OSDN Git Service

android-x86/system-bt.git
7 years agoBluetooth 5 advertising duration refactoring (1/4)
Jakub Pawlowski [Fri, 31 Mar 2017 03:41:02 +0000 (20:41 -0700)]
Bluetooth 5 advertising duration refactoring (1/4)

Expose both duration and maximum extended advertising events to limit
advertising time.

Test: manual
Bug: 30622771
Change-Id: I4475c322347899946b39e65026d8009e02c93759

7 years agoAVRCP CT: Only call handle_rc_features if target registered
Bailey Forrest [Fri, 31 Mar 2017 04:43:28 +0000 (21:43 -0700)]
AVRCP CT: Only call handle_rc_features if target registered

If target is not registered, handle_rc_features would clear feature
flags before handle_rc_ctrl_features can use them.

Bug: b/35807779
Test: AVRCP controller track info works on iOS.
Change-Id: Iee781f2fc926a053ba9cf30089d3a49e56b94539

7 years agoMerge "Include flags in all connectable packets"
Treehugger Robot [Thu, 30 Mar 2017 20:53:10 +0000 (20:53 +0000)]
Merge "Include flags in all connectable packets"

7 years agoInclude flags in all connectable packets
Jakub Pawlowski [Thu, 30 Mar 2017 18:52:53 +0000 (11:52 -0700)]
Include flags in all connectable packets

According to Bluetooth CSSv7, the flags field shall be included in all
connectable packets, not only legacy packets.

Test: manal
Bug: 30622771
Change-Id: I006f9bd6d31ca0673d78405f320b18623e6e5b16

7 years agoDelete bt_snoop.log when snoop logging is disabled
Ajay Panicker [Thu, 30 Mar 2017 17:33:19 +0000 (10:33 -0700)]
Delete bt_snoop.log when snoop logging is disabled

This patch will immediatly delete the snoop log if Bluetooth is on. If Bluetooth
is off it will delete the log as soon as Bluetooth starts again.

Test: Turn off logging while Bluetooth is on and off
Bug: 36718948
Change-Id: Ib660cd442c93f1a34c948d51c5cffc38695558d5

7 years agoBluetooth 5 periodic scan (2/3)
Jakub Pawlowski [Wed, 29 Mar 2017 21:28:43 +0000 (14:28 -0700)]
Bluetooth 5 periodic scan (2/3)

This patch wires calls/callbacks through stack to prepare for actual
implementation.

Test: manual
Bug: 30622771
Change-Id: Iac8c74a228f34f5ea4231a76848e22ea64f11d06

7 years agoMerge "Add new internal API: enableOptionalCodecs()/disableOptionalCodecs()"
Treehugger Robot [Wed, 29 Mar 2017 22:41:13 +0000 (22:41 +0000)]
Merge "Add new internal API: enableOptionalCodecs()/disableOptionalCodecs()"

7 years agoMerge "test-vendor-lib: Enable building for 64 bits"
Treehugger Robot [Wed, 29 Mar 2017 20:08:25 +0000 (20:08 +0000)]
Merge "test-vendor-lib: Enable building for 64 bits"

7 years agoAdd new internal API: enableOptionalCodecs()/disableOptionalCodecs()
Pavlin Radoslavov [Fri, 24 Feb 2017 18:35:30 +0000 (10:35 -0800)]
Add new internal API: enableOptionalCodecs()/disableOptionalCodecs()

This API can be used to enable the optional codecs, or disable them
and use only the mandatory SBC.
Internally, it is implemented by raising the SBC priority to
highest (so SBC will be used/selected), or reducing the SBC priority
to its default value (lowest).

Test: A2DP streaming and enabling/disabling/selecting optional codecs
Bug: 35873828
Change-Id: Ia1311d790441b7b5471c9fda7aeee8aa3306b16e

7 years agoAdd missing dependency on the Sony LDAC ABR library libldacBT_abr.so
Pavlin Radoslavov [Wed, 29 Mar 2017 01:40:32 +0000 (18:40 -0700)]
Add missing dependency on the Sony LDAC ABR library libldacBT_abr.so

Test: Code compilation
Bug: 35381097
Change-Id: Ib347319942aef4fef6bfd7edc23a4df004fc19fa

7 years agotest-vendor-lib: Enable building for 64 bits
Jorge E. Moreira [Mon, 27 Mar 2017 22:44:55 +0000 (15:44 -0700)]
test-vendor-lib: Enable building for 64 bits

Removes LOCAL_MODULE_PATH which is not supported for 64 bit builds.
Uses LOCAL_PROPRIETARY_MODULE to let the build system know it should
go in the vendor directory.

