b8bddb5823dc7309528d94faba3d0dfe2

5 years agoMerge "L2CAP: Handle no idle timeout (0xFFFF)"
Treehugger Robot [Thu, 23 Aug 2018 22:54:16 +0000 (22:54 +0000)]
Merge "L2CAP: Handle no idle timeout (0xFFFF)"

5 years agoL2CAP: Handle no idle timeout (0xFFFF)
Nitin Shivpure [Mon, 30 Jul 2018 12:44:54 +0000 (18:14 +0530)]
L2CAP: Handle no idle timeout (0xFFFF)

Precondition:
1. DUT acts as GATT server.
2. Remote acts as Gatt Client

Use case:
1. Remote device starts BLE scan
2. Remote device connects to DUT
3. Remote device keeps on sending notifications to DUT

Expected result:
Connection should be active after 18 hours

Observed Result:
Connection is dropped after 18 hours due to l2c_link_timeout.

Root Cause:
L2CAP is not properly handling no idle timeout(0xFFFF) and 65535000 ms
timer gets started after BLE connection, leading to connection getting dropped
after 18 hours.

Fix:
If there is no L2CAP dynamic ccb and no idle timeout is configured for
GATT or any other fixed  channel, then cancel existing LCB timer.

Test: No BLE disconnection seen after 18 hours

Bug: 112249120
Change-Id: I58766564cf62efb7af0869035115459bae37c558

5 years agoClear connection update pending flag when disconnected
Stanley Tng [Wed, 22 Aug 2018 21:43:57 +0000 (14:43 -0700)]
Clear connection update pending flag when disconnected

For the hearing aid profile and when we get disconnection
while there is a pending connection update, then clear the flag so that
the next connection update can proceed.

Bug: 112050972
Test: Manual test to check for regressions
Change-Id: I54635a680cf276cf6c9f94520943c3efa97e77ec

5 years agoMerge "A2DP-SINK: Use MessageLoopThread for decoding"
Treehugger Robot [Tue, 21 Aug 2018 23:47:01 +0000 (23:47 +0000)]
Merge "A2DP-SINK: Use MessageLoopThread for decoding"

5 years agoMerge "Fix "Disable Absolute Volume" Bluetooth developer option"
Ted Wang [Tue, 21 Aug 2018 02:32:49 +0000 (02:32 +0000)]
Merge "Fix "Disable Absolute Volume" Bluetooth developer option"

5 years agoDisable CFI verification for stack_rfcomm test that uses gtest+gmock
Jack He [Mon, 20 Aug 2018 22:46:24 +0000 (15:46 -0700)]
Disable CFI verification for stack_rfcomm test that uses gtest+gmock

Apparently, the CFI verification crashes unit tests that
are using gtest+gmock. Interestingly, those tests succeed for 32-bit
binaries, and fail for 64-bit binaries.

Bug: 71739588
Test: Running the 64-bit version of the unit tests in test/run_unit_tests.sh
Change-Id: Ic950087181a86d08af1b858272e946ef4125b78a

5 years agoA2DP-SINK: Use MessageLoopThread for decoding
Jack He [Thu, 16 Aug 2018 08:29:50 +0000 (01:29 -0700)]
A2DP-SINK: Use MessageLoopThread for decoding

* Instead of OSI thread, use MessageLoopThread for decoding
* Similar to A2DP source
* Change tBTIF_A2DP_SINK_CB into a proper class BtifA2dpSinkControlBlock
* Initialize variables in BtifA2dpSinkControlBlock's constructor
* Free un-freed variables in BtifA2dpSinkControlBlock::Reset() method

Bug: 110303473
Test: mm -j40, unit test, stream music
Change-Id: I21f5b0c9f88d2cadada6fc854070d31ac378244f

5 years agoFix "Disable Absolute Volume" Bluetooth developer option
Ted [Fri, 17 Aug 2018 10:49:37 +0000 (18:49 +0800)]
Fix "Disable Absolute Volume" Bluetooth developer option

Do not report device support absolute volume while
"Disable Absolute Volume" in developer option is enabled or
the remote device is listed in the interoperability database.

Test: manually adjust volume when Disable Absolute Volume is enabled
Bug: 112193362
Change-Id: Id63a22dba0d8a871035774fc86b66d32e9225b48

5 years agoA2DP Offload: Fix Max AVDTP MTU for Offload
Sunny Kapdi [Thu, 5 Jul 2018 21:31:21 +0000 (14:31 -0700)]
A2DP Offload: Fix Max AVDTP MTU for Offload

Fix the max AVDTP MTU for Offload to what
can fit in a 3DH5 packet after accounting
for AVDTP and L2CAP header.

Also:
 - Moved MAX_2MBPS_AVDTP_MTU and MAX_3MBPS_AVDTP_MTU to the common
   avdt_api.h so it can be used by the rest of the code
 - Removed previously incorrect value of BTA_AV_MAX_A2DP_MTU and
   replace it with the new (correct) value of MAX_3MBPS_AVDTP_MTU
 - Removed duplicate define of MAX_2MBPS_AVDTP_MTU and use
   the one from avdt_api.h

Bug: 78907438
Test: A2DP Streaming
Change-Id: I2c60c34c33e4164c9f7e2e7362bbb600482dec97

5 years agoMerge "Implement A2DP sink and AVRCP controller binder server"
Treehugger Robot [Sat, 18 Aug 2018 20:18:39 +0000 (20:18 +0000)]
Merge "Implement A2DP sink and AVRCP controller binder server"

5 years agoImplement A2DP sink and AVRCP controller binder server
Bailey Forrest [Tue, 31 Jan 2017 01:38:58 +0000 (17:38 -0800)]
Implement A2DP sink and AVRCP controller binder server

Also added bonded device management because this is needed for A2DP to
work with most devices.

