Re: Bluez stopped connecting A2DP sink

[Luiz Augusto von Dentz <luiz.dentz@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 3384 bytes --]

Hi,

On Thu, Oct 10, 2019 at 11:13 PM yayfortrees <yayfortrees@gmail.com> wrote:
>
> Hello,
>
> I've attached debug logs for 5.50 nopatch and 5.50 with the patch.
>
> To me it looked like it may be a timing issue - trying to connect the
> a2dp sink too early. As an experiment I put a sleep(1) at the top of
> the hs_cb function from your patch and the a2dp sink was connected.
> With the sleep I still get the "Device or resource busy" error early
> on, but it does reconnect the a2dp sink later in the connection and
> things work. It's not consistent though, maybe 1 of 5 attempts would
> fail to connect the a2dp profile.

Try with the following patch.

> I included that debug log as well.
>
> Hope this helps, thanks!
>
> On Thu, Oct 10, 2019 at 2:10 AM Luiz Augusto von Dentz
> <luiz.dentz@gmail.com> wrote:
> >
> > Hi,
> >
> > On Wed, Oct 9, 2019 at 10:51 AM Luiz Augusto von Dentz
> > <luiz.dentz@gmail.com> wrote:
> > >
> > > Hi,
> > >
> > > On Wed, Oct 9, 2019 at 1:56 AM yayfortrees <yayfortrees@gmail.com> wrote:
> > > >
> > > > Hello,
> > > >
> > > > I've attached btmon traces for 5.50 and 5.51. Thanks for taking a look!
> > > >
> > > > Let me know if you need anything else.
> > >
> > > Must likely the following error has something to do with it:
> > >
> > > = bluetoothd: a2dp-sink profile connect failed for 34:DF:2A:0D:F8:C2:
> > > Device or resource busy
> > >
> > > I will check why this could be happening.
> >
> > I tried this one and couldn't reproduce with the headsets I have,
> > could you please run bluetooth -dn and attach the logs here?
> >
> > > > On Tue, Oct 8, 2019 at 4:24 AM Luiz Augusto von Dentz
> > > > <luiz.dentz@gmail.com> wrote:
> > > > >
> > > > > Hi,
> > > > >
> > > > > On Mon, Oct 7, 2019 at 10:34 PM yayfortrees <yayfortrees@gmail.com> wrote:
> > > > > >
> > > > > > Hello,
> > > > > >
> > > > > > I recently upgraded bluez to 5.51 from 5.50 and my headphones stopped
> > > > > > using the A2DP sink during autoconnect. The headphones instead go to
> > > > > > HSP/HFP mode and the A2DP sink is disabled in pavucontrol/pactl. The
> > > > > > A2DP sink will only work when the device is initially paired.
> > > > > >
> > > > > > This had worked flawlessly for years. I was able to track down the
> > > > > > regression to this patch:
> > > > > > https://www.spinics.net/lists/linux-bluetooth/msg76180.html ([PATCH
> > > > > > BlueZ] policy: Add logic to connect a Sink). If I apply the patch to
> > > > > > 5.50, my headphones no longer connect to the A2DP sink, if I remove it
> > > > > > the problem goes away.
> > > > > >
> > > > > > Not sure what else I should include to help track this down. The
> > > > > > headphones I'm using are Beats Wireless (Solo Bluetooth).
> > > > > >
> > > > > > The only relevant log entry I see with the patch applied is:
> > > > > > a2dp-sink profile connect failed for 34:DF:2A:0D:F8:C2: Device or resource busy
> > > > >
> > > > > Can you collect the HCI traces of the problem using btmon? I didn't
> > > > > expect that to cause any problems, it should in theory solve them but
> > > > > lets see.
> > > > >
> > > > > > Let me know if any other information is needed.
> > > > > > Thanks.
> > > > >
> > > > >
> > > > >
> > > > > --
> > > > > Luiz Augusto von Dentz
> > >
> > >
> > >
> > > --
> > > Luiz Augusto von Dentz
> >
> >
> >
> > --
> > Luiz Augusto von Dentz



