2023-02-12 18:34:16

by Pauli Virtanen

Subject: [PATCH BlueZ] shared/bap: wait for GATT client ready before ASCS/PACS discovery

BAP client fails to find endpoints if GATT services are not resolved
when the GATT client is attached to the session. This condition
almost always occurs on the first BAP connection to a device between
bluetoothd restarts, and may occur also otherwise.

The BAP client code discovers ASCS/PACS services and characteristics at
client attach time.

Fix the connection issue by postponing PAC/ASE discovery until the
GATT client is ready, if it was not ready at attach time.
---

Notes:
Typical bluetoothctl output in the failing case, without this patch:

$ sudo systemctl restart bluetooth
$ bluetoothctl
...
[bluetooth]# connect XX:XX:XX:XX:XX:XX
Attempting to connect to XX:XX:XX:XX:XX:XX
[CHG] Device XX:XX:XX:XX:XX:XX Connected: yes
[CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000184e-0000-1000-8000-00805f9b34fb
[CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001850-0000-1000-8000-00805f9b34fb
[CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 03b80e5a-ede8-4b33-a751-6ce34ec4c700
Connection successful
[NEW] Primary Service (Handle 0x0000)
/org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0008
00001801-0000-1000-8000-00805f9b34fb
Generic Attribute Profile
...
[NEW] Descriptor (Handle 0x0000)
/org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0065/char0066/desc0069
00002901-0000-1000-8000-00805f9b34fb
Characteristic User Description
[CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001800-0000-1000-8000-00805f9b34fb
[CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001801-0000-1000-8000-00805f9b34fb
[CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000180a-0000-1000-8000-00805f9b34fb
[CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001844-0000-1000-8000-00805f9b34fb
[CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000184e-0000-1000-8000-00805f9b34fb
[CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001850-0000-1000-8000-00805f9b34fb
[CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 03b80e5a-ede8-4b33-a751-6ce34ec4c700
[CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: yes
[CHG] Device XX:XX:XX:XX:XX:XX Paired: yes
[xxx]# endpoint.list
[xxx]# disconnect
Attempting to disconnect from XX:XX:XX:XX:XX:XX
[CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: no
Successful disconnected
[CHG] Device XX:XX:XX:XX:XX:XX Connected: no
[bluetooth]# connect XX:XX:XX:XX:XX:XX
Attempting to connect to XX:XX:XX:XX:XX:XX
[CHG] Device XX:XX:XX:XX:XX:XX Connected: yes
Connection successful
[NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0
[NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0
[NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
[NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
[CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
[CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
[CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: yes
[xxx]# endpoint.list
Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0
Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0
[xxx]#

Endpoints and transports failed to appear on the first connect to the
BAP server, since GATT discovery was not yet completed when the BAP code
tried to discover the endpoints. Second connection succeeded.
Occasionally, it also happens that endpoints appear but transports do
not, but this is hard to reproduce.

With this patch:

$ sudo systemctl restart bluetooth
$ bluetoothctl
...
[bluetooth]# connect XX:XX:XX:XX:XX:XX
Attempting to connect to XX:XX:XX:XX:XX:XX
[CHG] Device XX:XX:XX:XX:XX:XX Connected: yes
[CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000184e-0000-1000-8000-00805f9b34fb
[CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001850-0000-1000-8000-00805f9b34fb
[CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 03b80e5a-ede8-4b33-a751-6ce34ec4c700
Connection successful
[NEW] Primary Service (Handle 0x0000)
/org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0008
00001801-0000-1000-8000-00805f9b34fb
Generic Attribute Profile
...
[NEW] Descriptor (Handle 0x0000)
/org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0065/char0066/desc0069
00002901-0000-1000-8000-00805f9b34fb
Characteristic User Description
[CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: yes
[CHG] Device XX:XX:XX:XX:XX:XX Paired: yes
[NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0
[NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0
[NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
[NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
[CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
[CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
[xxx]#

Now the first connection works properly.