Test: builds
Change-Id: Iba55d497d8cbe97911a41a5623248306a8954ec2

7 years agoosi: Update error messages
Myles Watson [Tue, 21 Mar 2017 17:28:20 +0000 (10:28 -0700)]
osi: Update error messages

Test: build
Change-Id: I5af34f01c47d8037d56c27e7e3fab3cc2f5fd2bc

7 years agoMerge "hci: Update error messages"
Treehugger Robot [Tue, 28 Mar 2017 20:37:09 +0000 (20:37 +0000)]
Merge "hci: Update error messages"

7 years agoMerge "l2cap: Use CHECK for memory allocation errors"
Treehugger Robot [Tue, 28 Mar 2017 20:29:26 +0000 (20:29 +0000)]
Merge "l2cap: Use CHECK for memory allocation errors"

7 years agoMerge "BTA HF_CLIENT: Always reply to +BCS"
Treehugger Robot [Tue, 28 Mar 2017 20:03:52 +0000 (20:03 +0000)]
Merge "BTA HF_CLIENT: Always reply to +BCS"

7 years agol2cap: Use CHECK for memory allocation errors
Myles Watson [Tue, 21 Mar 2017 17:16:44 +0000 (10:16 -0700)]
l2cap: Use CHECK for memory allocation errors

Test: LE connection
Change-Id: I43c170b40b125345b91e1c91a7993f0d138ac1c5

7 years agoBTA HF_CLIENT: Always reply to +BCS
Sanket Agarwal [Wed, 22 Feb 2017 19:55:34 +0000 (11:55 -0800)]
BTA HF_CLIENT: Always reply to +BCS

Phones don't like when we don't reply to +BCS codec-negotiation. We were
having this behavior because not always do we want to accept a
connection request over SCO. Instead of not replying to +BCS (which is
essentially a codec negotiation) we rather refuse the SCO request when
it comes in. This creates other issues such as some phones trying to
connect SCO in an infinite loop but it still keeps interoperability with
other devices such as iPhones and Microsoft phones intact

Bug: b/35469963
Bug: b/35431871
Bug: b/35419773

Test: SCO (Phone Audio) test with iPhone, Microsoft and Nexus phones

Change-Id: I5319edcb4da9d66c6d6b3a49742248c90dee98e6

7 years agohci: Update error messages
Myles Watson [Tue, 21 Mar 2017 17:21:15 +0000 (10:21 -0700)]
hci: Update error messages

Test: builds
Change-Id: I33544572122a166523a40dde090f276b102df247

7 years agoIntegration of LDAC Adaptive Bit Rate mechanism
fen wang [Wed, 15 Mar 2017 05:52:52 +0000 (14:52 +0900)]
Integration of LDAC Adaptive Bit Rate mechanism

LDAC ABR controls LDAC encoder bit rate based on A2DP transmit queue
length.
It can be used if the LDAC ABR library is installed on the device:
libldacBT_abr.so

Test: manual
Bug: 35381097
Change-Id: Idcc2fabe86acaf441f5a71f5d2bb66ad29252cdb

7 years agoAdd option to specify initial LE connection PHY (1/3)
Jakub Pawlowski [Sat, 25 Mar 2017 00:52:02 +0000 (17:52 -0700)]
Add option to specify initial LE connection PHY (1/3)

For whitelist connections we always use all possible PHYs, for direct
connection use PHY specified by client.

Test: manual
Bug: 30622771
Change-Id: I720f134e2800dc3d282135bb7ffbe3882117c680

7 years agoMerge "btm_sec: p_name fails matching security records"
Zach Johnson [Fri, 24 Mar 2017 21:19:30 +0000 (21:19 +0000)]
Merge "btm_sec: p_name fails matching security records"

7 years agobtm_sec: p_name fails matching security records
Martin Brabham [Thu, 16 Feb 2017 23:06:30 +0000 (15:06 -0800)]
btm_sec: p_name fails matching security records

When doing a comparison on p_name that is passed in against
the p_name in the struct will never match if the passed in
name is >= 21 characters.  This is due to the strlcpy replacing
the last character with a null termination character.

Without this patch, you will observe 2 security records for
"Android Network Acce" (which is supposed to be "Android Network Access Point")

BTM_SEC_SERVCE_NAME_LEN = 21

                                                    21
passed in: |A|n|d|r|o|i|d| |N|e|t|w|o|r|k| |A|c|c|e|s|s| |P|o|i|n|t|
           ---------------------------------------------------------
structval: |A|n|d|r|o|i|d| |N|e|t|w|o|r|k| |A|c|c|e|\0|

Bug: 34707848
Test: Compile, Manual
Change-Id: I24e0204d8faf54545c7b720a0fbc6765db9a1295

7 years agoExpose connection update callback (1/3)
Jakub Pawlowski [Fri, 24 Mar 2017 01:11:04 +0000 (18:11 -0700)]
Expose connection update callback (1/3)

Test: manual
Bug: 30622771
Change-Id: I94d25f6f22b42fb1432a9288c97b82503d57db86

7 years agohci: Timestamp HCI commands for timeouts
Myles Watson [Thu, 23 Mar 2017 21:34:40 +0000 (14:34 -0700)]
hci: Timestamp HCI commands for timeouts

Test: Start Bluetooth with COMMAND_PENDING_TIMOUT_MS=2
Change-Id: I393592ede7b9b0aab9c13f6c79756c24edd3d930

7 years agoSet preferred PHY and read PHY implementation (3/3)
Jakub Pawlowski [Thu, 23 Mar 2017 02:00:47 +0000 (19:00 -0700)]
Set preferred PHY and read PHY implementation (3/3)

Test: manual
Bug: 30622771
Change-Id: I4267238a0b5c7bc373ae1846ebd19a716881a4ec

7 years agohci: Cancel timer before running callbacks
Myles Watson [Thu, 23 Mar 2017 18:49:13 +0000 (11:49 -0700)]
hci: Cancel timer before running callbacks

