OSDN Git Service

android-x86/system-bt.git
7 years agoReset reconfig and suspend flags before connecting
Tsuyoshi Kamata [Wed, 4 Mar 2015 03:09:26 +0000 (12:09 +0900)]
Reset reconfig and suspend flags before connecting

When turning a carkit OFF during music playback, the flags are set to
false. After turning it back ON and re-connecting with the carkit, the
flags are not reset. This causes a potential disconnect when the pause
button is pushed on the carkit.

Bug: 28852454
Change-Id: If51c9b0a9d4180d5d3ca74963aa8186dd3f19e96

7 years agoMerge "Update the A2DP codec setup and selection mechanism"
Treehugger Robot [Tue, 25 Oct 2016 19:22:40 +0000 (19:22 +0000)]
Merge "Update the A2DP codec setup and selection mechanism"

7 years agoFix random crashes in HID related code
Jakub Pawlowski [Mon, 24 Oct 2016 22:00:12 +0000 (15:00 -0700)]
Fix random crashes in HID related code

Operation on characteristics/descriptors shouldn't access GATT database
when it's executed. This could happen while service rediscovery is in
progress.

Bug: 32240759
Test: connect to HID device
Change-Id: Ie2b6e6b451456204b1cea1e500df9a0ff949a9ef

7 years agoUpdate the A2DP codec setup and selection mechanism
Pavlin Radoslavov [Mon, 24 Oct 2016 03:39:46 +0000 (20:39 -0700)]
Update the A2DP codec setup and selection mechanism

* Update bta_av_co_audio_set_codec() so it can select among
  multiple available A2DP source codecs.
* Rename A2DP_SetCodec() to A2DP_SetSourceCodec() and update
  it to use tA2DP_CODEC_SEP_INDEX as an argument to specify
  the particular codec (instead of using SBC as default).

Also:
* Move the definition of AVDT_CODEC_SIZE from bt_target.h
  to avdt_api.h and increased its value from 10 to 20
* Add missing bta_av_co_audio_sink_has_scmst() check inside
  bta_av_co_find_peer_sink_supports_codec()
* Rename (inside bta_av_co.cc): cfg -> config

Bug: 30958229
Test: unit tests and A2DP streaming
Change-Id: I49fcf5063c3b6c4060abdfb60c2db171fa1ff747

7 years agoMerge "Dequeues direct connection request during cancel conn operation"
Treehugger Robot [Mon, 24 Oct 2016 18:54:29 +0000 (18:54 +0000)]
Merge "Dequeues direct connection request during cancel conn operation"

7 years agoDequeues direct connection request during cancel conn operation
Subramanian Srinivasan [Thu, 21 Aug 2014 23:46:04 +0000 (16:46 -0700)]
Dequeues direct connection request during cancel conn operation

When cancel connection request for a device is sent
from an app and if the current pending connection
request's BD address does not match with this device,
the entries of the connection request pending queue
are also checked. If BD address match occurs with an
entry in the connection request queue, the entry is
removed from the queue.

Change-Id: I1bf50a424d86ac53a5201fff742c822f4c8d1c0b

7 years agoMerge "Update macros: Add parens and do-while blocks"
Treehugger Robot [Mon, 24 Oct 2016 17:19:33 +0000 (17:19 +0000)]
Merge "Update macros: Add parens and do-while blocks"

7 years agoSerialize stack shutdown, state change callback and cleanup
Gurpreet Ghai [Mon, 12 Sep 2016 12:06:53 +0000 (17:36 +0530)]
Serialize stack shutdown, state change callback and cleanup

Use Case: Bluetooth On/Off scenario, User switch

Steps:
1. Turn ON Bluetooth
2. Switch user
Also observed on occurrence of enable/disable timeout during On/Off testing.

Failure: Crash while sending state change HAL callback.

Root Cause: In the event when stack cleanup is queued to stack manager
thread when stack shutdown is being processed, stack cleanup starts
execution as soon as stack shutdown is finished.
If the function event_signal_stack_down posted to btif thread at the
end of stack shutdown is executed after stack cleanup crash occurs.

Fix: Serialized the execution of stack shutdown, state change callback
execution and stack cleanup to happen in same order.

Test: code compilation.

Change-Id: Ic96205b5c304acb44eab53f4e2cb150726643bda

7 years agoUpdate macros: Add parens and do-while blocks
Myles Watson [Fri, 21 Oct 2016 23:25:30 +0000 (16:25 -0700)]
Update macros: Add parens and do-while blocks

Test: mma -j32
Change-Id: I7a77fb3e045e29c57cbf2e9c36b8867ca263a82f

7 years agobtif: Add do-while blocks to macros
Myles Watson [Fri, 21 Oct 2016 16:39:13 +0000 (09:39 -0700)]
btif: Add do-while blocks to macros

Test: mma -j32
Change-Id: Ie82465b8cef6c4f527fad13892abb365e15dad7c

7 years agoMerge "btif: Fix formatting in comments"
Treehugger Robot [Sat, 22 Oct 2016 03:29:07 +0000 (03:29 +0000)]
Merge "btif: Fix formatting in comments"

7 years agoMerge "Fix the GATT re-discovery failure when GATT service discovery is ongoing"
Treehugger Robot [Sat, 22 Oct 2016 00:54:38 +0000 (00:54 +0000)]
Merge "Fix the GATT re-discovery failure when GATT service discovery is ongoing"

7 years agobtif: Fix formatting in comments
Myles Watson [Fri, 21 Oct 2016 16:45:42 +0000 (09:45 -0700)]
btif: Fix formatting in comments

Test: mma -j32
Change-Id: I851e21d9f2935ab2afbf5e063f6df31e8d50c305

7 years agoFix the GATT re-discovery failure when GATT service discovery is ongoing
IHLHO KIM [Fri, 23 Sep 2016 07:43:54 +0000 (07:43 +0000)]
Fix the GATT re-discovery failure when GATT service discovery is ongoing

If 'BTA_GATTC_Refresh' is called when GATT service discovery is ongoing, GATT service discovery should be restarted after GATT service discovery is completed.
But it cannot be restarted actually because of missing set 'auto_update' state.
This patch set 'auto_update' state to start discovery again correctly.