On first reading, the spec does not seem to clearly comment if ASEs and
PACs could be removed/added by the server while client is connected. If
that's allowed, then we'd need some bigger changes here. Regardless,
waiting for GATT client ready before first scan seems good also in this
case.

src/shared/bap.c | 93 +++++++++++++++++++++++++++++++-----------------
1 file changed, 60 insertions(+), 33 deletions(-)

diff --git a/src/shared/bap.c b/src/shared/bap.c
index 22f2e6714..2d676d96f 100644
--- a/src/shared/bap.c
+++ b/src/shared/bap.c
@@ -158,6 +158,7 @@ struct bt_bap {
struct bt_att *att;
struct bt_bap_req *req;
unsigned int cp_id;
+ unsigned int client_ready_id;

unsigned int process_id;
unsigned int disconn_id;
@@ -3733,41 +3734,10 @@ static void bap_attach_att(struct bt_bap *bap, struct bt_att *att)
bap, NULL);
}

-bool bt_bap_attach(struct bt_bap *bap, struct bt_gatt_client *client)
+static void bap_attach_finish(struct bt_bap *bap)
{
bt_uuid_t uuid;

- if (queue_find(sessions, NULL, bap)) {
- /* If instance already been set but there is no client proceed
- * to clone it otherwise considered it already attached.
- */
- if (client && !bap->client)
- goto clone;
- return true;
- }
-
- if (!sessions)
- sessions = queue_new();
-
- queue_push_tail(sessions, bap);
-
- queue_foreach(bap_cbs, bap_attached, bap);
-
- if (!client) {
- bap_attach_att(bap, bap->att);
- return true;
- }
-
- if (bap->client)
- return false;
-
-clone:
- bap->client = bt_gatt_client_clone(client);
- if (!bap->client)
- return false;
-
- bap_attach_att(bap, bt_gatt_client_get_att(client));
-
if (bap->rdb->pacs) {
uint16_t value_handle;
struct bt_pacs *pacs = bap->rdb->pacs;
@@ -3798,7 +3768,7 @@ clone:

bap_notify_ready(bap);

- return true;
+ return;
}

bt_uuid16_create(&uuid, PACS_UUID);
@@ -3806,6 +3776,57 @@ clone:

bt_uuid16_create(&uuid, ASCS_UUID);
gatt_db_foreach_service(bap->rdb->db, &uuid, foreach_ascs_service, bap);
+}
+
+static void bap_attach_ready_cb(bool success, uint8_t att_ecode,
+ void *user_data)
+{
+ struct bt_bap *bap = user_data;
+
+ bap->client_ready_id = 0;
+
+ bap_attach_finish(bap);
+}
+
+bool bt_bap_attach(struct bt_bap *bap, struct bt_gatt_client *client)
+{
+ if (queue_find(sessions, NULL, bap)) {
+ /* If instance already been set but there is no client proceed
+ * to clone it otherwise considered it already attached.
+ */
+ if (client && !bap->client)
+ goto clone;
+ return true;
+ }
+
+ if (!sessions)
+ sessions = queue_new();
+
+ queue_push_tail(sessions, bap);
+
+ queue_foreach(bap_cbs, bap_attached, bap);
+
+ if (!client) {
+ bap_attach_att(bap, bap->att);
+ return true;
+ }
+
+ if (bap->client)
+ return false;
+
+clone:
+ bap->client = bt_gatt_client_clone(client);
+ if (!bap->client)
+ return false;
+
+ bap_attach_att(bap, bt_gatt_client_get_att(bap->client));
+
+ if (bt_gatt_client_is_ready(bap->client)) {
+ bap_attach_finish(bap);
+ } else {
+ bap->client_ready_id = bt_gatt_client_ready_register(
+ bap->client, bap_attach_ready_cb, bap, NULL);
+ }

return true;
}
@@ -3824,6 +3845,12 @@ void bt_bap_detach(struct bt_bap *bap)
if (!queue_remove(sessions, bap))
return;

+ if (bap->client_ready_id) {
+ bt_gatt_client_ready_unregister(bap->client,
+ bap->client_ready_id);
+ bap->client_ready_id = 0;
+ }
+
bt_gatt_client_unref(bap->client);
bap->client = NULL;

--
2.39.1

2023-02-12 20:05:43

by bluez.test.bot

Subject: RE: [BlueZ] shared/bap: wait for GATT client ready before ASCS/PACS discovery

This is automated email and please do not reply to this email!

