Kernel driver for MT6622

Kernel driver for MT6622

[Антон Бутанаев]

Hi, bluetooth developers!

I’m working on bluetooth driver for MT6622 bluetooth chip on MSB 2531
board. I need to be able to forward sound from that device to
bluetooth headset using A2DP profile. That MT6622 chip has UART line
for HCI control commands and I2S line for sound. I use bluez 5.50,
bluez-alsa 1.4.0 and kernel 3.10.69 in buildroot. I’ve already gone so
far that scanning devices, paring and connecting headset works now:

# bluetoothctl -- info C0:7A:A5:01:48:F0
Device C0:7A:A5:01:48:F0 (public)
Name: SVEN SEB-B270MV
Alias: SVEN SEB-B270MV
Class: 0x00240404
Icon: audio-card
Paired: yes
Trusted: no
Blocked: no
Connected: yes
LegacyPairing: no
UUID: Headset (00001108-0000-1000-8000-00805f9b34fb)
UUID: Audio Sink (0000110b-0000-1000-8000-00805f9b34fb)
UUID: A/V Remote Control Target (0000110c-0000-1000-8000-00805f9b34fb)
UUID: A/V Remote Control (0000110e-0000-1000-8000-00805f9b34fb)
UUID: Handsfree (0000111e-0000-1000-8000-00805f9b34fb)

But when I try to play sound, I see that I’m almost there: state
changed to SINK_STATE_PLAYING, then Starting IO loop: A2DP Source, but
after that I got “Broken pipe” error (here is some logs with debug
info):