-- 
Luiz Augusto von Dentz

[-- Attachment #2: 0001-audio-Fix-cancelling-disconnect-timeout.patch --]
[-- Type: text/x-patch, Size: 3060 bytes --]

From 4bbd477cf6ac05aff73ee74f18a1a418b1c1af9f Mon Sep 17 00:00:00 2001
From: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Date: Fri, 11 Oct 2019 15:40:04 +0300
Subject: [PATCH BlueZ] audio: Fix cancelling disconnect timeout

If AVDTP session has been connected but no stream has been setup calls
to service->connect would return -EBUSY to avoid stream setup collision
but it also takes a reference to the session which cancelling the
disconnect timeout and disabling the stream_setup logic that would
attempt to estabilish a stream.
---
 profiles/audio/sink.c   | 25 ++++++++++++-------------
 profiles/audio/source.c | 25 ++++++++++++-------------
 2 files changed, 24 insertions(+), 26 deletions(-)

diff --git a/profiles/audio/sink.c b/profiles/audio/sink.c
index 7cac21034..966440534 100644
--- a/profiles/audio/sink.c
+++ b/profiles/audio/sink.c
@@ -256,11 +256,18 @@ gboolean sink_setup_stream(struct btd_service *service, struct avdtp *session)
 	if (sink->connect_id > 0 || sink->disconnect_id > 0)
 		return FALSE;
 
-	if (session && !sink->session)
-		sink->session = avdtp_ref(session);
-
-	if (!sink->session)
-		return FALSE;
+	if (!sink->session) {
+		if (session)
+			sink->session = avdtp_ref(session);
+		else
+			sink->session = a2dp_avdtp_get(
+					btd_service_get_device(service));
+
+		if (!sink->session) {
+			DBG("Unable to get a session");
+			return FALSE;
+		}
+	}
 
 	sink->connect_id = a2dp_discover(sink->session, discovery_complete,
 								sink);
@@ -274,14 +281,6 @@ int sink_connect(struct btd_service *service)
 {
 	struct sink *sink = btd_service_get_user_data(service);
 
-	if (!sink->session)
-		sink->session = a2dp_avdtp_get(btd_service_get_device(service));
-
-	if (!sink->session) {
-		DBG("Unable to get a session");
-		return -EIO;
-	}
-
 	if (sink->connect_id > 0 || sink->disconnect_id > 0)
 		return -EBUSY;
 
diff --git a/profiles/audio/source.c b/profiles/audio/source.c
index 4081e1970..0ac20fe40 100644
--- a/profiles/audio/source.c
+++ b/profiles/audio/source.c
@@ -257,11 +257,18 @@ gboolean source_setup_stream(struct btd_service *service,
 	if (source->connect_id > 0 || source->disconnect_id > 0)
 		return FALSE;
 
-	if (session && !source->session)
-		source->session = avdtp_ref(session);
-
-	if (!source->session)
-		return FALSE;
+	if (!source->session) {
+		if (session)
+			source->session = avdtp_ref(session);
+		else
+			source->session = a2dp_avdtp_get(
+					btd_service_get_device(service));
+
+		if (!source->session) {
+			DBG("Unable to get a session");
+			return FALSE;
+		}
+	}
 
 	source->connect_id = a2dp_discover(source->session, discovery_complete,
 								source);
@@ -275,14 +282,6 @@ int source_connect(struct btd_service *service)
 {
 	struct source *source = btd_service_get_user_data(service);
 
-	if (!source->session)
-		source->session = a2dp_avdtp_get(btd_service_get_device(service));
-
-	if (!source->session) {
-		DBG("Unable to get a session");
-		return -EIO;
-	}
-
 	if (source->connect_id > 0 || source->disconnect_id > 0)
 		return -EBUSY;
 
-- 
2.21.0