Dear submitter,

Thank you for submitting the patches to the linux bluetooth mailing list.
This is a CI test results with your patch series:
PW Link:https://patchwork.kernel.org/project/bluetooth/list/?series=721100

---Test result---

Test Summary:
CheckPatch PASS 0.43 seconds
GitLint FAIL 0.53 seconds
BuildEll PASS 26.40 seconds
BluezMake PASS 787.59 seconds
MakeCheck PASS 11.08 seconds
MakeDistcheck PASS 147.95 seconds
CheckValgrind PASS 241.21 seconds
CheckSmatch PASS 319.48 seconds
bluezmakeextell PASS 95.77 seconds
IncrementalBuild PASS 604.62 seconds
ScanBuild PASS 954.53 seconds

Details
##############################
Test: GitLint - FAIL
Desc: Run gitlint
Output:
[BlueZ] shared/bap: wait for GATT client ready before ASCS/PACS discovery

WARNING: I3 - ignore-body-lines: gitlint will be switching from using Python regex 'match' (match beginning) to 'search' (match anywhere) semantics. Please review your ignore-body-lines.regex option accordingly. To remove this warning, set general.regex-style-search=True. More details: https://jorisroovers.github.io/gitlint/configuration/#regex-style-search
17: B2 Line has trailing whitespace: " "
60: B1 Line exceeds max length (132>80): " [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1"
61: B1 Line exceeds max length (132>80): " [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0"
67: B2 Line has trailing whitespace: " "
73: B2 Line has trailing whitespace: " "
75: B2 Line has trailing whitespace: " "
101: B1 Line exceeds max length (132>80): " [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1"
102: B1 Line exceeds max length (132>80): " [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0"
104: B2 Line has trailing whitespace: " "
106: B2 Line has trailing whitespace: " "


---
Regards,
Linux Bluetooth

2023-02-12 20:46:03

by Luiz Augusto von Dentz

Subject: Re: [PATCH BlueZ] shared/bap: wait for GATT client ready before ASCS/PACS discovery

Hi Pauli,

On Sun, Feb 12, 2023 at 10:41 AM Pauli Virtanen <[email protected]> wrote:
>
> BAP client fails to find endpoints if GATT services are not resolved
> when the GATT client is attached to the session. This condition
> almost always occurs on the first BAP connection to a device between
> bluetoothd restarts, and may occur also otherwise.

Hmm, this shouldn't happen if you have caching enabled, did you
disable it for some reason?

> The BAP client code discovers ASCS/PACS services and characteristics at
> client attach time.
>
> Fix the connection issue by postponing PAC/ASE discovery until the
> GATT client is ready, if it was not ready at attach time.

This should probably be done by the core, it should check if the
services are cached then it can call accept, otherwise it needs to
wait for the client to be ready, perhaps there is something not quite
right with the way to probe services as afaik if we don't have any
cache the driver shouldn't have been called to begin with.

