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

Change-Id: I03f022b6f18c7d1ec7f139a7d67fd60ec6f15ba2

Test: compiles
Change-Id: I0a16498d82d12e1b7597290e8b778efb5a7b3906

Also removed last usage of a GKI_* function.

Test: compiles
Change-Id: If9f8beca5d8485b363027fd64ac4005e9e499721

Test: compile using ninja

Test: code compilation
Change-Id: Ia43d7530a598a526764023c47bb20fdedda6fe64

Test: compile with ninja
Change-Id: I0f2ff28a03de05ea96bed5ba43eace9609d8deb0

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

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

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

Test: code compilation.
Change-Id: I726a8df99c4a8e7af8130c0ca63306e67c4486c3

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

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

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)

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

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

Use int32_t and int16_t instead of SINT32 and SINT16

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

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

Test: top-level compilation and running A2DP

Change-Id: Ibbd3176b86667885666bfe29025ec2c120fafce6

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

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.

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

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

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

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

Also, fixed C++ related compilation errors.

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

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

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

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