Bug: 36507488
Test: Turn Bluetooth on/off, scan for devices
Change-Id: Ia9334d83b2191504acdab50f777b6d10073226e0

7 years agoMerge "Increase size of written bytes for snprintf for AT+BIND"
Treehugger Robot [Thu, 23 Mar 2017 20:47:55 +0000 (20:47 +0000)]
Merge "Increase size of written bytes for snprintf for AT+BIND"

7 years agoFix a race condition inside command_timed_out() when printing log info
Pavlin Radoslavov [Thu, 23 Mar 2017 19:24:57 +0000 (12:24 -0700)]
Fix a race condition inside command_timed_out() when printing log info

Release the lock after we don't need anymore access to the wait_entry
that was protected by the lock.
Note: We are keeping the lock right before restarting the Bluetooth
process, hence it doesn't matter we are holding it for a bit longer
than before.

Test: Code compilation
Change-Id: Id393a285fce9b9e28efcf0f55fcc32821c5a72b6

7 years agoMerge "AVRCP version interop pereference (1/3)"
Treehugger Robot [Thu, 23 Mar 2017 17:49:56 +0000 (17:49 +0000)]
Merge "AVRCP version interop pereference (1/3)"

7 years agoMerge "main: Update error handling"
Treehugger Robot [Thu, 23 Mar 2017 13:21:52 +0000 (13:21 +0000)]
Merge "main: Update error handling"

7 years agoIncrease size of written bytes for snprintf for AT+BIND
Ajay Panicker [Wed, 15 Feb 2017 20:23:35 +0000 (12:23 -0800)]
Increase size of written bytes for snprintf for AT+BIND

snprintf includes the null terminator for size of string to write to buffer.
Having a buffer size of one would only write a null terminator.

Bug: 35348420
Test: HFP Test with carkit
Change-Id: Id81abca562259371520b891e84ec76f6cb434b0c

7 years agoAVRCP version interop pereference (1/3)
Ajay Panicker [Sat, 18 Mar 2017 21:58:06 +0000 (14:58 -0700)]
AVRCP version interop pereference (1/3)

Add a mechanism to allow the user to select the AVRCP version broadcasted
through SDP in the developer options. This helps solve some interoperability
issues such as with Honda Carkits where broadcasting AVRCP version 1.6 causes
media metadata to fail.

Bug: 35956792
Test: Select AVRCP version through developer option
Merged-In: Ib55af4406131178846d38ef4da33fb72cb2cd02f
Change-Id: Ib55af4406131178846d38ef4da33fb72cb2cd02f

7 years agoReduce sniff intervals for active mode.
Ivan Podogov [Tue, 3 Jan 2017 21:20:33 +0000 (21:20 +0000)]
Reduce sniff intervals for active mode.

According to the Bluetooth HID specification, a pointing device
should have a recommended report rate of approximately 80 Hz, and
sniff interval of 18 slots for the sniff subrating mode.
Current settings set the sniff interval to 54 slots, which is
aproximately 29.6 Hz - not really enough to emulate a mouse.

Test: with a pointing device app
Change-Id: Ic3365137bcf3f5608ada37fbfa0c3233f6ef83d7

7 years agomain: Update error handling
Myles Watson [Tue, 21 Mar 2017 17:24:20 +0000 (10:24 -0700)]
main: Update error handling

Test: builds
Change-Id: Ibc6b1d94f49bd711e7844ae769c384f76d357f22

7 years agoMerge "Add liblog dependency to libbt-vendor."
Treehugger Robot [Mon, 20 Mar 2017 22:35:18 +0000 (22:35 +0000)]
Merge "Add liblog dependency to libbt-vendor."

7 years agoBluetooth 5 AdvertisingSet implementation (4/4)
Jakub Pawlowski [Sat, 18 Mar 2017 01:05:38 +0000 (18:05 -0700)]
Bluetooth 5 AdvertisingSet implementation (4/4)

Test: manual
Bug: 30622771
Change-Id: Ie1c1588383a83296cabd29a0481f7d80a1168263

7 years agoReport back TX Power value
Jakub Pawlowski [Fri, 17 Mar 2017 19:31:42 +0000 (12:31 -0700)]
Report back TX Power value

For LE Extended Advertising report the value returned by controller.
For VSC Advertising return the value send to controller.
For Legacy Advertising use 0.

Test: manual
Bug: 30622771
Change-Id: Ifc49af86e7f064c91ed2d20e6f87532159d1bffa

7 years agoDo not use Multi-Adv instance 0 in VSC
Jakub Pawlowski [Mon, 20 Mar 2017 17:33:25 +0000 (10:33 -0700)]
Do not use Multi-Adv instance 0 in VSC

Bug: 36365653
Test: manual
Change-Id: If44b2f0206bab91fa77f2a1ff68d83e1593f28e2

7 years agoMerge "Change default AVRCP version to 1.4"
Treehugger Robot [Mon, 20 Mar 2017 21:19:05 +0000 (21:19 +0000)]
Merge "Change default AVRCP version to 1.4"

7 years agoAdd liblog dependency to libbt-vendor.
Alex Deymo [Mon, 20 Mar 2017 20:55:10 +0000 (13:55 -0700)]
Add liblog dependency to libbt-vendor.

lubcutils doesn't bundle liblog anymore. This fixes the build in boards
defining BOARD_HAVE_BLUETOOTH_LINUX.