Change-Id: Id18f04705fef253261e52836849e4e6020e42a05

7 years agoAvoid skipping SDP after link key derivation
Nitin Arora [Wed, 13 Apr 2016 22:53:03 +0000 (15:53 -0700)]
Avoid skipping SDP after link key derivation

Use Case:
In case of cross key derivation of BR link key from a secure LTK,
the corresponding BR transport needs to go through the SDP routine
and alert the upper layers regarding the bond completion of the
BR transport. This specific procedure is currently getting skipped
in an attempt to prevent spurious link key notifications from causing
bond state changes

Fix:
Added an additional check to make sure that the LE pairing is not
ongoing before skipping the SDP. This uses the sequence of operations
where the LE LTK is generated first and link key is derived from the
LTK.

Test: mma -j32
Change-Id: Ife242c93c3adf90d581ac0bbf4896dd164b9f8e2

7 years agoMerge "Reject AVRCP Metadata command with invalid length"
Treehugger Robot [Fri, 21 Oct 2016 20:48:55 +0000 (20:48 +0000)]
Merge "Reject AVRCP Metadata command with invalid length"

7 years agoMerge "Disable HSP profile during BT turn off"
Treehugger Robot [Fri, 21 Oct 2016 20:16:07 +0000 (20:16 +0000)]
Merge "Disable HSP profile during BT turn off"

7 years agoDisable HSP profile during BT turn off
Satish Kodishala [Wed, 9 Mar 2016 13:27:01 +0000 (18:57 +0530)]
Disable HSP profile during BT turn off

Usecase:
1. Enable HSP profile only.
2. Do Network reset.
3. Connect to headset supporting HSP profile.

Failure:
Not able to connect to headset after network reset.

Rootcause:
During BT turn off, HFP profile is disabled instead of
HSP profile. This led to the HSP profile registration
from apps failed during turn on after network reset.

Fix:
Add a check to see if HFP or HSP profile is enabled
during BT turn off.

Test: code compilation
Change-Id: Ie7bd7b546c377f4a060bc89916640035a047d086

7 years agoReject AVRCP Metadata command with invalid length
Anubhav Gupta [Wed, 21 May 2014 14:33:00 +0000 (20:03 +0530)]
Reject AVRCP Metadata command with invalid length

This change ensures that DUT rejects AVRCP Meta command
having invalid data length to avoid crash while processing
the invalid command further.
This crash was seen while running codenomicon test suit.

Change-Id: I8a591d601636a0fce268b5b965f2bd691c24c319

7 years agoPrevent multiple register of service change indication
Nitin Arora [Thu, 16 Jun 2016 22:27:59 +0000 (15:27 -0700)]
Prevent multiple register of service change indication

Use Case:
Pair DUT to a remote LE device. The host registers for the service
change indication twice, once when the connection callback is
received and another time, when the SMP process completes

Failure:
There are a few issues caused by this.
First, the duplicate action of service change indication registeration
is unnecessary.
The registeration that follows the SMP process adds a gatt_if to the
apps holding the current link. This gatt_if (=1) never disconnects
after registeration failure or success. And hence the host can never
physically disconnect the link.

Fix:
This change removes the service change registeration at SMP
process completion

Test: manual
Change-Id: I0d7566e2b4a9d01aa7926cdbe9f528c3941fcfeb

7 years agoMerge "Remove multiplexing capability in AVDTP"
Treehugger Robot [Fri, 21 Oct 2016 16:12:53 +0000 (16:12 +0000)]
Merge "Remove multiplexing capability in AVDTP"

7 years agobtif: Fixup clang-format
Myles Watson [Wed, 19 Oct 2016 16:53:56 +0000 (09:53 -0700)]
btif: Fixup clang-format

In src/btif_sock_l2cap.cc:
Move a comment before the define and run clang-format again.

Fix formatting in comments that were more than 80-characters wide.

Test: mma -j32
Change-Id: I17465a1425d1de9a28f0675fe65c8529359a87af

7 years agobtif: Apply clang-format
Myles Watson [Wed, 19 Oct 2016 16:50:22 +0000 (09:50 -0700)]
btif: Apply clang-format

clang-format doesn't understand block quotes of this form:

/*  This is not handled well
**  because there are two asterisks
**/

cd btif/

 # Replace '**' at the beginning of the line with ' *'

sed 's/^[*][*]/ \*/' -i include/* src/* test/* co/*
clang-format --style=file -i src/* include/* test/* co/*

Test: mma -j32
Change-Id: I2477eae5480602d5b2fee5ec89c9ed7888022341

7 years agoRemove multiplexing capability in AVDTP
Pavlin Radoslavov [Fri, 21 Oct 2016 01:30:05 +0000 (18:30 -0700)]
Remove multiplexing capability in AVDTP

Remove code guarded by AVDT_MULTIPLEXING.
This functionality is not used, and appears broken.

Test: code compilation and A2DP streaming
Change-Id: Ib87a905c61cff6b3b41d1858b90666d846847d0d

7 years agoMerge "Various small warning/build fixes for Linux"
Treehugger Robot [Thu, 20 Oct 2016 22:16:03 +0000 (22:16 +0000)]
Merge "Various small warning/build fixes for Linux"

7 years agoReset device security flags when pairing fails
Andre Eisenbach [Tue, 18 Oct 2016 00:32:45 +0000 (17:32 -0700)]
Reset device security flags when pairing fails

Bug: 29998634
Test: manual
(cherry picked from commit 54ee9431e8bf0ec1c47d7306fd7a3d0975590733)

Change-Id: I8bd0ce5b797e30009260f7fe673f389d066db9ac

7 years agoVarious small warning/build fixes for Linux
Jakub Pawlowski [Thu, 20 Oct 2016 17:19:50 +0000 (10:19 -0700)]
Various small warning/build fixes for Linux

Test: build with ninja
Change-Id: Ic0f4d23e4a56a42be6e660f2afec83400ea6daa6

7 years agoVarious small fixes for Linux build
Jakub Pawlowski [Thu, 20 Oct 2016 06:31:13 +0000 (23:31 -0700)]
Various small fixes for Linux build