# aplay -D btheadset /my/maybe-next-time.wav
../shared/ctl-client.c:108: Connecting to socket: /var/run/bluealsa/hci0
bluealsa: ctl.c:596: Received new connection: 8
bluealsa: ctl.c:609: New client accepted: 8
bluealsa: ctl.c:634: +-+-
../shared/ctl-client.c:238: Getting transport for C0:7A:A5:01:48:F0 type 0x41
bluealsa: ctl.c:634: +-+-
bluealsa-pcm.c:583: Setting constraints: 3
Playing WAVE '/my/maybe-next-time.wav' : Signed 16 bit Little Endian,
Rate 44100 Hz, Stereo
bluealsa-pcm.c:307: Initializing HW: 3
../shared/ctl-client.c:397: Requesting PCM open for C0:7A:A5:01:48:F0
bluealsa: ctl.c:329: PCM requested for C0:7A:A5:01:48:F0 type 0x41
bluetoothd[668]: profiles/audio/transport.c:media_owner_create() Owner
created: sender=:1.1
bluetoothd[668]: profiles/audio/avdtp.c:avdtp_ref() 0x1ecc348: ref=2
bluetoothd[668]: profiles/audio/a2dp.c:a2dp_sep_lock() SEP 0x1ec7cd8 locked
bluetoothd[668]: profiles/audio/avdtp.c:avdtp_ref() 0x1ecc348: ref=3
bluetoothd[668]: profiles/audio/a2dp.c:setup_ref() 0x1ecc2a8: ref=1
bluetoothd[668]: profiles/audio/transport.c:transport_set_state()
State changed /org/bluez/hci0/dev_C0_7A_A5_01_48_F0/fd0:
TRANSPORT_STATE_IDLE -> TRANSPORT_STATE_REQUESTING
bluetoothd[668]: profiles/audio/transport.c:media_request_create()
Request created: method=Acquire id=4
bluetoothd[668]: profiles/audio/transport.c:media_owner_add() Owner
:1.1 Request Acquire
bluetoothd[668]:
profiles/audio/transport.c:media_transport_set_owner() Transport
/org/bluez/hci0/dev_C0_7A_A5_01_48_F0/fd0 Owner :1.1
bluetoothd[668]: profiles/audio/avdtp.c:session_cb()
bluetoothd[668]: profiles/audio/avdtp.c:avdtp_parse_resp() START
request succeeded
bluetoothd[668]: profiles/audio/a2dp.c:start_cfm() Source 0x1ec7cd8: Start_Cfm
bluetoothd[668]: /org/bluez/hci0/dev_C0_7A_A5_01_48_F0/fd0: fd(17) ready
bluetoothd[668]: profiles/audio/transport.c:media_owner_remove() Owner
:1.1 Request Acquire
bluetoothd[668]: profiles/audio/transport.c:transport_set_state()
State changed /org/bluez/hci0/dev_C0_7A_A5_01_48_F0/fd0:
TRANSPORT_STATE_REQUESTING -> TRANSPORT_STATE_ACTIVE
bluetoothd[668]: profiles/audio/a2dp.c:setup_unref() 0x1ecc2a8: ref=0
bluetoothd[668]: profiles/audio/a2dp.c:setup_free() 0x1ecc2a8
bluetoothd[668]: profiles/audio/avdtp.c:avdtp_unref() 0x1ecc348: ref=2
bluetoothd[668]: profiles/audio/avdtp.c:avdtp_sep_set_state() stream
state changed: OPEN -> STREAMING
bluetoothd[668]: profiles/audio/sink.c:sink_set_state() State changed
/org/bluez/hci0/dev_C0_7A_A5_01_48_F0: SINK_STATE_CONNECTED ->
SINK_STATE_PLAYING
bluetoothd[668]: profiles/audio/transport.c:transport_update_playing()
/org/bluez/hci0/dev_C0_7A_A5_01_48_F0/fd0 State=TRANSPORT_STATE_ACTIVE
Playing=1
bluealsa: ba-transport.c:669: New transport: 18 (MTU: R:672 W:895)
bluealsa: ctl.c:634: +-+-
bluealsa-pcm.c:330: FIFO buffer size: 4096
bluealsa-pcm.c:336: Selected HW buffer: 6 periods x 16380 bytes <= 98284 bytes
bluealsa-pcm.c:351: Initializing SW: 3
bluealsa-pcm.c:351: Initializing SW: 3
bluealsa-pcm.c:351: Initializing SW: 3
bluealsa-pcm.c:373: Prepared: 3
bluealsa-pcm.c:351: Initializing SW: 3
bluealsa: bluez.c:1396: Signal: PropertiesChanged:
org.bluez.MediaTransport1: State
bluealsa: ba-transport.c:546: State transition: 0 -> 2
bluealsa: ba-transport.c:105: Created new IO thread: A2DP Source (SBC)
bluealsa: io.c:428: Starting IO loop: A2DP Source (SBC)
bluealsa-pcm.c:228: Starting: 3
../shared/ctl-client.c:444: Requesting PCM resume for C0:7A:A5:01:48:F0
bluealsa: ba-transport.c:546: State transition: 2 -> 2
bluealsa: ctl.c:634: +-+-
bluealsa-pcm.c:122: Starting IO loop: 6
bluetoothd[668]: profiles/audio/media.c:media_endpoint_exit()
media_endpoint_exit
bluetoothd[668]: profiles/audio/media.c:media_endpoint_exit()
media_endpoint_exit
bluetoothd[668]: src/adapter.c:adapter_service_remove() /org/bluez/hci0
bluetoothd[668]: src/adapter.c:remove_uuid() sending remove uuid
command for index 0
bluetoothd[668]: src/sdpd-service.c:remove_record_from_server()
Removing record with handle 0x10004
bluetoothd[668]: profiles/audio/media.c:release_endpoint() sender=:1.1
path=/hci0/A2DP/SBC/Sink/1
bluetoothd[668]: Endpoint unregistered: sender=:1.1 path=/hci0/A2DP/SBC/Sink/1
bluetoothd[668]: profiles/audio/media.c:media_endpoint_destroy()
sender=:1.1 path=/hci0/A2DP/SBC/Sink/1
bluetoothd[668]: profiles/audio/avdtp.c:avdtp_unregister_sep() SEP
0x1ec7d00 unregistered: type:1 codec:0 seid:2
bluetoothd[668]: profiles/audio/media.c:media_endpoint_exit()
media_endpoint_exit
ALSA lib bluealsa-pcm.c:201:(io_thread) PCM FIFO write error: Broken pipe
bluealsa-pcm.c:92: IO thread cleanup: 3
underrun!!! (at least 22.550 ms long)
aplay: xrun:1642: xrun: prepare error: No such device
bluealsa-pcm.c:268: Stopping: 3

Now I’m struggling at debugging the problem. It’s actually quite hard
for me since bluez uses event driven architecture and call stacks in
debugger usually point to some event dispatcher and it’s difficult to
understand what code actually triggered what action.
It’s unclear for me, what cased what, whether media_endpoint_exit() call:

bluetoothd[668]: profiles/audio/media.c:media_endpoint_exit()
media_endpoint_exit

cased broken pipe:

ALSA lib bluealsa-pcm.c:201:(io_thread) PCM FIFO write error: Broken pipe

since it appears earlier in logs, or vice versa.