Bug: None
Test: `make droid` in such board.
Change-Id: Ic9b7ba045c3a291eb33a606ffabc41c45c9ca3e2

7 years agoChange default AVRCP version to 1.4
Ajay Panicker [Fri, 17 Mar 2017 22:10:34 +0000 (15:10 -0700)]
Change default AVRCP version to 1.4

Poor support for AVRCP 1.6 in many carkits means the default should be 1.4 until we can work out interop for different models.

Bug: 35956792
Test: Connect Phone to Honda Carkit and play music
Change-Id: Ie48ba3f045cca5edda49399d137f2f756be0ff8a

7 years agoHIDD: Fix possible out-of-bounds error.
Ivan Podogov [Mon, 20 Mar 2017 13:40:39 +0000 (13:40 +0000)]
HIDD: Fix possible out-of-bounds error.

Bug: 32282212
Test: manual
Change-Id: I6ca91bb8e278db9e4121f22567a27c175661b537

7 years agoBase classes should have virtual destructors
Pavlin Radoslavov [Sun, 19 Mar 2017 02:24:05 +0000 (19:24 -0700)]
Base classes should have virtual destructors

Each base class in service/common/bluetooth now has a virtual
destructor. This fixes memory delete issues that were exposed
when running the unit tests with ASAN enabled.

Test: Running unit tests with ASAN enabled
Change-Id: Ie944eb80784de4ebac3c7276fabed6acb5fec6c0

7 years agoMerge "Abort using SIGABRT instead of SIGKILL on command_timeout"
Treehugger Robot [Fri, 17 Mar 2017 19:19:04 +0000 (19:19 +0000)]
Merge "Abort using SIGABRT instead of SIGKILL on command_timeout"

7 years agoDivide advertise data when needed
Jakub Pawlowski [Fri, 17 Mar 2017 05:10:32 +0000 (22:10 -0700)]
Divide advertise data when needed

Bug: 30622771
Test: BleAdvertisingManagerTest.test_data_sender
Change-Id: I447fed753b08cef766d99ea8dfa47b1212a9ce03

7 years agoBluetooth 5 Periodic Advertising
Jakub Pawlowski [Fri, 17 Mar 2017 00:15:47 +0000 (17:15 -0700)]
Bluetooth 5 Periodic Advertising

This patch implements the initialization of Bluetooth 5 advertising.

Bug: 30622771
Test: manual
Change-Id: If06a6c827f1d8097aa3f019b99ebcb3933742d6c

7 years agoAbort using SIGABRT instead of SIGKILL on command_timeout
Andre Eisenbach [Fri, 17 Mar 2017 04:58:53 +0000 (21:58 -0700)]
Abort using SIGABRT instead of SIGKILL on command_timeout

This will produce a stack trace and flag for automated crash reporting.

Bug: 36026072
Test: net_test_bluetooth; device off/on/off tests
Change-Id: I4d4380d61028261ba7bb4599e39993cae1d906a5

7 years agoLE Maximum Advertising Data Length (4/4)
Jakub Pawlowski [Fri, 17 Mar 2017 01:19:20 +0000 (18:19 -0700)]
LE Maximum Advertising Data Length (4/4)

Add ability to check maximum advertising data length.

Bug: 30622771
Test: manual
Change-Id: I0f3c806046157633f8e2106c1b2700c4277a2b4d

7 years agoMerge "Use the correct tBTA_PAN type when copying the data in a callback"
Treehugger Robot [Fri, 17 Mar 2017 04:00:31 +0000 (04:00 +0000)]
Merge "Use the correct tBTA_PAN type when copying the data in a callback"

7 years agoUse the correct buffer size when copying device inquiry response data
Pavlin Radoslavov [Fri, 17 Mar 2017 02:09:31 +0000 (19:09 -0700)]
Use the correct buffer size when copying device inquiry response data

Bug: 36372989
Test: Running ASAN build
Change-Id: I6016e7609ba24db632222ff4613017e3be30b09c

7 years agoUse the correct tBTA_PAN type when copying the data in a callback
Pavlin Radoslavov [Fri, 17 Mar 2017 01:54:55 +0000 (18:54 -0700)]
Use the correct tBTA_PAN type when copying the data in a callback

This fixes stack-buffer-overflow issue found using ASAN.
Previously, the original data had type "struct tBTA_PAN_SET_ROLE"
and similar, and eventually "memcpy(..., sizeof(tBTA_PAN))" would
copy data beyond the end of the data buffer.

Bug: 36367964
Test: Running ASAN build
Change-Id: I47210a501378023168a0dd71381e93a5051a4c71

7 years agoBluetooth 5 AdvertisingSet implementation (3/4)
Jakub Pawlowski [Sun, 12 Mar 2017 00:46:00 +0000 (16:46 -0800)]
Bluetooth 5 AdvertisingSet implementation (3/4)

This patch wires up fist methods of AdvertisingSet, making it possible
to start advertising and stop advertising. It also replaces legacy
implemementation with calls to new implementation.