* add "-latomic" which was missing since we moved to clang
* empty LOG_EVENT_INT stub added for Linux
* add link libraries to fix stack_unittest
* add missing headers

Test: build with ninja
Change-Id: I1fe0b0b615db1da5ad083c162cb207b08979ce15

7 years agoMove libhardware dependency to third_party folder on Linux
Jakub Pawlowski [Thu, 20 Oct 2016 01:35:18 +0000 (18:35 -0700)]
Move libhardware dependency to third_party folder on Linux

Thanks to this change we no longer have to configure paths by hand

Test: build with ninja
Change-Id: I04dfaa8f16e52a25efcc241019bae40d4201fb8b

7 years agoosi: Apply clang-format
Myles Watson [Wed, 19 Oct 2016 20:15:34 +0000 (13:15 -0700)]
osi: Apply clang-format

cd osi/
clang-format -i --style=file include/*.h include/socket_utils/* src/*.cc \
    src/socket_utils/* src/protos/* test/*

Test: mma -j32
Change-Id: I659e586076f1e2ec8f687cd33f441700b8d1f823

7 years agoGet rid of dependency on core from linux build (1/2)
Jakub Pawlowski [Wed, 19 Oct 2016 03:22:14 +0000 (20:22 -0700)]
Get rid of dependency on core from linux build (1/2)

When building bluetooth stack for linux, don't include dependencies from
core.

Test: build with ninja and mma
Change-Id: I02aeb178327c9f6c0414cc37624aecc58fb0876e

7 years agoMerge "Abstract the adjustment of A2DP codec parameters"
Treehugger Robot [Wed, 19 Oct 2016 07:11:14 +0000 (07:11 +0000)]
Merge "Abstract the adjustment of A2DP codec parameters"

7 years agoRemove Android a2dp bridge from generic build configuration
Jakub Pawlowski [Wed, 19 Oct 2016 01:45:21 +0000 (18:45 -0700)]
Remove Android a2dp bridge from generic build configuration

Code in audio_a2dp_hw folder contains Android-specific code for talking
to Android media libraries. It is not useful on other platforms, and
only adds unnecessary dependencies.

Test: compile with ninja
Change-Id: Ia0313332b4456ba0287587da66cdf2cce045f89c

7 years agoAbstract the adjustment of A2DP codec parameters
Pavlin Radoslavov [Wed, 19 Oct 2016 00:19:55 +0000 (17:19 -0700)]
Abstract the adjustment of A2DP codec parameters

Replaced hard-coded SBC-specific hack for updating some
of its parameters with an API abstraction: A2DP_AdjustCodec()
Also, added the corresponding unit tests.

Bug: 30958229
Test: manual A2DP testing, added new unit tests
Change-Id: I51a0a019d107362f9c24829408d426a5403b0a8e

7 years agoMerge "Use the default LE connection timeout if the remote device has no preference"
Treehugger Robot [Wed, 19 Oct 2016 00:50:29 +0000 (00:50 +0000)]
Merge "Use the default LE connection timeout if the remote device has no preference"

7 years agoMerge "Remove unused task IDs"
Treehugger Robot [Wed, 19 Oct 2016 00:30:01 +0000 (00:30 +0000)]
Merge "Remove unused task IDs"

7 years agoUse the default LE connection timeout if the remote device has no preference
Jakub Pawlowski [Tue, 18 Oct 2016 21:56:00 +0000 (14:56 -0700)]
Use the default LE connection timeout if the remote device has no preference

Change-Id: I03f022b6f18c7d1ec7f139a7d67fd60ec6f15ba2

7 years agoMerge "Add the appropriate statement to "case:" without "break""
Treehugger Robot [Tue, 18 Oct 2016 20:48:22 +0000 (20:48 +0000)]
Merge "Add the appropriate statement to "case:" without "break""

7 years agoMerge "Remove GAP_CONN_POST_EVT_INCLUDED"
Treehugger Robot [Tue, 18 Oct 2016 20:44:08 +0000 (20:44 +0000)]
Merge "Remove GAP_CONN_POST_EVT_INCLUDED"

7 years agoRemove unused task IDs
Andre Eisenbach [Tue, 18 Oct 2016 19:16:07 +0000 (12:16 -0700)]
Remove unused task IDs

Test: compiles
Change-Id: I0a16498d82d12e1b7597290e8b778efb5a7b3906

7 years agoRemove GAP_CONN_POST_EVT_INCLUDED
Andre Eisenbach [Tue, 18 Oct 2016 18:58:57 +0000 (11:58 -0700)]
Remove GAP_CONN_POST_EVT_INCLUDED

Also removed last usage of a GKI_* function.

Test: compiles
Change-Id: If9f8beca5d8485b363027fd64ac4005e9e499721

7 years agoLinux build fix
Jakub Pawlowski [Tue, 18 Oct 2016 16:56:29 +0000 (09:56 -0700)]
Linux build fix

Test: compile using ninja

7 years agoAdd the appropriate statement to "case:" without "break"
Pavlin Radoslavov [Tue, 18 Oct 2016 17:12:26 +0000 (10:12 -0700)]
Add the appropriate statement to "case:" without "break"

Test: code compilation
Change-Id: Ia43d7530a598a526764023c47bb20fdedda6fe64

7 years agoAdd clang configuration for ninja build scripts
Jakub Pawlowski [Tue, 18 Oct 2016 16:56:57 +0000 (09:56 -0700)]
Add clang configuration for ninja build scripts

Test: compile with ninja
Change-Id: I0f2ff28a03de05ea96bed5ba43eace9609d8deb0

7 years agobtif_rc: Don't memset STL mutex objects
Greg Kaiser [Tue, 18 Oct 2016 14:04:24 +0000 (07:04 -0700)]
btif_rc: Don't memset STL mutex objects

Switching to STL mutex/recursive_mutex objects left us with
code which memset these objects.  These are opaque, and need to
be initialized/destroyed via constructors/destructors, not
through directly setting their memory.

