The flakiness is caused by usage of fakeosi
and mock timer implementation. This change
replaces fakeosi and mock timer with real
implementation.

Bug: 311769946
Test: atest net_test_btif_hh
Change-Id: I81434dca67a34a4fc209abb551d5ab3f9842eadc

* changes:
  Security: Move parsing to HCI: sec requests
  Security: Move parsing to HCI: local OOB
  Security: Move parsing to HCI: Remote OOB
  Security: Move parsing to HCI: SP complete
  Security: Move parsing to HCI: IO Cap Rsp
  Security: Move parsing to HCI: cancel complete
  Use a single crypto_toolbox (GD)
  HCI: Add error codes in 5.3 to PDL

* changes:
  Unify the callback trigger in check_status_with_id
  Divide long single gap data in le_advertising_manager
  Use raw payload for data raw commands builder
  Add new flag for divide long single gap data

Bug: 301661850
Test: mma packages/modules/Bluetooth
Change-Id: I7e26602e32ebd4b40e88a32ee4b77ae42998b7d7

Bug: 310711695
Test: atest net_test_bta

Change-Id: I2effe4e7524db465214140abee1aed54f9c64c94

Test: m Bluetooth
Test: atest bluetooth_test_gd_unit
Bug: 303546333
Change-Id: I721a66c901b8ee48074eeb710c56898117feb747

When host send long single entry gap data > 251 bytes, gd
advertising_manager would reject instead of dividing into fragments.
We need to support this usage as it's allowed by Spec. For
advertising/scan response data, we need to divide the data if length is 252 ~ 255 bytes.
For periodic advertising data, we need to divide if length is 253 ~ 255
bytes.

Change is under flag. Tested with
adb shell device_config put bluetooth com.android.bluetooth.flags.divide_long_single_gap_data true

NO_IFTTT=Modification was to fix missing libraries failure

Test: atest bluetooth_test_gd_unit
Test: manual test with long gap data and check air traces
Bug: 310217895
Bug: 303546333
Change-Id: I94bc4dbadb7a53b7008b2dca4659dbd489503318

The security code should not have to parse packets.

Bug: 301661850
Test: mma -j32
Change-Id: Ib93760e92cfc432a7cb0927dc9c96bc00f813b70

The security code should not have to parse packets.

Bug: 301661850
Test: mma -j32
Change-Id: I6da2007d4af48ca9180ffbd0e9ea411d21c15580

The security code should not have to parse packets.

Bug: 301661850
Test: mma -j32
Change-Id: I0791250571fdacadaf297e3c32b6f0e159905c68

The security code should not have to parse packets.

Bug: 301661850
Test: mma -j32
Change-Id: I941ca006afd6a5e2f77dbd8222a8cbf81dbebc20

The security code should not have to parse packets.

Bug: 301661850
Test: mma -j32
Change-Id: Ib3f5b0deb4a011cce2c1e7fa27716b5330419045

The security code should not have to parse packets.

Also pass a peer address by value.

Bug: 301661850
Test: mma -j32
Change-Id: Ie72a5c5559f2ce19923540b345d68d83004e43c0

Bug: 301661850
Test: mma -j32
Change-Id: I3ddce69a3a112727632f78ab40a2b2432d973616

Bug: 301661850
Test: mma -j32
Change-Id: I4260e64d4891cfef1258ce3b169ce2ce8f35eb19

A gatt read response with non valid and out of sync status should
not trigger storying and proceeding to connected state of device.
Informing about connected device leads to invalid behavior e.g. not
exposing indirectly PACS/ASCS attribute values.

Tag: #feature
Bug: 304854941
Test: atest UnicastTest#DisconnectAclBeforeGettingReadResponses
Change-Id: Ifabbd97f3fa18f88b3d21592f6c6a3de55289028

* changes:
  Database: Reduce scope of the cache lock
  Database: release cache lock before callback

Bug: 291815510
Test: atest AdapterServiceTest
Change-Id: I961bc33be30af37c4cc5353847c3ace1205eeb66

Merge "Handle Bond state changes through the adapter service instead of relying on broadcast receivers." into main

Bug: 310009436
Test: atest net_test_bta

Change-Id: Ib24d79e69569bd249d449dff0fafaaaf5bfb7e80

Bug: 311238169
Test: atest net_test_btif

Change-Id: Idc8b1b4f3d66d248ebb11797a1cbd567b616c954

This allows ScanManager constructor to take a custom looper as a
parameter, allowing easier testing with use of TestLooper class.

Bug: 309702502
Test: atest BluetoothInstrumentationTests
Change-Id: I0a036c5db99d517e799873f695aec6532307f33e

We do not need to lock the cache when checking parameter.
This is no-op change that will help if there is any unseen deadlock. It
also just a good practice to not lock when we do not need it

Bug: 310837813
Test: atest DatabaseManagerTest | current test should still pass
Change-Id: Ifa029c35196bfe0da3daa83e3b229d9048e92518

Bug: 310837813
Fix: 310837813
Test: atest setCustomMetadata_reentrantCallback_noDeadLock
Change-Id: I370a549a24b5a2e8793a394e56de934fa527ebba

This reverts commit b335f433.

Reason for revert: The original CL can be submitted independently. It doesn't need PDD review.

Change-Id: I19a15107a2ca81ea687c3025b5b8a241b850aef3

Bug: 308497929
Test: m Bluetooth | No test require as flag is added for follow-up use
Change-Id: I781d37d7188b7e9a0c3a24add00be0666224ef42

Bug: 308497929
Test: m Bluetooth | No test require when adding sysprop
Change-Id: I538da9e91f501d21f65f234b3e403faef68e57fb

Test: m Bluetooth
Test: atest bluetooth_test_gd_unit
Bug: 303546333
Change-Id: I4228a47a22418e61678163327339d88db9653e55