I suspect that kernel bluetooth driver for MT6622 chip which I develop
should provide some channel for bluetoothd daemon to be able to
forward PCM data to the chip and without such channel broken pipe
happens.
The main obstacle for me now is that I’m having hard time finding any
documentation about what low level interface should kernel driver
provide for bluez for A2DP sound playing via I2S. I also miss big
picture of how bluez communicates with bluetooth chip via UART and I2S
lines, what should be done at kernel driver level and how. Do I need
to integrate with sound kernel driver? I’ve found some presentation in
youtube but it was too general for me, I need more details about
bluetooth audio.
After spending some time debugging bluetoothd daemon code I suspect
that bluetoothd opens a socket or pipe for PCM data, and passes its
handle to bluez-alsa to put PCM data into. But since kernel driver
doesn’t implement such socket the problem arises.
Could you please point me to some documentation, presentation,
tutorial or, best of all, example kernel driver code? Any help would
be very appreciated.

Anton

Re: Kernel driver for MT6622

[Антон Бутанаев]

The problem was in SBC: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=856487
The following patch fixes it.

diff --git a/Linux/Chicago/buildroot/package/sbc/0001-Add-config.h-in-sbc-sbc_primitives.h.patch
b/Linux/Chicago/buildroot/package/sbc/0001-Add-config.h-in-sbc-sbc_primitives.h.patch
new file mode 100644
index 000000000..3ed5d8b4a
--- /dev/null
+++ b/Linux/Chicago/buildroot/package/sbc/0001-Add-config.h-in-sbc-sbc_primitives.h.patch
@@ -0,0 +1,33 @@
+From e460c51017a7d1e04511ea4d293296a3d422049a Mon Sep 17 00:00:00 2001
+From: Nobuhiro Iwamatsu <iwamatsu@nigauri.org>
+Date: Mon, 3 Apr 2017 01:42:26 +0900
+Subject: [PATCH] Add config.h in sbc/sbc_primitives.h
+
+When we use "--enable-high-precision" option of configure, this does not work
+effectively. Because SBC_HIGH_PRECISION enabled by this option is not available
+in "sbc/sbc_primitives.h".
+This adds config.h to "sbc/sbc_primitives.h" to use SBC_HIGH_PRECISION.
+
+Signed-off-by: Nobuhiro Iwamatsu <iwamatsu@nigauri.org>
+---
+ sbc/sbc_primitives.h | 4 ++++
+ 1 file changed, 4 insertions(+)
+
+diff --git a/sbc/sbc_primitives.h b/sbc/sbc_primitives.h
+index e01c957..b060484 100644
+--- a/sbc/sbc_primitives.h
++++ b/sbc/sbc_primitives.h
+@@ -27,6 +27,10 @@
+ #ifndef __SBC_PRIMITIVES_H
+ #define __SBC_PRIMITIVES_H
+
++#ifdef HAVE_CONFIG_H
++#include <config.h>
++#endif
++
+ #define SCALE_OUT_BITS 15
+ #define SBC_X_BUFFER_SIZE 328
+
+--
+2.11.0
+
diff --git a/Linux/Chicago/buildroot/package/sbc/sbc.mk
b/Linux/Chicago/buildroot/package/sbc/sbc.mk
index 77661c189..fbba27902 100644
--- a/Linux/Chicago/buildroot/package/sbc/sbc.mk
+++ b/Linux/Chicago/buildroot/package/sbc/sbc.mk
@@ -11,5 +11,6 @@ SBC_INSTALL_STAGING = YES
 SBC_DEPENDENCIES = libsndfile host-pkgconf
 SBC_LICENSE = GPL-2.0+ (programs), LGPL-2.1+ (library)
 SBC_LICENSE_FILES = COPYING COPYING.LIB
+SBC_CONF_OPTS = --enable-high-precision

 $(eval $(autotools-package))

