Bug: 298680973
Test: mmm packages/modules/Bluetooth
(cherry picked from https://googleplex-android-review.googlesource.com/q/commit:a94ef57ebc136a62cbf322a9f577fcb1caf93da7)
Merged-In: Iccb65bc4ede79b656b5b17d3596d99744c168861
Change-Id: Iccb65bc4ede79b656b5b17d3596d99744c168861

* changes:
  Fix Bluetooth app dependency
  Jarjar to prevent conflict w fwk-bluetooth.jar
  Jarjar to prevent conflict w system/framework.jar

Bug: 291521892
Test: mma
Test: m com.android.btservices
Change-Id: Ica18ee891c97919f9a421154dff9aaf7ba3d8898

Merge "Avatar: Modify "test_music_start_dual_device" test case for system behavior changed" into main

* changes:
  Nit clean of unused declaration
  Add ble sysprop for vnd_included

Bug: 260493132
Test: mma
Test: m com.android.btservices
Change-Id: I9b623b351d8e4274725171bb9fca0993d0510719

Test: m com.android.btservices
Bug: 282061758
Change-Id: I43317d2f3b41588a17bf40f8d1591e22d8aa31dc

For testing legacy device and service discovery

Bug: 298523541
Test: atest bt_host_test_bta
Change-Id: I8e8e260ada3f90e2b77b2a6aba756effa613aa92

* changes:
  Force reorder-init-list as error
  Fix reorder-init-list in power telemetry

Test: atest avatar
Bug: 301325580

Change-Id: Ia32bdcd905992b97a67eae17930f0edd73b903be

* changes:
  Consistent storage keys
  Remove entire storage section when the device is unbonded

Volume control service was missing information about group being active
and was no able to change the volume for the active group.

Issue could be observed after phone restart or when switching between
two LE HS

Bug: 302142886
Test: atest VolumeControlServiceTest
Tag: #feature
Change-Id: Idc11e503d2c0cea67ce0738fabde0adb0797a358

Failing scenario fixed in this patch
1. ANdroid bonded with Dual Mode device over BR/EDR and later over LE
   using CTKD
2. CSIS having cached Advertising Reports, sends to Java RPA of already
   bonded device as a next set member
3. The above leads to read second set member being no bonded.

Bug: 290914667
Test: atest bluetooth_csis_test   for regression
Test: manual 10 times pairing in the raw
Tag: #feature
Change-Id: I7e543ff4feeb4dab650bdeb5bb60593ba065c1f8

Merge "let HapClient profile to update connection policy with other LE audio GATT based profile" into main

Since we fixed all warning, it would be better to not introduce any
futur regression, hence adding the flag by default

Bug: 297606425
Test: m Bluetooth | this is no-op, only impact build
Change-Id: Iac0b15bae84408804c5f9c1d4428a3201484d500

Follow up of aosp/2656937

Bug: 290393336
Test: m Bluetooth | no-op change, only need to build
Change-Id: Iebec9d1e683510eb703414e830bb6b83b7cd3bd5

The original BLE_VND_INCLUDED define method cannot be override
on runtime and some targets need to disable this feature(Eg: hikey960 board)

Test: m .
Test: bluetooth enabled by default after boot with the hikey960 board
Bug: 302408538

Change-Id: I6da644a22056c8198b0dcd63b98c12fe829d487f
Signed-off-by: Yongqin Liu <yongqin.liu@linaro.org>

```
stack_power_telemetry.cc:518:7: warning: ISO C++ requires field
designators to be specified in declaration order;
field 'channel_type' will be initialized after field 'src'
[-Wreorder-init-list]
      .src.cid = static_cast<uint16_t>(src_id),
      ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
stack_power_telemetry.cc:517:23: note: previous initialization for field
'channel_type' is here
      .channel_type = channel_type,
                      ^~~~~~~~~~~~
 ```

Bug: 297606425
Test: m Bluetooth | Only build needed no-op change
Change-Id: Ifad027799e58a93afb11ea97c54c30adefb104ea

Bug: 299365655
Test: m .

Change-Id: Ia3cbf8614f91b99d429a9f60bd23064e795a608c

1. Literal strings shouldn't be used as keys
2. Keys should use a consistent naming convention
3. Key groups like LE keys are should be bound together

Test: m com.android.btservices
Bug: 298479810
Change-Id: Ief56997ac18e6190bab1ef8fb031b610ff5f855f

Bug: 303926933
Test: m .
No-Typo-Check: legacy code

Change-Id: Ie277ed2b19195ee90f6bfc7e82ece8c198e1efb0

Every bonded device property is stored under a section corresponding to
the device. BTIF storage function for removing the device removes the
individual properties. However as the properties are added in the BT
stack, chances of some properties left alone during device removal
increases.
For example, "AddrType" property is not removed when a device is
removed. As a result, when reconnecting/repairing with such device, this
old property unintentionally gets assigned to the device.

Test: m com.android.btservices
Bug: 298479810
Change-Id: I5c1bceb708727e17771cb222abe79f986e55a573

Currently, GATT service discovery is scheduled immediately after
connection to new LE device.

Depending on timing of Link Layer, we might not read remote version
information before checking for Robust Caching support. In this case, we
would assume device doesn't support it.

After this patch, we would wait for "remote version information" command
to finish, and go back to schedulsed service discovery. This way, we
would alawys use Robust Cache when newly connected device supports it.

Test: Bond with a set of devices, where both of them have same GATT
    database hash. Verify service discovery is triggered only once in snoop
    log.
Bug: 293249869
Change-Id: Id87ce6f8aa4a4f23a33d963dbbc28ef0891a6d75

MediaSessionCompat will crash if the player is not
responding and this will propagate to AVRCP.
We need to handle these crashes too and update the
remote device.
As all crashes should be handled the same here,
replace SecurityException by Exception.

Bug: 290576021
Tag: #feature
Test: atest net_test_btpackets
Test: atest BluetoothInstrumentationTests
Change-Id: I43a37017e9ee49e44dc6f631813ec3eb4ba96a6f

The check of BT_PROPERTY_TYPE_OF_DEVICE property requires size of 4
bytes, but some of the read attempts only provide 1 byte of data.
This causes those read attempts to fail.

Even though the underlying data fits within 1 byte, there are a lot
of places which assume 4 bytes of data, so changing this to 1 byte
breaks those assumption and might introduce unwanted side effects.

This CL modifies those failed reads to provide exactly 4 bytes.
Additionally, this also replaces every accessors that uses enum to use
uint32_t instead, since enum size is not guaranteed to be 4 bytes.

Bug: 300418914
Tag: #floss
Test: m -j
Test: Manually, make sure we don't fail reading dev_type from storage
      by adding logs and just scan some advertisements.

Change-Id: Ie2a880813ed5e4666813e02bb6914ae52950bb09