Bug: 30622771
Test: sl4a ConcurrentBleAdvertisingTest
Change-Id: I843d8fc12ba1928c87e256a0fb6d730c9fa200e6

7 years agoFix net_test_stack_multi_adv
Jakub Pawlowski [Thu, 16 Mar 2017 21:47:55 +0000 (14:47 -0700)]
Fix net_test_stack_multi_adv

Change-Id: Ic28aa4a120d973bbf3f05d8022bc2c2e619562e2

7 years agoFix an HCI race condition when transmitting a packet
Pavlin Radoslavov [Thu, 16 Mar 2017 11:54:21 +0000 (04:54 -0700)]
Fix an HCI race condition when transmitting a packet

There is a race condition when calling
event_command_ready() -> transmit_fragment() -> hci_transmit()

If right after hci_transmit() there is thread context switch
and another thread executes filter_incoming_event() for the same
command, the corresponding packet/command will be taken off the
commands_pending_response list and free()-ed.

However, after the execution on the first thread continues
within transmit_fragment(), the execution logic will continue using
the "packet" that was already free()-ed by the other thread.

To prevent this from happening, the "commands_pending_response_mutex"
within event_command_ready() has to protect the transmit_fragment()
execution and the update_command_response_timer() function right after it.

Also:
 * Changed the "commands_pending_response_mutex" to recursive_mutex
 * Added "commands_pending_response_mutex" protection in few other
   places where "commands_pending_response" itself is used.

Bug: 36205494
Test: Running ASAN build
Change-Id: I63677ad1f2b28683c321631e9e29e4f01628d269

7 years agotest_vendor: Stop watching before closing the FD
Myles Watson [Wed, 15 Mar 2017 19:14:36 +0000 (12:14 -0700)]
test_vendor: Stop watching before closing the FD

Test: net_test_bluetooth BluetoothTest.AdapterRepeatedEnableDisable
Change-Id: I72f5c9105b535ad5a5c41bc15dc9a3ad5f4e023f

7 years agoFix bta_dm_act.cc compile error.
Alex Deymo [Wed, 15 Mar 2017 07:16:58 +0000 (00:16 -0700)]
Fix bta_dm_act.cc compile error.

When BTM_LOCAL_IO_CAPS == BTM_IO_CAP_NONE the code fails to compile due
to a missing comma. This was introducted a couple of months ago by
2e8e9f42132d0297a94c0852a1fbf45c9bc348e5 but it went undetected.

This patch simply adds back the missing comma.

Bug: 35804180
Test: `make` on a device with BTM_LOCAL_IO_CAPS set to BTM_IO_CAP_NONE.
Change-Id: I0a924abec4fe6612c5dbb72416123ca391551523

7 years agoMerge "Add a log message in the beginning of function hci_initialize()"
Treehugger Robot [Wed, 15 Mar 2017 01:08:24 +0000 (01:08 +0000)]
Merge "Add a log message in the beginning of function hci_initialize()"

7 years agoParse extended advertising packets
Jakub Pawlowski [Mon, 13 Mar 2017 23:09:35 +0000 (16:09 -0700)]
Parse extended advertising packets

Test: manual
Bug: 30622771
Change-Id: I286b8d636115385ee5951284629a1894026d9b5b

7 years agoAdd a log message in the beginning of function hci_initialize()
Pavlin Radoslavov [Sat, 11 Mar 2017 04:51:45 +0000 (20:51 -0800)]
Add a log message in the beginning of function hci_initialize()

Bug: 36043456
Test: Code compilation
Change-Id: Ib588c95a64100f5b2b3567f1bac9fe5e9160e4c0

7 years agoAdd "LE Remove Advertising Set Command"
Jakub Pawlowski [Sun, 12 Mar 2017 00:40:52 +0000 (16:40 -0800)]
Add "LE Remove Advertising Set Command"

Bug: 30622771
Test: manual
Change-Id: Ie0df73475359c69af8df672454fe0c0374523c88

7 years agohci: Stop the thread before hci_close()
Myles Watson [Tue, 14 Mar 2017 18:58:10 +0000 (11:58 -0700)]
hci: Stop the thread before hci_close()

Bug:36026072
Test: Switch users (toggle Bluetooth under load)
Change-Id: I3fb901b5c47021c708c1fb548858406225065d00

7 years agoLE Periodic Advertising - advertiser interface
Lakshmipathi K [Mon, 13 Mar 2017 17:35:12 +0000 (10:35 -0700)]
LE Periodic Advertising - advertiser interface

Bug: 30622771
Test: manual
Change-Id: I358b872c9ffb52f007db4f111520c05b9521aa7f

7 years agoMerge "Bluetooth 5 feature check implementation (1/3)"
Treehugger Robot [Sat, 11 Mar 2017 02:03:42 +0000 (02:03 +0000)]
Merge "Bluetooth 5 feature check implementation (1/3)"

7 years agohci: Close the HAL before cleaning up pending command queue and timers
Myles Watson [Fri, 10 Mar 2017 21:08:59 +0000 (13:08 -0800)]
hci: Close the HAL before cleaning up pending command queue and timers

This prevents errors in case of event callbacks being received or timers
being fired after the pending command queue has been deleted.

Bug: 36026072
Test: Turn Bluetooth off/on while scanning and file transfer
Change-Id: I7bb699e929a8b40e42a3543b5445885ae6b1b975

7 years agoBluetooth 5 feature check implementation (1/3)
Jakub Pawlowski [Fri, 10 Mar 2017 02:11:43 +0000 (18:11 -0800)]
Bluetooth 5 feature check implementation (1/3)

Wire the new feature check API to actual values received from the
controller.