We change our memset calls to only zero out the parts of the
structs which don't have these objects.

Test: Recompile
Change-Id: Idf666e935f4094ad4cfca7cdafa523f777a7e769

7 years agoRenamed most C files to C++: *.c to *.cc
Pavlin Radoslavov [Sat, 15 Oct 2016 02:34:48 +0000 (19:34 -0700)]
Renamed most C files to C++: *.c to *.cc

Also:
 - Fixed C++ related compilation errors.
 - Added missing 'extern "C"' guards in some of the header files.
 - Added missing LOCAL_CPP_EXTENSION to Android.mk files.
 - Added-back btif/src/btif_mce.cc and bta/mce/bta_mce_* to
   btif/Android.mk and bta/Android.mk respectively.
 - Fixed the alphabetical ordering of the *.cc files in some
   of the Android.mk files.
 - Added missing Copyright header to "osi/include/list.h"
 - Updated "osi/src/wakelock.cc" to use C++ std::string
   instead of dynamic allocation of C-style strings.

Test: code compilation, unit tests, and A2DP streaming
Change-Id: Ia2f7215ed9df32775c701b68fc86b09875b942c7

7 years agoRemoved obsoleted define statements from bt_target.h
Pavlin Radoslavov [Tue, 18 Oct 2016 00:38:02 +0000 (17:38 -0700)]
Removed obsoleted define statements from bt_target.h

The removed statements were accidentally re-introduced
by an earlier CL that wasn't cherry-picked / rebased properly.

Test: code compilation.
Change-Id: I726a8df99c4a8e7af8130c0ca63306e67c4486c3

7 years agoMerge "Add various missing includes"
Treehugger Robot [Tue, 18 Oct 2016 01:38:08 +0000 (01:38 +0000)]
Merge "Add various missing includes"

7 years agoAdd various missing includes
Mark Salyzyn [Mon, 17 Oct 2016 16:43:39 +0000 (09:43 -0700)]
Add various missing includes

Test: Build errors
Bug: 30465923
Change-Id: Ib902d9d11d1f346d4dab0678d7f409f11a35c1b2

7 years agoMake advertising instance count part of the HCI interface
Jakub Pawlowski [Mon, 10 Oct 2016 21:25:52 +0000 (14:25 -0700)]
Make advertising instance count part of the HCI interface

The number of advertising instances will ultimately depend on the HCI
interface used. BleAdvertisingManagerImpl should not depend on a global
function for that.

Test: Covered by BleAdvertiseApiTest sl4a test
Bug: 30622771
Change-Id: I1399de3f4289708f7218eae9c00ac7372e4246db

7 years agoImplement AVRCP Controller (Client) Browsing.
Sanket Agarwal [Wed, 11 May 2016 19:30:49 +0000 (12:30 -0700)]
Implement AVRCP Controller (Client) Browsing.

Following is included as part of the features:
a) Browsing for all 4 scopes (Media, VFS, Player and Now Playing)
b) Player selection and song selection from browse list.

The change constructs the following (from lower to upper protocol
layers):
1. AVCTP Browse (bta/ & stack/)
                -- Connection Handling
                -- Constructing browse commands
                -- Parsing browse responses
2. AVRCP  (btif/)
                -- JNI interaction/API
                -- Connection handling to Java
                -- Delegating request and responses to Java

Bug: 28791287
Change-Id: Ibc97ded93cb9c469778ea1e37733390d561cd4cd
(cherry picked from commit 9ea8d07c9286a1f4d338dd64ee02266e324d28e5)

7 years agoFormat A2DP-related code with clang-format
Pavlin Radoslavov [Fri, 14 Oct 2016 23:13:54 +0000 (16:13 -0700)]
Format A2DP-related code with clang-format

clang-format -style=file -i stack/a2dp/a2dp_* stack/include/a2dp_* \
        btif/include/btif_a2dp* btif/src/btif_a2dp*

Test: top-level compilation and running A2DP
Change-Id: I66f0a047192b783ac0147def480754daf617dd8d

7 years agoRemoved unused *_DYNAMIC_MEMORY conditional statements
Pavlin Radoslavov [Fri, 14 Oct 2016 22:47:27 +0000 (15:47 -0700)]
Removed unused *_DYNAMIC_MEMORY conditional statements

Test: compilation from the top-level directory
Change-Id: I181e6ae2e71c232ca0dc733dabda5cb0fc2672d8

7 years agoMerge "Use proper types in SBC related code"
Treehugger Robot [Sat, 15 Oct 2016 04:51:06 +0000 (04:51 +0000)]
Merge "Use proper types in SBC related code"

7 years agoUse proper types in SBC related code
Jakub Pawlowski [Sat, 15 Oct 2016 00:14:16 +0000 (17:14 -0700)]
Use proper types in SBC related code

Use int32_t and int16_t instead of SINT32 and SINT16

Test: organoleptic assessment of audio quality from Android and Linux
Change-Id: Ia4d8a5f08163a90240382fd102082f5aab9611c5

7 years agoMerge "Add device record for BR/EDR devices with derived LE keys"
Treehugger Robot [Sat, 15 Oct 2016 00:34:00 +0000 (00:34 +0000)]
Merge "Add device record for BR/EDR devices with derived LE keys"

7 years agoMerge "A2DP-related renaming: a2d_* to a2dp_*"
Treehugger Robot [Sat, 15 Oct 2016 00:09:42 +0000 (00:09 +0000)]
Merge "A2DP-related renaming: a2d_* to a2dp_*"

7 years agoAdd device record for BR/EDR devices with derived LE keys
Subramanian Srinivasan [Fri, 14 Oct 2016 22:43:34 +0000 (15:43 -0700)]
Add device record for BR/EDR devices with derived LE keys

Loads BLE keys stored in NVRAM to btm_sec database for BR/EDR
devices with cross transport derived LE keys after BT reset.
This change also makes sure that after BT reset, while loading
the bonded devices from NVRAM, the bonded devices(with device
type BR/EDR and having derived BLE keys) are also added to the
resolving list after loading the IRK into btm_sec database.