> ---
>
> Notes:
> Typical bluetoothctl output in the failing case, without this patch:
>
> $ sudo systemctl restart bluetooth
> $ bluetoothctl
> ...
> [bluetooth]# connect XX:XX:XX:XX:XX:XX
> Attempting to connect to XX:XX:XX:XX:XX:XX
> [CHG] Device XX:XX:XX:XX:XX:XX Connected: yes
> [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000184e-0000-1000-8000-00805f9b34fb
> [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001850-0000-1000-8000-00805f9b34fb
> [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 03b80e5a-ede8-4b33-a751-6ce34ec4c700
> Connection successful
> [NEW] Primary Service (Handle 0x0000)
> /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0008
> 00001801-0000-1000-8000-00805f9b34fb
> Generic Attribute Profile
> ...
> [NEW] Descriptor (Handle 0x0000)
> /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0065/char0066/desc0069
> 00002901-0000-1000-8000-00805f9b34fb
> Characteristic User Description
> [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001800-0000-1000-8000-00805f9b34fb
> [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001801-0000-1000-8000-00805f9b34fb
> [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000180a-0000-1000-8000-00805f9b34fb
> [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001844-0000-1000-8000-00805f9b34fb
> [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000184e-0000-1000-8000-00805f9b34fb
> [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001850-0000-1000-8000-00805f9b34fb
> [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 03b80e5a-ede8-4b33-a751-6ce34ec4c700
> [CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: yes
> [CHG] Device XX:XX:XX:XX:XX:XX Paired: yes
> [xxx]# endpoint.list
> [xxx]# disconnect
> Attempting to disconnect from XX:XX:XX:XX:XX:XX
> [CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: no
> Successful disconnected
> [CHG] Device XX:XX:XX:XX:XX:XX Connected: no
> [bluetooth]# connect XX:XX:XX:XX:XX:XX
> Attempting to connect to XX:XX:XX:XX:XX:XX
> [CHG] Device XX:XX:XX:XX:XX:XX Connected: yes
> Connection successful
> [NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0
> [NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0
> [NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
> [NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
> [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
> [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
> [CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: yes
> [xxx]# endpoint.list
> Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0
> Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0
> [xxx]#
>
> Endpoints and transports failed to appear on the first connect to the
> BAP server, since GATT discovery was not yet completed when the BAP code
> tried to discover the endpoints. Second connection succeeded.
> Occasionally, it also happens that endpoints appear but transports do
> not, but this is hard to reproduce.
>
> With this patch:
>
> $ sudo systemctl restart bluetooth
> $ bluetoothctl
> ...
> [bluetooth]# connect XX:XX:XX:XX:XX:XX
> Attempting to connect to XX:XX:XX:XX:XX:XX
> [CHG] Device XX:XX:XX:XX:XX:XX Connected: yes
> [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000184e-0000-1000-8000-00805f9b34fb
> [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001850-0000-1000-8000-00805f9b34fb
> [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 03b80e5a-ede8-4b33-a751-6ce34ec4c700
> Connection successful
> [NEW] Primary Service (Handle 0x0000)
> /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0008
> 00001801-0000-1000-8000-00805f9b34fb
> Generic Attribute Profile
> ...
> [NEW] Descriptor (Handle 0x0000)
> /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0065/char0066/desc0069
> 00002901-0000-1000-8000-00805f9b34fb
> Characteristic User Description
> [CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: yes
> [CHG] Device XX:XX:XX:XX:XX:XX Paired: yes
> [NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0