Test: manual
Bug: 30622771
Change-Id: I2eefb7d5bb00d89ac7ea803191ba05f830080ade

7 years agosmp_keys simplification
Jakub Pawlowski [Fri, 24 Feb 2017 18:57:57 +0000 (10:57 -0800)]
smp_keys simplification

Test: manual
Change-Id: Ia38604d496b8ae50934c42b68f2424afd1c6320c

7 years agohci: Reduce HCI timeout to 2 seconds
Myles Watson [Mon, 6 Mar 2017 17:07:37 +0000 (09:07 -0800)]
hci: Reduce HCI timeout to 2 seconds

Test: Bluetooth starts/stops, Bluetooth scan works
Change-Id: I4b04d71d57f45956898ff1a99e495195b54f4e8e

7 years agoFix for dual-mode devices advertising with a single address
Jack He [Thu, 16 Feb 2017 08:06:15 +0000 (03:06 -0500)]
Fix for dual-mode devices advertising with a single address

* BTM_ReadRemoteDeviceName checks if a device is connectable over LE before
  making a connection. However, when it is not connectable over LE, it could
  be connectable over BR/EDR.
* For Dual Mode device, the device type should be 0x3 and is not equal to
  BT_DEVICE_TYPE_BREDR, hence we cannot check equality to determine whether
  a device is BR/EDR, but need to AND them and check the result.
* btm_ble_read_remote_name() already checks if device is connectable in LE,
  and we do not need to do it in BTM_ReadRemoteDeviceName().
* This fix removes the check from BTM_ReadRemoteDeviceName
* This fix also adds error checking to return values of BTM_ReadRemoteDeviceName
* This fix moves inquiry database query to respective remote name
  request procedures, as the result of the query is used in those
  procedures instead of in BTM_ReadRemoteDeviceName()

Bug: 35348392
Test: make, pair with device that advertise on both LE and BR/EDR
Change-Id: I10d1aa9dd6e2447f9b9d40b1fea9370b5088a57d

7 years agoMerge "cleanup: no assigns in conditionals"
Treehugger Robot [Thu, 2 Mar 2017 17:05:10 +0000 (17:05 +0000)]
Merge "cleanup: no assigns in conditionals"

7 years agoDisable broken unittest
Bailey Forrest [Wed, 1 Mar 2017 19:49:14 +0000 (11:49 -0800)]
Disable broken unittest

Change-Id: I9c7dbd8dfdb440a31e4c98e93ee533fe96ee7b0e

7 years agoAllow framework to compile with BTM_IO_CAP_NONE
Pulkit Bhuwalka [Wed, 1 Mar 2017 00:11:34 +0000 (16:11 -0800)]
Allow framework to compile with BTM_IO_CAP_NONE

Using BTM_IO_CAP_NONE currently causes compilation to fail since
bta_dm_sp_cback ends up unused out a previous #ifdef that only kicks
in if BTM_IO_CAP_NONE is not set. Plugging the entire function into
a #ifdef so consistency is maintained and the framework can compile.

Bug: 35257446
Test: Built locally and run on device. Pairing without display using
JustWorks works.

Change-Id: Ibd7af6454253ccd731ff79602f327a52b8fe585b

7 years agoMerge "Always restrict the AVDTP MTU for SBC codec to MAX_2MBPS_AVDTP_MTU(663)"
Treehugger Robot [Wed, 1 Mar 2017 03:25:04 +0000 (03:25 +0000)]
Merge "Always restrict the AVDTP MTU for SBC codec to MAX_2MBPS_AVDTP_MTU(663)"

7 years agoMerge "Ignore AAC bitrate from the A2DP Sink device if below a threshold"
Treehugger Robot [Wed, 1 Mar 2017 01:46:28 +0000 (01:46 +0000)]
Merge "Ignore AAC bitrate from the A2DP Sink device if below a threshold"

7 years agoMerge "btif_sock_thread: Set thread_id = -1 after join"
Treehugger Robot [Wed, 1 Mar 2017 00:28:52 +0000 (00:28 +0000)]
Merge "btif_sock_thread: Set thread_id = -1 after join"

7 years agoAlways restrict the AVDTP MTU for SBC codec to MAX_2MBPS_AVDTP_MTU(663)
Pavlin Radoslavov [Tue, 28 Feb 2017 23:30:21 +0000 (15:30 -0800)]
Always restrict the AVDTP MTU for SBC codec to MAX_2MBPS_AVDTP_MTU(663)

Some devices claim they can handle large MTU and large SBC bitpool.
However, they are struggling and eventually over time they
start delaying the "Rcvd Number of Completed Packets" reports.
Now we always limit the payload MTU to MAX_2MBPS_AVDTP_MTU(663) -
with the overhead of (4 bytes L2CAP Header + 12 bytes AVDTP Header)
such packet can fit within 2-DH5 payload size of 679 bytes.