Test: vendor
Change-Id: I1c9164a947526cf2ea36156458d148f5cbee778f

7 years agoA2DP-related renaming: a2d_* to a2dp_*
Pavlin Radoslavov [Fri, 14 Oct 2016 22:14:37 +0000 (15:14 -0700)]
A2DP-related renaming: a2d_* to a2dp_*

Test: top-level compilation and running A2DP

Change-Id: Ibbd3176b86667885666bfe29025ec2c120fafce6

7 years agoFixes setting of service uuid mask during BLE APCF scan
Subramanian Srinivasan [Mon, 24 Aug 2015 19:17:03 +0000 (12:17 -0700)]
Fixes setting of service uuid mask during BLE APCF scan

Fixes setting of incorrect service uuid mask bytes in vendor
specific command when service uuid mask length is less than
the length of service uuid APCF filter.
eg:ServiceUuid:12131215-0000-1000-8000-00805F9B34FB
SvcMaskUuid:0000FFFF-0000-1000-8000-00805F9B34FB
In the above eg, user is trying to apply 32 bit service uuid
mask on 32 bit service uuid. But the stack treats the
service uuid mask as 16 bits and inserts garbage values for
the first 16 bits of service uuid mask in the VS command
since the MSB bits of the mask are zeroes. This change fixes
the scenario where service uuid mask length is less than
service uuid length.

Change-Id: Ie1aece1162b1ba7f4b7a3e9eea7eb5a074f85d62

7 years agoHCI HAL: Retry writing HCI data on UART socket for EAGAIN errors
Lawrance Liu [Wed, 21 Sep 2016 15:01:13 +0000 (23:01 +0800)]
HCI HAL: Retry writing HCI data on UART socket for EAGAIN errors

If HCI HAL receives error code "EAGAIN" when transmitting data on UART
socket, retry transmit data again.

Purpose :
When doing system suspend/resume aging test, the HCI HAL may transmit
data during system resuming. In this case, there exist some timing which
BT controller does not allow HCI HAL to transmit data.

Change-Id: Ic1f13c76d8e735c941800a318a0388ca42b1fd72
Test: System suspend/resume aging test.

7 years agoMerge "Renamed files: *.c to *.cc in bta/av and stack/a2dp"
Treehugger Robot [Fri, 14 Oct 2016 20:21:15 +0000 (20:21 +0000)]
Merge "Renamed files: *.c to *.cc in bta/av and stack/a2dp"

7 years agoUpdate hf structure on phone state change only if SLC is up
Satish Kodishala [Thu, 4 Feb 2016 04:05:59 +0000 (09:35 +0530)]
Update hf structure on phone state change only if SLC is up

Usecase:
1. Enable multi-hf
2. Pair and connect A2DP to a headset(hs1).
3. Pair and connect HFP to another headset(hs2)
4. Accept call on AG
5. Disconnect hs2 by powering off headset
6. Disconnect the call
7. Power on the hs2

Failure:
No A2DP playback on hs1 though music player is playing.

Rootcause:
When there is a call on AG, we update both hf structures about
the call. However, when any headset gets disconnected, only its
hf structure is cleared. After the call is ended, we don't clear
hf structures since there is no hf connection. When A2DP queries
for active call, we return true incorrectly since the hs1's structures
are not updated when the call is ended.

Fix:
Update hf structures on phone state change only if SLC is connected
on it. Updating both hf structures regardless of the SLC state causes
hf state to indicate wrong status.

Test: manual
Change-Id: I487792038d97a2bac77639d7c699f49fc0b18f97

7 years agoUpdate GATT channel state to closing during disconnection
Nitin Arora [Thu, 14 May 2015 01:39:50 +0000 (18:39 -0700)]
Update GATT channel state to closing during disconnection

This change ensures that the GATT channel state is
moved to CLOSING state while GATT disconnect is
requested by the host, specifically when the GATT
state is OPEN.
This is needed to prevent any furthur GATT operations,
once GATT disconnect is issued.

Change-Id: I8511caa477881de4f4eaf18b84f23530e1958fe9

7 years agoSkip LE key derivation if BR key is temporary
Nitin Arora [Wed, 19 Aug 2015 22:47:23 +0000 (15:47 -0700)]
Skip LE key derivation if BR key is temporary

This change skips the LE LTK derivation in case the BR pairig
is not deticated bonding and neither side has requested MITM.
I.e. OPP file transfer require such bonds. This change resolve
the issue of residual IRK in the resolving list.

Change-Id: Ifc9dfc097564f603a466a5bd9b245c89f1bb82aa

7 years agoAdd null checks for L2CAP socket callback
Srinu Jella [Wed, 25 Mar 2015 07:36:38 +0000 (13:06 +0530)]
Add null checks for L2CAP socket callback

Use Case: Bluetooth process crashed while sending the file
to remote device.

Steps: Send a file over L2CAP (OBEX over L2CAP) to remote device

Failure: BT process will crash and restarted automatically

Root Cause: L2CAP socket callback reset to null on error
condition, and when other function try to dereference it, this
leads to BT crash.

Fix: Added null checks for L2CAP socket callback

Change-Id: I2e4f20278fcc8a09bd4dbd507a6c4147e0de93c1

7 years agoRenamed files: *.c to *.cc in bta/av and stack/a2dp
Pavlin Radoslavov [Fri, 14 Oct 2016 19:17:08 +0000 (12:17 -0700)]
Renamed files: *.c to *.cc in bta/av and stack/a2dp

Also, fixed C++ related compilation errors.

Test: code compilation from top-level directory
Change-Id: I60367ee7bc54b891cb05bb3f88dd11ebe5b2ca83

7 years agoMerge "Rectify size passed to copy service name to security records"
Treehugger Robot [Fri, 14 Oct 2016 19:49:05 +0000 (19:49 +0000)]
Merge "Rectify size passed to copy service name to security records"

7 years agoMerge "Remove bonded device information on pairing rejected"
Treehugger Robot [Fri, 14 Oct 2016 19:45:26 +0000 (19:45 +0000)]
Merge "Remove bonded device information on pairing rejected"

7 years agoMerge "Cleanup L2CAP socket properly on close"
Treehugger Robot [Fri, 14 Oct 2016 19:43:05 +0000 (19:43 +0000)]
Merge "Cleanup L2CAP socket properly on close"

7 years agoFix HFP AT command BIA failures
Devin Kim [Wed, 7 Sep 2016 04:47:47 +0000 (13:47 +0900)]
Fix HFP AT command BIA failures

Running the PTE test case TC_AG_IIA_BV_01_I, a failure occurs due to
missing initialization of tBTA_AG_VAL.