> [NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0
> [NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
> [NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
> [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
> [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
> [xxx]#
>
> Now the first connection works properly.
>
> On first reading, the spec does not seem to clearly comment if ASEs and
> PACs could be removed/added by the server while client is connected. If
> that's allowed, then we'd need some bigger changes here. Regardless,
> waiting for GATT client ready before first scan seems good also in this
> case.
>
> src/shared/bap.c | 93 +++++++++++++++++++++++++++++++-----------------
> 1 file changed, 60 insertions(+), 33 deletions(-)
>
> diff --git a/src/shared/bap.c b/src/shared/bap.c
> index 22f2e6714..2d676d96f 100644
> --- a/src/shared/bap.c
> +++ b/src/shared/bap.c
> @@ -158,6 +158,7 @@ struct bt_bap {
> struct bt_att *att;
> struct bt_bap_req *req;
> unsigned int cp_id;
> + unsigned int client_ready_id;
>
> unsigned int process_id;
> unsigned int disconn_id;
> @@ -3733,41 +3734,10 @@ static void bap_attach_att(struct bt_bap *bap, struct bt_att *att)
> bap, NULL);
> }
>
> -bool bt_bap_attach(struct bt_bap *bap, struct bt_gatt_client *client)
> +static void bap_attach_finish(struct bt_bap *bap)
> {
> bt_uuid_t uuid;
>
> - if (queue_find(sessions, NULL, bap)) {
> - /* If instance already been set but there is no client proceed
> - * to clone it otherwise considered it already attached.
> - */
> - if (client && !bap->client)
> - goto clone;
> - return true;
> - }
> -
> - if (!sessions)
> - sessions = queue_new();
> -
> - queue_push_tail(sessions, bap);
> -
> - queue_foreach(bap_cbs, bap_attached, bap);
> -
> - if (!client) {
> - bap_attach_att(bap, bap->att);
> - return true;
> - }
> -
> - if (bap->client)
> - return false;
> -
> -clone:
> - bap->client = bt_gatt_client_clone(client);
> - if (!bap->client)
> - return false;
> -
> - bap_attach_att(bap, bt_gatt_client_get_att(client));
> -
> if (bap->rdb->pacs) {
> uint16_t value_handle;
> struct bt_pacs *pacs = bap->rdb->pacs;
> @@ -3798,7 +3768,7 @@ clone:
>
> bap_notify_ready(bap);
>
> - return true;
> + return;
> }
>
> bt_uuid16_create(&uuid, PACS_UUID);
> @@ -3806,6 +3776,57 @@ clone:
>
> bt_uuid16_create(&uuid, ASCS_UUID);
> gatt_db_foreach_service(bap->rdb->db, &uuid, foreach_ascs_service, bap);
> +}
> +
> +static void bap_attach_ready_cb(bool success, uint8_t att_ecode,
> + void *user_data)
> +{
> + struct bt_bap *bap = user_data;
> +
> + bap->client_ready_id = 0;
> +
> + bap_attach_finish(bap);
> +}
> +
> +bool bt_bap_attach(struct bt_bap *bap, struct bt_gatt_client *client)
> +{
> + if (queue_find(sessions, NULL, bap)) {
> + /* If instance already been set but there is no client proceed
> + * to clone it otherwise considered it already attached.
> + */
> + if (client && !bap->client)
> + goto clone;
> + return true;
> + }
> +
> + if (!sessions)
> + sessions = queue_new();
> +
> + queue_push_tail(sessions, bap);
> +
> + queue_foreach(bap_cbs, bap_attached, bap);
> +
> + if (!client) {
> + bap_attach_att(bap, bap->att);
> + return true;
> + }
> +
> + if (bap->client)
> + return false;
> +
> +clone:
> + bap->client = bt_gatt_client_clone(client);
> + if (!bap->client)
> + return false;
> +
> + bap_attach_att(bap, bt_gatt_client_get_att(bap->client));
> +
> + if (bt_gatt_client_is_ready(bap->client)) {
> + bap_attach_finish(bap);
> + } else {
> + bap->client_ready_id = bt_gatt_client_ready_register(
> + bap->client, bap_attach_ready_cb, bap, NULL);
> + }
>
> return true;
> }
> @@ -3824,6 +3845,12 @@ void bt_bap_detach(struct bt_bap *bap)
> if (!queue_remove(sessions, bap))
> return;
>
> + if (bap->client_ready_id) {
> + bt_gatt_client_ready_unregister(bap->client,
> + bap->client_ready_id);
> + bap->client_ready_id = 0;
> + }
> +
> bt_gatt_client_unref(bap->client);
> bap->client = NULL;
>
> --
> 2.39.1
>