Bug: 34127145
Test: A2DP SBC streaming to headsets and carkit
Change-Id: I2f32becba703d77d36be46158f68b8a9ef3f299a

7 years agoIgnore AAC bitrate from the A2DP Sink device if below a threshold
Pavlin Radoslavov [Fri, 17 Feb 2017 17:53:52 +0000 (09:53 -0800)]
Ignore AAC bitrate from the A2DP Sink device if below a threshold

Some Sink devices might advertise AAC zero bitrate or a very small
value (e.g., 24576). Ignore values that are below a threshold
(currently set to 64000), because those seem invalid.
For such devices, the actual streaming bitrate will be computed
based on the MTU.

Test: Code compilation
Change-Id: Ibe3de6e53681a618f388b41a01f366ad462b7a1d

7 years agoMerge "osi: Use atomic_exchange to protect is_joined"
Treehugger Robot [Tue, 28 Feb 2017 23:07:36 +0000 (23:07 +0000)]
Merge "osi: Use atomic_exchange to protect is_joined"

7 years agobtif_sock_thread: Set thread_id = -1 after join
Myles Watson [Tue, 28 Feb 2017 19:14:26 +0000 (11:14 -0800)]
btif_sock_thread: Set thread_id = -1 after join

Bug: 35690239
Test: Switch users with Bluetooth enabled
     (Disables/Enables Bluetooth under high load)
Change-Id: I6e09547cc8ffa3f01bc523ce4c18fdabfe08df99

7 years agocleanup: no assigns in conditionals
Marie Janssen [Wed, 22 Feb 2017 16:35:29 +0000 (08:35 -0800)]
cleanup: no assigns in conditionals

Some assigns snuck into our conditionals, making the code harder to
read / follow.

Also remove some unnecessary null checks after osi_malloc that are
nearby.

Test: compiles, sanity check with device
Change-Id: Ia544da44060d9fc4363af9fbcb4112f3190f75ab

7 years agobtif: Remove redundant thread_{stop|join}
Myles Watson [Tue, 28 Feb 2017 19:16:03 +0000 (11:16 -0800)]
btif: Remove redundant thread_{stop|join}

Test: Switch users with Bluetooth enabled
     (Disables/Enables Bluetooth under high load)
Change-Id: I2414fd26a0380644700d66a8fdec1fc54a1cd2f6

7 years agobtif_sock: Use atomic operations for thread_handle
Myles Watson [Tue, 28 Feb 2017 19:05:52 +0000 (11:05 -0800)]
btif_sock: Use atomic operations for thread_handle

Test: Switch users with Bluetooth enabled
     (Disables/Enables Bluetooth under high load)
Change-Id: I9039f97c090f693e798b8082fbb027e2028c8b40

7 years agoosi: Use atomic_exchange to protect is_joined
Myles Watson [Tue, 28 Feb 2017 19:01:03 +0000 (11:01 -0800)]
osi: Use atomic_exchange to protect is_joined

Test: Switch users with Bluetooth enabled
     (Disables/Enables Bluetooth under high load)
Change-Id: I34c5579b75620970f756fff28ccc07cf76599a94

7 years agoIncrease max remote ext feature pages to 3
Venkata Jagadeesh Garaga [Wed, 9 Nov 2016 05:10:20 +0000 (10:40 +0530)]
Increase max remote ext feature pages to 3

Use Case: DUT unable to pair with Xbox HID device

Steps:
1. Turn on BT
2. Start Inquiry
3. Pair with device with three extended feature pages

Failure: Unable to pair with HID device

Root Cause: DUT not responding for btm_read_remote_ext_features_complete
as we are not supporting three extended pages. So Security procedure also
not getting started.

Fix: Increase max remote ext feature pages to three from
2 as some latest remotes supporting three extended pages.
As per spec we can support 255 extended pages.

Test: Pair with device with three extended feature pages

Bug: 32999575

Change-Id: I061c78a7aaa1ab58a2fc554d026305a0124eee77

7 years agoFix libldac build on Linux
Jakub Pawlowski [Tue, 28 Feb 2017 01:51:44 +0000 (17:51 -0800)]
Fix libldac build on Linux

In build/BUILD.gn we set EXPORT_SYMBOL value, and it's used across the
whole stack. However, third party libraries require this value to be
cleared - otherwise the symbols are not exported properly.

Test: manual

7 years agoFix initial advertising command order
Jakub Pawlowski [Fri, 24 Feb 2017 19:35:07 +0000 (11:35 -0800)]
Fix initial advertising command order

Set the address during SetParameters, for VSC implementation, and right
after setting parameters for BT5 implementation.

Test: sl4a ConcurrentBleAdvertisingTest
Bug: 33922058
Bug: 34962013
Change-Id: Idcc60a36884607de191066d8be1ee3bc459e86fc

7 years agobtsnd_hcic_ble_rand refactor
Jakub Pawlowski [Fri, 24 Feb 2017 17:49:59 +0000 (09:49 -0800)]
btsnd_hcic_ble_rand refactor

Make btsnd_hcic_ble_rand accept base::Callback instead of function
pointer. This will result in great code simplifications in further
patches.

