When advertiser is not started because too many advertisers are
registered, return proper error code.

Bug: 30622771
Test: sl4a ConcurrentBleAdvertisingTest
Change-Id: Ie8a94b964170edbbf1c56f39b03ad98f745a4861

Bug: 32995283
Test: Sanity test with various devices
Change-Id: I20a0b1571c044170ba5a8792d8e990a6790cbce2

BT_TRACE_VERBOSE turned on constants-as-strings in
logs in many parts of the stack, which is preferable
to obscure numbers.

Test: m -j37, check logspam is not worse than normal on a device
Change-Id: I1d24f1c1337290a8a2bcd9a0d937c9e0f1f1c611

AdvertiseManager.startAdvertising() was implemented in Java layer for
historical reasons. Now that we have nice native callbacks and tests,
it can be moved into native code, where it will be shared between
Android daemon and bluetoothtbd.

Bug: 30622771
Test: native tests added to cover startAdvertising
Change-Id: Idf5d3c53b6ad21d8c7e239b5b150b88e073bb709

Fixes logic in A2DP_CodecInfoMatchesCapabilitySbc that could otherwise
reject valid SBC codec parameters, leading to A2DP being disconnected.

Example:
Tesla Model S - min bitpool = 2; max bitpool = 60
Android - min bitpool = 2; max bitpool = 53

Bug: 32688022
Test: Added test to test_a2dp_is_source_codec_supported to cover this bug
Change-Id: If02bab1c47f926f82a636184cf3cbd1edbbe223a

Change-Id: I82d2bb25470d46516e0ab6114bf85cb93f9023ac

This reverts commit 5db07990.

Despite the original patch intention, it actually causes various remote
devices (tested Android Auto, Logitech Speakerphone, Tesla Model S and
many more) to no connect A2DP (only HFP) in various re-connect
scenarios.

This is due to the fact that the stream end point type is not correctly
determined yet in some of these scenarios causing the AVDT stream to be
closed prematurely and thus the A2DP connection to fail.

Test: manual; various devices
Fixes: 32688022

Use AVRC_ADV_CTRL_INCLUDED just to control features instead of
controlling features and inclusion of functions.

Change-Id: Ifff0d7006c3d8df1d57c79ace19ee60487c2f0a5

Bug: 24400656
Test: Previously crashed Bluetooth, does not crash after patch
Change-Id: Ia381abca72e46d369a0de1749e42d1b6273a321a

AVRCP controller side support is always included now.

Test: croot system/bt && mma -j37

Change-Id: I22260b4f123b352f70cbab54a19b5f8f5f171825

uinput isn't used anymore; all media commands go through the HAL

Test: croot system/bt && mma -j37

Change-Id: I83ab133794683a26e250ea10927fa95b2e2cb28a

This fixes a potential NULL-pointer check order reverse
in case failed to get TCB using the TCB index

Test: mm -j 8
Change-Id: I8fcc168d7d077f8df7c7abb9c325c5168fa8cd39

This fixes a potential NULL-pointer access in case
the ACL DB for the requested BDA is not found.

Test: mm -j 8

Change-Id: I86ab66730aa0d9a143ddde851bd33a05a405bb7f

We are passing the value in pointer (UINT_TO_PTR) hence no copy is
required for the context transfer (simply use the UINT value in the
event field instead)

Change-Id: I43625c3a4d68cd8cbadd51b1fccc4138583082b6

Bug: 30622771
Test: still builds
Change-Id: I4003a1241e611c9c1f2a21090ea3ba7530b0643a

Ensure BleAdvertisingManager::OnAdvertisingStateChanged() is called
after an advertising state change to re-enable advertising after
a connection.

This patch also fixes the registration test, which was broken since
one more advertising instance was freed for general usage.

Test: net_test_stack_multi_adv native test
Change-Id: I76887d98cb3ec386ebc0f7fad95170b8c18b9116

Remove the useless "error:" label and NULL-pointer check block

Test: mm -j 8

Change-Id: I218f5f1e2741a49a0fe9bec749dea2484a1df2eb

GetElementAttributes response and GetItemAttributes response share the
same format and require the same checks for length.

