Bug: 232048916
Tag: #logging
Test: gd/cert/run --sl4a
Change-Id: If5a2970aa7b1ebdb85e59f2f9dd494f0837d1379

rerun-if-changed should not contain `-I`, and only the topmost directory
is needed because cargo scans for the directory recursively.

Bug: 221310414
Tag: #floss
Test: Manual - `cargo build` and check that topshim is only built once
unless a file inside /system is touched.

Change-Id: I4558d260498c79bc0a6d5362e36504159464c918

Bug: 232258524
Test: gd/cert/run
Tag: #refactor
BYPASS_LONG_LINES_REASON: Bluetooth likes 120 lines

Change-Id: I7a57eadc6186746fc400742bf8c7adea8965e8de

Bug: 232258524
Test: gd/cert/run
Tag: #refactor
BYPASS_LONG_LINES_REASON: Bluetooth likes 120 lines

Change-Id: I211c7bc2b38dc9d75a4be087a99e0b2ab8e22c17

* erase items before calling std::map::emplace in le_acl_facade
* add more logging to LE and classic ACL facades
* reduce GrpcEventQueue's wait time from 500ms to 100ms so that
  it can be terminated faster
* force a wait of 100ms after closing gRPC streams in Python

Bug: 228619929
Bug: 235872679
Bug: 232048916
Test: gd/cer/run --sl4a
Change-Id: Idbb12b55db96b00bb695c54af1fd0efcc1f65841

* changes:
  Revert "Disable sniff mode for Phonak device"
  Do not perform SSR when avdtp start

* changes:
  SL4A: Throw exception when test fails and clear paired devices
  Test: Zip dynamic libraries and push to device if not present

Call std::map::erase before std::map::emplace
* std::map::emplace does not replace existing value if the key
  already exists. The newly constructed element will be destroyed
  immediately if there already is an element with the key in the
  container. Hence, if we want to emplace a new value, we should
  always erase the old one or use std::map::insert_or_assign
* If the value is move-only, insert_or_assign may not work

Avoid shutting down gRPC server in signal handler
* See https://github.com/grpc/grpc/issues/24884
* Create a separate shutdown thread for this purpose

Make sure all facade server values are deleted when exit
* To avoid memory leak

Do not call server->Wait() on a deleted gRPC server
* Instead, use a std::uniptr to hold the server object until the
  wrapper class is freed

Bug: 228619929
Bug: 235872679
Bug: 232048916
Change-Id: Idf64df4ca61e0820ec9d48fdc9b0d191ddf0d352

Test: none
Change-Id: Ia1fc5d261fbb8b3d25c3694fafd4e90614f18fcb

To better reflect the actual range of the volume.

Bug: 194853979
Tag: #floss
Test: build, deploy, connect and check dbus-monitor
BYPASS_LONG_LINES_REASON: Bluetooth likes 120 char lines

Change-Id: I996fda92e2cb65f77912e93c3526d71f075d4377

* Return False no longer makes the test fail in mobly
  we should raise exception instead
* Clear paired devices before tests so that they do not interfer with
  later tests

Bug: 232048916
Tag: #test
Test: gd/cert/run --sl4a_sl4a --clean
Change-Id: Id1e4cb30e28a13203848cb5c406ecb784fe2315c

* Bluetooth mainline modules keeps its own copies of dynamic libraries
* Hence now GD cert binaries have to keep another copy when running

Test: gd/cert/run --sl4a
Bug: 232048916
Tag: #gd-refactor
Change-Id: I716b47f6b808cd42b0713f55ba08544011652730

BUG: 194853979
Tag: #floss
Test: build, deploy, connect and test by sending DBus message
BYPASS_LONG_LINES_REASON: Bluetooth likes 120 char lines

Change-Id: Ifd8139e514eedd9abf5f2bd00930046aafdaf1ea

Bug: 229748367
Tag: #floss
Test: Build
BYPASS_LONG_LINES_REASON: Bluetooth likes 120 char lines

Change-Id: I2922102c321b3e0894288da2ef33c84f054b92ba

Append the Bluetooth address parameter for SetVolume DBus message so the
audio client can use it to set HFP device's speaker volume.

Bug: 229748367
Tag: #floss
Test: build, deploy, connect and test by sending DBus message
BYPASS_LONG_LINES_REASON: Bluetooth likes 120 char lines

Change-Id: I39c87f41b25d3b684ef0c7bf4a04107390514fcd

* changes:
  floss: Add socket manager implementation
  floss: Serialize Option<T> and std::fs::File in dbus
  floss: Refactor insertion via dbus projection

Add IBluetoothSocketManager and wire it into the stack.

Bug: 233123287
Tag: #floss
Test: Manual tests on ChromeOS
Change-Id: I5c6660cffa49ac445daab2ca3d3cbc24804d112c

In order to represent open file descriptors, add serialization support
for Option<T> and std::fs::File (eventually, we will have
Option<std::fs::File>).

Bug: 233123489
Tag: #floss
Test: ./build.py
Change-Id: I0d23f257ae2c1b6aee10edcc1adcd3ee8840a485

Dbus projection currently requires that a single dbus object path only
has a single interface associated with it. This results in multiple
objects being created for several interfaces.

This change refactors generate_dbus_exporter to provide an Interface
token which must be separately inserted with the object path. It also
provides a mechanism to have Mixins, a single object that contains
multiple interface implementations.

Bug: 233123489
Tag: #floss
Test: ./build.py and manual test on ChromeOS
Change-Id: I82f2cd18d02276bf26f77541ba3592a73ca558fa

std::map::emplace does not replace existing value if the key
already exists.

The newly constructed element will be destroyed immediately
if there already is an element with the key in the container.

Bug: 235872679
Tag: #test
Test: gd/cert/run --sl4a --clean

Change-Id: I340b723e805c0b94128bf6665690571f09dc9529

We use EATT only for LE and this fix incorrect value provided to the
EATT implementation as peer_mtu.

It could result on incorrect GATT fragmentation, when EATT was created
by remote device.

Tag: #feature
Test: atest BluetoothInstrumentationTests
Bug: 235783869
Change-Id: I4dde8eeb60ab0b045960a81718b428acbc0c34f7

This makes it much easier to see the added and removed lines in the
logs. Use ‘less -R’ when viewing a log file to allow the escape
sequences to show up on the terminal.

Test: a provoked error shows a colorful diff: http://screen/bEf5puAqZGVbC5w
Change-Id: I3318015ec563e47be5317298464ef527d1809424

This allows us to show what we searched for and where. It makes the
error messages much more useful when you run ‘cargo test’ without
having the necessary binaries in your ‘$PATH’.

Test: ‘cargo test’ without having run ‘m pdl’
Change-Id: I98b3cc3c9d10fa930bc449faddc5c6186583ef54

We now show the unified diff relative to the known-good code. That is,
if we generate a line too much, this like shows up as an added line.
Before it would show up as a deleted line, which seems wrong.

Test: atest pdl_inline_tests
Change-Id: I9ecb24c4f79b65221131adfe0e44db4529a6a478

When user triggers a volume change through the Handsfree device, we need
to notify audio stack about the change to have the volume synced.

Bug: 229748367
Tag: #floss
Test: build, deploy, connect and check dbus-monitor
BYPASS_LONG_LINES_REASON: Bluetooth likes 120 char lines

Change-Id: Id14162af924df0b9c22ae02653050edfb115680e