Test: sl4a ConcurrentBleAdvertisingTest
Change-Id: I9933caa78b20d0011ce3e17fd72d5740bcc4ae6d

7 years agotest_vendor: Replace __FUNCTION__ with __func__
Myles Watson [Tue, 13 Dec 2016 17:47:55 +0000 (09:47 -0800)]
test_vendor: Replace __FUNCTION__ with __func__

Test: Bluetooth settings still works
Change-Id: I4de24cceb9beb262ce7f96e7291288b70618f694

7 years agoRevert "HID: Fix scroll issue with Apple Magic Mouse"
Ajay Panicker [Fri, 24 Feb 2017 17:37:30 +0000 (17:37 +0000)]
Revert "HID: Fix scroll issue with Apple Magic Mouse"

Reverting due to the fact that this causes many HID issues with fugu.
The general problem seems to be that this patch requires kernel 3.18
while fugu is currently on 3.10. Will investigate further later but
reverting for now to fix fugu.

This reverts commit 3e2201163b609aab6dcc0778158e790c3201353b.
Fixes: 35667962
Fixes: 35712367
Change-Id: I536da368fc9796b26dc52892eda6501772683330

7 years agoMerge "Set the AAC audioMuxVersion to "2""
Treehugger Robot [Fri, 24 Feb 2017 02:48:23 +0000 (02:48 +0000)]
Merge "Set the AAC audioMuxVersion to "2""

7 years agoMerge "Update the AAC bitrate settings"
Treehugger Robot [Fri, 24 Feb 2017 01:01:15 +0000 (01:01 +0000)]
Merge "Update the AAC bitrate settings"

7 years agoDisplay passkey when bonding
Jakub Pawlowski [Thu, 23 Feb 2017 22:43:08 +0000 (14:43 -0800)]
Display passkey when bonding

Test: manual
Change-Id: I238dfadf401b90d52caa4c381ea068b48091c092

7 years agoMerge "Clear state mask when scan stopped"
Treehugger Robot [Fri, 24 Feb 2017 00:14:26 +0000 (00:14 +0000)]
Merge "Clear state mask when scan stopped"

7 years agoSet the AAC audioMuxVersion to "2"
Pavlin Radoslavov [Thu, 23 Feb 2017 21:57:15 +0000 (13:57 -0800)]
Set the AAC audioMuxVersion to "2"

This change is needed, because there are Sink devices that
expect a fixed latmBufferFullness value.

Bug: 35715579
Test: A2DP streaming to various devices
Change-Id: Ia27490920cda273cee15bfd7e643b85a5c5e428d

7 years agoUpdate the AAC bitrate settings
Pavlin Radoslavov [Thu, 23 Feb 2017 21:49:54 +0000 (13:49 -0800)]
Update the AAC bitrate settings

Use AACENC_PEAK_BITRATE to limit the maximum bitrate as limited by
the MTU, and use AACENC_BITRATE to set the encoder bitrate.

Previously, both were set to the same value. That was disabling
the AAC bitreservoir, and eventually affecting the audio quality.

Bug: 35715570
Test: A2DP streaming to various devices
Change-Id: I6adac4889f9b2d7509182df8961ddfada02ee8ac

7 years agoDon't build vendor lib on Linux
Bluetooth Build Test [Wed, 22 Feb 2017 19:52:30 +0000 (11:52 -0800)]
Don't build vendor lib on Linux

The stack does no longer use vendor libs. We use hci_layer_linux.cc now.

Test: manual
Change-Id: I6c64337280c073647faf0bd775d75c5d700e7911

7 years agoClear state mask when scan stopped
Jakub Pawlowski [Thu, 23 Feb 2017 21:55:02 +0000 (13:55 -0800)]
Clear state mask when scan stopped

Test: manual test
Change-Id: Icc47679aaba92afd7f3eab48416065d3a741a745

7 years agoLog which advertising HCI interface is used
Jakub Pawlowski [Thu, 23 Feb 2017 17:50:31 +0000 (09:50 -0800)]
Log which advertising HCI interface is used

This is useful for debugging advertising issues.

Test: manual
Change-Id: Ia51eccb1ff7922a5864572a89e77404299965862

7 years agoMerge "Close leftover Audio-to-Bluetooth HAL connections before reopening"
Treehugger Robot [Thu, 23 Feb 2017 18:26:03 +0000 (18:26 +0000)]
Merge "Close leftover Audio-to-Bluetooth HAL connections before reopening"

7 years agoFix stack shutdown on Linux
Jakub Pawlowski [Thu, 23 Feb 2017 01:14:29 +0000 (17:14 -0800)]
Fix stack shutdown on Linux

Right now, hci_close() is getting stuck on stopping the reader thread.
In order to fix that, we must have a way to exit the blocking read when
shutting down the stack.

Test: manual
Change-Id: I3a7eeef6d8ed223fc96ba9eeacca37006a5c9739

7 years agoClose leftover Audio-to-Bluetooth HAL connections before reopening
Pavlin Radoslavov [Thu, 23 Feb 2017 06:07:17 +0000 (22:07 -0800)]
Close leftover Audio-to-Bluetooth HAL connections before reopening

If the Audio-to-Bluetooth HAL reopens a connection on the Data or
Control channel without closing first the old one, the UIPC select(2)
loop gets confused and could spin with 100% CPU usage.
Now we check for channels that are reopened, and explicitly close
them before reopening.

Also, fix a check whether a file descriptor is valid: 0 is a valid
file descriptor.

Test: A2DP testing with a headset: connect/play/stop
Bug: 35394629
Change-Id: I7ae988ac77f25da56e59e347363e7c473e7f69a9