Bug: 31325270
Test: PTS
Change-Id: I683eccd53d40e79ec03545166b18ffa1922f0fb2

7 years agoRFCOMM: Prevent sending data in BTA_JV_ST_NONE state
Srinu Jella [Mon, 7 Sep 2015 08:10:07 +0000 (13:40 +0530)]
RFCOMM: Prevent sending data in BTA_JV_ST_NONE state

This patch adds state check in bta_jv_rfcomm_write, to prevent sending
data in BTA_JV_ST_NONE state. This would prevent an ASSERT in the code
due to a null handle. The handle might have already been set to null
during the closing of the RFCOMM connection.

Change-Id: Ibc6b1fb7aa6dce9a4acb52d724af3023fad9a9ae

7 years agoRectify size passed to copy service name to security records
Gurpreet Ghai [Thu, 9 Jun 2016 09:06:35 +0000 (14:36 +0530)]
Rectify size passed to copy service name to security records

Use case: Services having name length greater than max length
always take new slot in security records. After few enable
disable cycles in BLE ON scenario, the record gets completely
occupied making further connection impossible.

Steps:
1. Install a BLE App to create BLE always ON scenario.
2. Run a few cycles of Bluetooth(BREDR) enable-disable.

Failure: Once security records are completely occupied, new
connections are not possible because of socket creation
failure due to unavailability of security id.

Root cause: Size passed to copy service name does not include
bit for null termination. So, for services having name size
greater than that size have their last bit truncated for null
termination.
During next enable, the comparison with same size always fails
causing service to take new slot in security records array.

Fix: Increased the size passed to copy service name by 1 after
considering null termination.

Change-Id: I3527df6610932f60fd85bcdd14d5b8a527c79722

7 years agoRemove bonded device information on pairing rejected
Srinu Jella [Thu, 17 Mar 2016 11:53:48 +0000 (17:23 +0530)]
Remove bonded device information on pairing rejected

Use Case: Remove bonded device information from the BTIF layer storage
when authentication failed.

Failure: Remote device appeared in the paired list after authentication
failed.

Steps:
1. Create a connection from DUT to remote.
2. Remove linkkey in remote.
3. Create connection from DUT to remote, which is in paired list of
DUT, Remote rejected the connection and authentication failed.
But after BT OFF/ON, remote device appeared in the paired list

Root Cause: Not removing device if remoted rejected the pairing
request with reason pairing not allowed

Fix: Remove bonded device information from the BTIF layer storage
on rejected with pairing not allowed reason from remote

Change-Id: Ic9e5e1ed70d304edd3dbca82a73833753fa0093c

7 years agoCleanup L2CAP socket properly on close
Srinu Jella [Thu, 7 Apr 2016 13:35:44 +0000 (19:05 +0530)]
Cleanup L2CAP socket properly on close

Use case:
1. Sometimes pairing pop up occurs during Android beam
   transfer though it is not required.
2. No OPP Tx request over OBEX if we cancel during file transfer.

Steps:
1. Pair and connect from DUT to remote.
2. Go to gallery and send a file.
3. Accept same file and stop it from remote during transfer.
4. Repeat step 2-4 10 times.

Failure:
OPP Tx should not fail.

Root cause:
Close request was not properly handled from BTIF layer.

Fix:
1. Close the request properly from BTIF layer
   to close the socket variables and instruct the BTA layer
   close from there as well.

2. Clean up L2CAP socket properly on closure to free
   the entry registered with security manager to avoid
   leak in security manager records.

3. Close the socket for all valid handles including handle 0.

Change-Id: I4e37dcd858af258fbd64fbfb2fbf0083bd743e06

7 years agoAdd missing header files for ninja
Jakub Pawlowski [Fri, 14 Oct 2016 16:09:26 +0000 (09:09 -0700)]
Add missing header files for ninja

Ninja is complaining about unknown size_t and CHAR_BIT.

Test: compile with ninja
Change-Id: I5cf16dbeba964b8553b4e97d9ee21f68891956e3

7 years agoAdd a mechanism to dump A2DP codec stats
Pavlin Radoslavov [Wed, 12 Oct 2016 16:45:30 +0000 (09:45 -0700)]
Add a mechanism to dump A2DP codec stats

Also, added-back A2DP SBC frames related stats:
  - media_read_total_limited_frames
  - media_read_max_limited_frames
  - media_read_limited_count
  - media_read_total_expected_frames
  - media_read_max_expected_frames
  - media_read_expected_count

A2DP SBC State:
  Frames expected (total/max/ave)                         : 9131 / 8 / 6
  Frames limited (total/max/ave)                          : 0 / 0 / 0
  Counts (expected/limited)                               : 1322 / 0

Bug: 30958229
Test: manual
Change-Id: Id6ddea5eb9d39cf03cae84c79968081d56984e91

7 years agoAdd include for CHAR_BIT definition
Jakub Pawlowski [Thu, 13 Oct 2016 23:49:07 +0000 (16:49 -0700)]
Add include for CHAR_BIT definition

ninja is failing to compile due to missing limits.h include

Change-Id: I4c8eb0df7fb521c44358401ed1ce5cafe29384bb

7 years agoMerge "BTIF A2DP Source and Sink refactor and cleanup"
Treehugger Robot [Fri, 14 Oct 2016 01:55:33 +0000 (01:55 +0000)]
Merge "BTIF A2DP Source and Sink refactor and cleanup"

7 years agoBTIF A2DP Source and Sink refactor and cleanup
Pavlin Radoslavov [Mon, 3 Oct 2016 01:34:46 +0000 (18:34 -0700)]
BTIF A2DP Source and Sink refactor and cleanup