пн, 27 мая 2019 г. в 10:26, Антон Бутанаев <anton.butanaev@gmail.com>:
>
> Hi, bluetooth developers!
>
> I’m working on bluetooth driver for MT6622 bluetooth chip on MSB 2531
> board. I need to be able to forward sound from that device to
> bluetooth headset using A2DP profile. That MT6622 chip has UART line
> for HCI control commands and I2S line for sound. I use bluez 5.50,
> bluez-alsa 1.4.0 and kernel 3.10.69 in buildroot. I’ve already gone so
> far that scanning devices, paring and connecting headset works now:
>
> # bluetoothctl -- info C0:7A:A5:01:48:F0
> Device C0:7A:A5:01:48:F0 (public)
> Name: SVEN SEB-B270MV
> Alias: SVEN SEB-B270MV
> Class: 0x00240404
> Icon: audio-card
> Paired: yes
> Trusted: no
> Blocked: no
> Connected: yes
> LegacyPairing: no
> UUID: Headset (00001108-0000-1000-8000-00805f9b34fb)
> UUID: Audio Sink (0000110b-0000-1000-8000-00805f9b34fb)
> UUID: A/V Remote Control Target (0000110c-0000-1000-8000-00805f9b34fb)
> UUID: A/V Remote Control (0000110e-0000-1000-8000-00805f9b34fb)
> UUID: Handsfree (0000111e-0000-1000-8000-00805f9b34fb)
>
> But when I try to play sound, I see that I’m almost there: state
> changed to SINK_STATE_PLAYING, then Starting IO loop: A2DP Source, but
> after that I got “Broken pipe” error (here is some logs with debug
> info):
>
> # aplay -D btheadset /my/maybe-next-time.wav
> ../shared/ctl-client.c:108: Connecting to socket: /var/run/bluealsa/hci0
> bluealsa: ctl.c:596: Received new connection: 8
> bluealsa: ctl.c:609: New client accepted: 8
> bluealsa: ctl.c:634: +-+-
> ../shared/ctl-client.c:238: Getting transport for C0:7A:A5:01:48:F0 type 0x41
> bluealsa: ctl.c:634: +-+-
> bluealsa-pcm.c:583: Setting constraints: 3
> Playing WAVE '/my/maybe-next-time.wav' : Signed 16 bit Little Endian,
> Rate 44100 Hz, Stereo
> bluealsa-pcm.c:307: Initializing HW: 3
> ../shared/ctl-client.c:397: Requesting PCM open for C0:7A:A5:01:48:F0
> bluealsa: ctl.c:329: PCM requested for C0:7A:A5:01:48:F0 type 0x41
> bluetoothd[668]: profiles/audio/transport.c:media_owner_create() Owner
> created: sender=:1.1
> bluetoothd[668]: profiles/audio/avdtp.c:avdtp_ref() 0x1ecc348: ref=2
> bluetoothd[668]: profiles/audio/a2dp.c:a2dp_sep_lock() SEP 0x1ec7cd8 locked
> bluetoothd[668]: profiles/audio/avdtp.c:avdtp_ref() 0x1ecc348: ref=3
> bluetoothd[668]: profiles/audio/a2dp.c:setup_ref() 0x1ecc2a8: ref=1
> bluetoothd[668]: profiles/audio/transport.c:transport_set_state()
> State changed /org/bluez/hci0/dev_C0_7A_A5_01_48_F0/fd0:
> TRANSPORT_STATE_IDLE -> TRANSPORT_STATE_REQUESTING
> bluetoothd[668]: profiles/audio/transport.c:media_request_create()
> Request created: method=Acquire id=4
> bluetoothd[668]: profiles/audio/transport.c:media_owner_add() Owner
> :1.1 Request Acquire
> bluetoothd[668]:
> profiles/audio/transport.c:media_transport_set_owner() Transport
> /org/bluez/hci0/dev_C0_7A_A5_01_48_F0/fd0 Owner :1.1
> bluetoothd[668]: profiles/audio/avdtp.c:session_cb()
> bluetoothd[668]: profiles/audio/avdtp.c:avdtp_parse_resp() START
> request succeeded
> bluetoothd[668]: profiles/audio/a2dp.c:start_cfm() Source 0x1ec7cd8: Start_Cfm
> bluetoothd[668]: /org/bluez/hci0/dev_C0_7A_A5_01_48_F0/fd0: fd(17) ready
> bluetoothd[668]: profiles/audio/transport.c:media_owner_remove() Owner
> :1.1 Request Acquire
> bluetoothd[668]: profiles/audio/transport.c:transport_set_state()
> State changed /org/bluez/hci0/dev_C0_7A_A5_01_48_F0/fd0:
> TRANSPORT_STATE_REQUESTING -> TRANSPORT_STATE_ACTIVE
> bluetoothd[668]: profiles/audio/a2dp.c:setup_unref() 0x1ecc2a8: ref=0
> bluetoothd[668]: profiles/audio/a2dp.c:setup_free() 0x1ecc2a8
> bluetoothd[668]: profiles/audio/avdtp.c:avdtp_unref() 0x1ecc348: ref=2
> bluetoothd[668]: profiles/audio/avdtp.c:avdtp_sep_set_state() stream
> state changed: OPEN -> STREAMING
> bluetoothd[668]: profiles/audio/sink.c:sink_set_state() State changed
> /org/bluez/hci0/dev_C0_7A_A5_01_48_F0: SINK_STATE_CONNECTED ->
> SINK_STATE_PLAYING
> bluetoothd[668]: profiles/audio/transport.c:transport_update_playing()