Test: play media on carkit, see media. especially with long items.
Bug: 32407250
Bug: 30571638
Change-Id: I8623e7d662f7a39112b7527b6f5ab63c5e32379c

This patch unifies advertising API. On startup, if VSC multi advertising
is avaliable, it will be used, otherwise
BleAdvertiserLegacyHciInterfaceImpl will be used, which provides one
advertising instance.

Test: sl4a tests are passing, all proper tests updated accordingly
Change-Id: I73753962e3a33a980e8a9a1174c8d88611a47e6b

When adding services to GATT database, applications should always provide
it in big endian format.

Bug: 32750783
Change-Id: I5264357bdbb6bab453b63baae6563cde86d20bac

This patch fixes bug introduced in commit
e5d56f55
which would cause unnecessary connection parameter update requests
to be send. It should be sent immediately after connecting only.

Bug: 32563079
Bug: 28435172
Change-Id: Ibd9301a990f12a94e8043b9c29a480f068251ba8

Use case: DUT unable to do connect with all devices in
Multipairing case.

Steps:
1. pair with HS
2. disconnect HS.
3. reconnect HS from DUT.

Failure: Authentication failure occurred due to LMP
transaction collision DUT not deleting link key which
lead to unable to connect to device.

Root Cause: Not posting the Authentication failure
event to application layer, while removing the link-key.

Fix: Post the Authentication failure event to application
and removing the link-key.

Test: code compilation.

Change-Id: I079d8ce89c1fea66f0624127514cec8405bfdd3d

Change the thread that the timeout posts a connection request event to
so that the state machine handles calling the callback on the correct
thread to prevent a JNI crash.

Bug: 32754737
Test: Manually paired a device that normally crashes phone
Change-Id: I2c3384feaa404cc583e81b2df994b7cbd079d25d

Bug: 32780409
Test: try pairing with nRF52DK using OOB LE SC
Change-Id: I3c165843bb76c372b76bdc18a7d9226345d39037

When address type is not parsed, creating bond to devices not using
random address is impossible.

Bug: 32780409
Test: try pairing with nRF52DK using random address
Change-Id: I95c4ebc7bd03cd1b895c9f17deef2e57d8b57d2a

Test: mma -j32
Change-Id: I1df0e71cba73cfde0d973498698c7d6fa017e14a

clang-format -i --style=file device/include/* \
  device/src/* device/test/*

Test: mma -j32
Change-Id: I1a0f9e1cd838c025fca487347358cc0079265004

Test: mma -j32
Change-Id: Ic945ac421b1918527105c59674eda89afd3d4126

clang-format -i --style=file bta/*/*

(twice)

Test: mma -j32
Change-Id: Ib118b1dfb6a34f9a5bfe153d65841e9041165264

bta: Separate includes of different types

Include "bta_sys.h" in bta/dm/bta_dm_int.h

Test: mma -j32
Change-Id: I292bc6518230201e1175937b9c6766c19709c9d5

Double asterisks at the beginning of the line
  sed -i s,"^[*][*]"," *", bta/*/*

Double asterisks after a space
  sed -i s,"^ [*][*]\([^*]\)"," *\1", bta/*/*
  sed -i s,"^ [*][*]$"," *", bta/*/*

Test: mma -j32
Change-Id: Ib83b802c7000176683b4e7f24d1255b3c2c6c01c

clang-format -i --style=file include/*

Test: mma -j32
Change-Id: Ibfb31af3fa82cfd6021fdc3203381b91e57bb71e

In preparation for clang-format, fix a few comments that
would have been mangled.

Test: mma -j32
Change-Id: I7194818e93432f8bad4d7c114a93025e217aa694

Test: mma -j32
Change-Id: I2d0df65001c82b37f001ba4697e58c206f2a4b9c

This reverts commit ec7a3f8f.

Change-Id: I862272c682409778dce356c8cba497ea4b78b14d