* Moved bta/av/bta_av_sbc.c to stack/a2dp/a2d_sbc_up_sample.c
  and renamed the corresponding bta_av_sbc_up_sample_* functions to
  a2d_sbc_up_sample_*
  Also, renamed bta_av_sbc_init_up_sample() to a2d_sbc_init_up_sample()
* Moved A2DP SBC Encoder-related tasks, state and constants from
  btif_a2dp_source.cc to a2d_sbc_encoder.c
* Introduced A2DP encoder callbacks interface tA2D_ENCODER_INTERFACE,
  and added A2D_GetEncoderInterface() to acccess the interface for a codec.
* Updated a2d_sbc_encoder_init() to specify callbacks:
  - a2d_source_read_callback_t - a callback to read audio data for encoding
  - a2d_source_enqueue_callback_t - a callback to enqueue A2DP source
    packets for transmission
* Removed BTIF_MEDIA_TRSCD_* - the encoding is always PCM
* Removed conditionally defined constants and replaced them by
  local definitions:
  - BTIF_MEDIA_BITRATE_STEP -> A2D_SBC_BITRATE_STEP
  - BTIF_A2DP_DEFAULT_BITRATE -> A2D_SBC_DEFAULT_BITRATE
  - BTIF_A2DP_NON_EDR_MAX_RATE -> A2D_SBC_NON_EDR_MAX_RATE
* Hard-coded the read data timeout period A2DP_DATA_READ_POLL_MS to 10ms
* Merged the UNDERRUN read errors into the UNDERFLOW errors.
* Removed the following statistics, because they are not very useful:
  - media_read_total_limited_frames
  - media_read_max_limited_frames
  - media_read_limited_count
  - media_read_total_expected_frames
  - media_read_max_expected_frames
  - media_read_expected_count
* Updated the implementation of buffer_overruns_max_count to
  use the (new) counter tx_queue_max_dropped_messages instead of
  media_read_max_expected_frames
* Replaced usage of LOG_DEBUG() in stack/a2d/ with LOG_VERBOSE():
  the former is enabled by default on some of the builds.
* Renamed:
  - btif_a2dp_sink_audio_focus_state_t -> btif_a2dp_sink_focus_state_t
  - btif_a2dp_sink_set_audio_focus_state_req() ->
    btif_a2dp_sink_set_focus_state_req()
  - tBTIF_A2DP_SINK_CB.RxSbcQ -> rx_audio_queue
  - btif_a2dp_source_start_aa_req() -> btif_a2dp_source_start_audio_req()
  - btif_a2dp_source_stop_aa_req() -> btif_a2dp_source_stop_audio_req()
  - btif_a2dp_source_aa_readbuf() -> btif_a2dp_source_audio_readbuf()
  - tBTIF_A2DP_SOURCE_CB.TxAaQ -> tx_audio_queue
  - BTIF_A2DP_SOURCE_MEDIA_TIMER_MS -> A2D_SBC_ENCODER_INTERVAL_MS
  - tA2D_AV_MEDIA_FEEDINGS -> tA2D_FEEDING_PARAMS
  - tBTIF_A2DP_SOURCE_INIT_AUDIO -> tA2D_ENCODER_INIT_PARAMS
  - tBTIF_A2DP_SOURCE_UPDATE_AUDIO -> tA2D_ENCODER_UPDATE_PARAMS
  - tBTIF_A2DP_SOURCE_INIT_AUDIO_FEEDING -> tBTIF_A2DP_SOURCE_INIT_FEEDING

Bug: 30958229
Test: TestTracker/65192
Change-Id: Iba0a9694bda3dba73f211f1bde25821497fa6a06

7 years agoUse proper socket type in uipc for non-android
Jakub Pawlowski [Thu, 13 Oct 2016 22:46:30 +0000 (15:46 -0700)]
Use proper socket type in uipc for non-android

Change-Id: I5e901ca381c816945146b508770a2aa54d9cdd79

7 years agoUpdate BUILD.gn files
Jakub Pawlowski [Thu, 13 Oct 2016 22:43:15 +0000 (15:43 -0700)]
Update BUILD.gn files

Test: compiled using ninja

7 years agoSplit btif_media_task into Source, Sink and Control
Pavlin Radoslavov [Fri, 23 Sep 2016 23:36:47 +0000 (16:36 -0700)]
Split btif_media_task into Source, Sink and Control

* btif/include/btif_a2dp.h and btif/src/btif_a2dp.cc implement
  the entry points for the BTIF A2DP module.

* btif/include/btif_a2dp_source.h and btif/src/btif_a2dp_source.cc implement
  the BTIF A2DP Source component.

* btif/include/btif_a2dp_sink.h and btif/src/btif_a2dp_sink.cc implement
  the BTIF A2DP Sink component.

* btif/include/btif_a2dp_control.h and btif/src/btif_a2dp_control.cc
  implement the A2DP control mechanism for the audio channel from the
  Media Framework.

Also:
* Removed BTA_AV_SBC_HDR_SIZE and used A2D_SBC_MPL_HDR_LEN instead.
* Removed BTIF_AV_SINK_FOCUS_REQ_EVT, because it is not used.
* Removed many of the "#if (BTA_AV_SINK_INCLUDED == TRUE)" guards.
* Removed "#ifdef USE_AUDIO_TRACK" guard, and always compile the
  corresponding code.
* Removed tBTIF_AV_MEDIA_FEEDINGS_PCM_STATE and moved its state
  to tBTIF_AV_MEDIA_FEEDINGS_STATE .