Bug: 32657860
Test: Run on device, modified tests pass.
Change-Id: I862e41b6594b7cec9f61094d300577574cc8bbc6

5 years agoMerge "Replace period_ms_t with uint64_t"
Treehugger Robot [Sat, 18 Aug 2018 00:30:46 +0000 (00:30 +0000)]
Merge "Replace period_ms_t with uint64_t"

5 years agoReplace period_ms_t with uint64_t
Jack He [Thu, 16 Aug 2018 05:17:23 +0000 (22:17 -0700)]
Replace period_ms_t with uint64_t

* There is no need to define a unit for millisecond in our system since
  uint64_t is the largest integer that is supported
* period_ms_t results in many unnecessary dependency on
  osi/include/time.h
* Used text editor to do the directory wise replacement, basically
    sed -i 's/period_ms_t/uint64_t/g' *
* Need to modify several method and variable names to indicate that they
  are using millisecond as time unit

Bug: 110303473
Test: build, no functional change as period_ms_t was a typedef of
      uint64_t

Change-Id: I497c9a146a97bab7c7477b358847dad958b92034

5 years agoMerge "HID: Remove device from database in error case"
Treehugger Robot [Thu, 16 Aug 2018 18:27:34 +0000 (18:27 +0000)]
Merge "HID: Remove device from database in error case"

5 years agoMerge "Enable SNIFF mode for HFP"
Treehugger Robot [Wed, 15 Aug 2018 02:41:17 +0000 (02:41 +0000)]
Merge "Enable SNIFF mode for HFP"

5 years agoMerge "Fix reliable write."
Treehugger Robot [Wed, 15 Aug 2018 00:11:21 +0000 (00:11 +0000)]
Merge "Fix reliable write."

5 years agoMerge "HFP: Only active BT device can hung up call"
Treehugger Robot [Tue, 14 Aug 2018 23:27:34 +0000 (23:27 +0000)]
Merge "HFP: Only active BT device can hung up call"

5 years agoMerge "Check startup_future before using it"
Treehugger Robot [Tue, 14 Aug 2018 23:21:53 +0000 (23:21 +0000)]
Merge "Check startup_future before using it"

5 years agoFix reliable write.
Qiyu Hu [Wed, 13 Jun 2018 15:08:17 +0000 (08:08 -0700)]
Fix reliable write.

We cannot simply assume the write is terminated in reliable write. When
the reliable write value is longer than MTU allows, the current
implementation can only send whatever MTU allows and naively set the
status to GATT_SUCCESS, in the name of "application should verify handle
offset and value are matched or not". That's why MTU negotiation is a
workaround as people mention in b/37031096, which just fits all the write
value into a single request.

This also blocks our test on CtsVerifier.

Bug: 37031096
Test: Manual test and confirm that we don't simply send partial value
Change-Id: I907877608f4672f24c002e630e58bf9133937a5e

5 years agoHID: Remove device from database in error case
Hemant Gupta [Tue, 19 Sep 2017 10:00:56 +0000 (15:30 +0530)]
HID: Remove device from database in error case

Usecase:
1) Pair and connect with remote hid device.
2) Disconnect with remote hid device.
3) Unpair DUT from remote hid device.
4) Send connect request from hid device by pressing any button.
5) Try to pair and connect DUT with remote hid device.

Expected Result
1) Connection should fail after step 4 if remote device does not initaite pairing.
2) Connection should succeed with remote device and hid device should be added to
paired list and hid operations should work properly.

Observed Result
HID operations from remote device starts working without device added to paired list
and UI stucked at pairing.

Root Cause:
After step 4, device should be removed from database when incoming connection is
rejected.

Fix:
Perform virtual unplug for device when connection request for hid is received from
unpaired device.

Test: Issue is not reproducible after stress testing with above change

Bug: 68176469
Change-Id: I9bc72ccc7a849c0d6acf3230a75e1fe95949b3fe

5 years agoEnable SNIFF mode for HFP
Ugo Yu [Tue, 7 Aug 2018 01:55:11 +0000 (09:55 +0800)]
Enable SNIFF mode for HFP

Bug: 112285337
Test: Manual
Change-Id: Ibb019d11cf93051a198df4efb7bd79f31c15a9c0

5 years agoHFP: Only active BT device can hung up call
minle.zuo [Wed, 25 Jul 2018 09:02:51 +0000 (17:02 +0800)]
HFP: Only active BT device can hung up call

Some inactive devices can't send hung up key to host when
HFP connected and active call exist.
And other inactive devices can send hung up call by
sending hung up command.

Solution: To keep consistency of this behavior,
only allow active BT device to hung up call.

Test: Active device can hung up call and inactive can't hung up

Change-Id: Iec1ae689f1d40c15199707e9e925814f71ded619