> /org/bluez/hci0/dev_C0_7A_A5_01_48_F0/fd0 State=TRANSPORT_STATE_ACTIVE
> Playing=1
> bluealsa: ba-transport.c:669: New transport: 18 (MTU: R:672 W:895)
> bluealsa: ctl.c:634: +-+-
> bluealsa-pcm.c:330: FIFO buffer size: 4096
> bluealsa-pcm.c:336: Selected HW buffer: 6 periods x 16380 bytes <= 98284 bytes
> bluealsa-pcm.c:351: Initializing SW: 3
> bluealsa-pcm.c:351: Initializing SW: 3
> bluealsa-pcm.c:351: Initializing SW: 3
> bluealsa-pcm.c:373: Prepared: 3
> bluealsa-pcm.c:351: Initializing SW: 3
> bluealsa: bluez.c:1396: Signal: PropertiesChanged:
> org.bluez.MediaTransport1: State
> bluealsa: ba-transport.c:546: State transition: 0 -> 2
> bluealsa: ba-transport.c:105: Created new IO thread: A2DP Source (SBC)
> bluealsa: io.c:428: Starting IO loop: A2DP Source (SBC)
> bluealsa-pcm.c:228: Starting: 3
> ../shared/ctl-client.c:444: Requesting PCM resume for C0:7A:A5:01:48:F0
> bluealsa: ba-transport.c:546: State transition: 2 -> 2
> bluealsa: ctl.c:634: +-+-
> bluealsa-pcm.c:122: Starting IO loop: 6
> bluetoothd[668]: profiles/audio/media.c:media_endpoint_exit()
> media_endpoint_exit
> bluetoothd[668]: profiles/audio/media.c:media_endpoint_exit()
> media_endpoint_exit
> bluetoothd[668]: src/adapter.c:adapter_service_remove() /org/bluez/hci0
> bluetoothd[668]: src/adapter.c:remove_uuid() sending remove uuid
> command for index 0
> bluetoothd[668]: src/sdpd-service.c:remove_record_from_server()
> Removing record with handle 0x10004
> bluetoothd[668]: profiles/audio/media.c:release_endpoint() sender=:1.1
> path=/hci0/A2DP/SBC/Sink/1
> bluetoothd[668]: Endpoint unregistered: sender=:1.1 path=/hci0/A2DP/SBC/Sink/1
> bluetoothd[668]: profiles/audio/media.c:media_endpoint_destroy()
> sender=:1.1 path=/hci0/A2DP/SBC/Sink/1
> bluetoothd[668]: profiles/audio/avdtp.c:avdtp_unregister_sep() SEP
> 0x1ec7d00 unregistered: type:1 codec:0 seid:2
> bluetoothd[668]: profiles/audio/media.c:media_endpoint_exit()
> media_endpoint_exit
> ALSA lib bluealsa-pcm.c:201:(io_thread) PCM FIFO write error: Broken pipe
> bluealsa-pcm.c:92: IO thread cleanup: 3
> underrun!!! (at least 22.550 ms long)
> aplay: xrun:1642: xrun: prepare error: No such device
> bluealsa-pcm.c:268: Stopping: 3
>
> Now I’m struggling at debugging the problem. It’s actually quite hard
> for me since bluez uses event driven architecture and call stacks in
> debugger usually point to some event dispatcher and it’s difficult to
> understand what code actually triggered what action.
> It’s unclear for me, what cased what, whether media_endpoint_exit() call:
>
> bluetoothd[668]: profiles/audio/media.c:media_endpoint_exit()
> media_endpoint_exit
>
> cased broken pipe:
>
> ALSA lib bluealsa-pcm.c:201:(io_thread) PCM FIFO write error: Broken pipe
>
> since it appears earlier in logs, or vice versa.
>
> I suspect that kernel bluetooth driver for MT6622 chip which I develop
> should provide some channel for bluetoothd daemon to be able to
> forward PCM data to the chip and without such channel broken pipe
> happens.
> The main obstacle for me now is that I’m having hard time finding any
> documentation about what low level interface should kernel driver
> provide for bluez for A2DP sound playing via I2S. I also miss big
> picture of how bluez communicates with bluetooth chip via UART and I2S
> lines, what should be done at kernel driver level and how. Do I need
> to integrate with sound kernel driver? I’ve found some presentation in
> youtube but it was too general for me, I need more details about
> bluetooth audio.
> After spending some time debugging bluetoothd daemon code I suspect
> that bluetoothd opens a socket or pipe for PCM data, and passes its
> handle to bluez-alsa to put PCM data into. But since kernel driver
> doesn’t implement such socket the problem arises.
> Could you please point me to some documentation, presentation,
> tutorial or, best of all, example kernel driver code? Any help would
> be very appreciated.
>
> Anton