Bug: 30958229
Test: TestTracker/65192
Change-Id: I20bc52a1d7a7f03c92628a1562f14b7df3ebb445

7 years agoMerge "cleanup: Remove BT_USE_TRACES and BT_TRACE_APPL"
Treehugger Robot [Thu, 13 Oct 2016 00:22:49 +0000 (00:22 +0000)]
Merge "cleanup: Remove BT_USE_TRACES and BT_TRACE_APPL"

7 years agoFix file permissions for source code files
Pavlin Radoslavov [Wed, 12 Oct 2016 19:54:05 +0000 (12:54 -0700)]
Fix file permissions for source code files

Removed file execute permissions for:
 - bta/include/bta_hh_co.h
 - embdrv/sbc/decoder/Android.mk
 - embdrv/sbc/Android.mk
 - embdrv/Android.mk
 - btif/co/bta_ag_co.cc

Test: visual inspection of file permissions
Change-Id: I746debb1e8ceaa1cc14f867641b25cb42c04c7be

7 years agocleanup: Remove BT_USE_TRACES and BT_TRACE_APPL
Marie Janssen [Wed, 12 Oct 2016 19:40:57 +0000 (12:40 -0700)]
cleanup: Remove BT_USE_TRACES and BT_TRACE_APPL

Test: mma in system/bt

Change-Id: I357eb44a97f3935262252cf97c13cef078a95e4d

7 years agoMerge "btif: migrate from pthread locks to std::mutex"
Marie Janssen [Wed, 12 Oct 2016 16:17:42 +0000 (16:17 +0000)]
Merge "btif: migrate from pthread locks to std::mutex"

7 years agoMoved the typedef of period_ms_t from alarm.h to time.h
Pavlin Radoslavov [Wed, 12 Oct 2016 00:46:45 +0000 (17:46 -0700)]
Moved the typedef of period_ms_t from alarm.h to time.h

Test: code compilation
Change-Id: Ib61126d7921d088b045dca6c5b69a5379603fcc3

7 years agobtif: migrate from pthread locks to std::mutex
Marie Janssen [Mon, 10 Oct 2016 20:38:30 +0000 (13:38 -0700)]
btif: migrate from pthread locks to std::mutex

Test: run unit tests / sanity connection to devices
Change-Id: I293c3600affd229fea67cdd6624eba7f186cbcb7

7 years agoManually merge commit 'cc15f06acc3a1447318fefae00b3c10d9b7ae42a' to fix
Stephen Li [Tue, 11 Oct 2016 17:27:31 +0000 (10:27 -0700)]
Manually merge commit 'cc15f06acc3a1447318fefae00b3c10d9b7ae42a' to fix
automerger

BUG:32069275
Change-Id: I60cc50b21fe0d5a5464bd3acc5897ba7c4e05b73

7 years agostack: remove dead (#if 0'd) code
Marie Janssen [Mon, 10 Oct 2016 16:49:14 +0000 (09:49 -0700)]
stack: remove dead (#if 0'd) code

Test: compiles
Change-Id: I34f68c2b166a28622155358ff75478af8d18bed0

7 years agoFix improper connection check for absolute volume
Ajay Panicker [Tue, 11 Oct 2016 03:43:42 +0000 (20:43 -0700)]
Fix improper connection check for absolute volume

Test: Manually tested
Bug: 32060762
Change-Id: Ia6ca97c119d421aef1136f97afccb3f0c126334d

7 years agoReplace all uses of sprintf() with snprint()
George Burgess IV [Wed, 2 Mar 2016 22:00:19 +0000 (14:00 -0800)]
Replace all uses of sprintf() with snprint()

- sprintf() does not limit the length of the character string when writing
  to a buffer and may result in buffer overflow
- snprintf() requires the maximum write length as a parameter. When the
  maximum length supported is smaller than the reserved buffer length,
  the call will not result in buffer overflow

Bug: 31859081
Test: TestTracker/64195/3975
Change-Id: I519f8ef7b9b162fd79094f89148250d783c734c0

7 years agoMerge "Revert "Add dumpsys support for LE connection parameter updates""
Treehugger Robot [Mon, 10 Oct 2016 23:44:10 +0000 (23:44 +0000)]
Merge "Revert "Add dumpsys support for LE connection parameter updates""

7 years agoRemove Jabra headsets from absolute volume blacklist
Andre Eisenbach [Tue, 13 Sep 2016 16:57:51 +0000 (09:57 -0700)]
Remove Jabra headsets from absolute volume blacklist

Firmware fixes are available for affected headsets (Thanks, Jabra!!).

Change-Id: I60b42366d1e8155895cc58fff69f6404415a0406

7 years agoConvert advertising HAL from struct into class (3/3)
Jakub Pawlowski [Thu, 6 Oct 2016 23:52:30 +0000 (16:52 -0700)]
Convert advertising HAL from struct into class (3/3)

Bug: 30622771
Test: all related tests were updated
Change-Id: I9695284ae249fa427ae4b2f3e9649da951102c6c

7 years agoRemove return value of BTM_VendorSpecificCommand function
Jakub Pawlowski [Mon, 10 Oct 2016 19:36:49 +0000 (12:36 -0700)]
Remove return value of BTM_VendorSpecificCommand function

The return value of BTM_VendorSpecificCommand is currently just an
information wether the callback was passed to it or not, but many places
through stack use this value as error indication. Remove it.

Test: unit tests pass
Change-Id: Idf73fc49fb25e52c3aeb64f2305b4498ecacd9a4

7 years agoMerge "Remove return value of BTM_VendorSpecificCommand function"
Treehugger Robot [Mon, 10 Oct 2016 22:21:36 +0000 (22:21 +0000)]
Merge "Remove return value of BTM_VendorSpecificCommand function"

7 years agoRevert "Add dumpsys support for LE connection parameter updates"
Andre Eisenbach [Thu, 6 Oct 2016 22:42:22 +0000 (22:42 +0000)]
Revert "Add dumpsys support for LE connection parameter updates"

Reverting for now after talking to Jacky because this causes  circular dependencies as lower layer (stack/) functions are calling higher layer APIs (btif/) directly.

To restore a change like this, the data should be collected at a lower layer in the stack and reported through polling or a callback.

This reverts commit 8ec8ca4a875eed8b28e6714899f49781537411ae.

Change-Id: I748843864ae7198ea021fe70d2643a62097ab029

7 years agoMerge "L2CAP: Ensure handle is not null before attempting to close a socket"
Treehugger Robot [Mon, 10 Oct 2016 20:19:29 +0000 (20:19 +0000)]
Merge "L2CAP: Ensure handle is not null before attempting to close a socket"