--
Luiz Augusto von Dentz

2023-02-12 21:50:29

by Pauli Virtanen

Subject: Re: [PATCH BlueZ] shared/bap: wait for GATT client ready before ASCS/PACS discovery

Hi Luiz,

su, 2023-02-12 kello 12:45 -0800, Luiz Augusto von Dentz kirjoitti:
> Hi Pauli,
>
> On Sun, Feb 12, 2023 at 10:41 AM Pauli Virtanen <[email protected]> wrote:
> >
> > BAP client fails to find endpoints if GATT services are not resolved
> > when the GATT client is attached to the session. This condition
> > almost always occurs on the first BAP connection to a device between
> > bluetoothd restarts, and may occur also otherwise.
>
> Hmm, this shouldn't happen if you have caching enabled, did you
> disable it for some reason?

Caching has not been disabled. The files in /var/lib/bluetooth/*/cache
also get updated so presumably it is working.

> > The BAP client code discovers ASCS/PACS services and characteristics at
> > client attach time.
> >
> > Fix the connection issue by postponing PAC/ASE discovery until the
> > GATT client is ready, if it was not ready at attach time.
>
> This should probably be done by the core, it should check if the
> services are cached then it can call accept, otherwise it needs to
> wait for the client to be ready, perhaps there is something not quite
> right with the way to probe services as afaik if we don't have any
> cache the driver shouldn't have been called to begin with.

bap_accept gets called from

#0 bap_accept (service=0x604000005f50) at profiles/audio/bap.c:1304
#1 0x0000000000631755 in service_accept (service=0x604000005f50, initiator=true) at src/service.c:203
#2 0x0000000000668561 in device_accept_gatt_profiles (device=0x617000000080) at src/device.c:3739
#3 0x0000000000676128 in gatt_client_init (device=0x617000000080) at src/device.c:5161
#4 0x00000000006782e5 in device_attach_att (dev=0x617000000080, io=0x60c00000b2c0) at src/device.c:5331
#5 0x00000000006788e5 in att_connect_cb (io=0x60c00000b2c0, gerr=0x0, user_data=0x617000000080) at src/device.c:5378
#6 0x000000000053de23 in connect_cb (io=0x60c00000b2c0, cond=G_IO_OUT, user_data=0x60300001f8d0) at btio/btio.c:229
#7 0x00007f4955d16cbf in g_main_context_dispatch () from target:/lib64/libglib-2.0.so.0
#8 0x00007f4955d6c598 in g_main_context_iterate.constprop () from target:/lib64/libglib-2.0.so.0
#9 0x00007f4955d1628f in g_main_loop_run () from target:/lib64/libglib-2.0.so.0
#10 0x00000000007df9ca in mainloop_run () at src/shared/mainloop-glib.c:66
#11 0x00000000007e075a in mainloop_run_with_signal (func=0x558597 <signal_callback>, user_data=0x0) at src/shared/mainloop-notify.c:188
#12 0x0000000000558e97 in main (argc=1, argv=0x7ffd399f7788) at src/main.c:1304

which does not wait for GATT client ready. The comment above that

5157│ /*
5158│ * Notify notify existing service about the new connection so they can
5159│ * react to notifications while discovering services
5160│ */

looks like this is intentional. If it's like this, then services would
need to explicitly wait.

Didn't look yet how come it knows the gatt service is there then,
however I guess it only looks for the PACS service uuid, as I think
nothing tells device.c also ASCS would be needed.

Best,
Pauli

>
> > ---
> >
> > Notes:
> > Typical bluetoothctl output in the failing case, without this patch:
> >
> > $ sudo systemctl restart bluetooth
> > $ bluetoothctl
> > ...
> > [bluetooth]# connect XX:XX:XX:XX:XX:XX
> > Attempting to connect to XX:XX:XX:XX:XX:XX
> > [CHG] Device XX:XX:XX:XX:XX:XX Connected: yes
> > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000184e-0000-1000-8000-00805f9b34fb
> > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001850-0000-1000-8000-00805f9b34fb
> > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 03b80e5a-ede8-4b33-a751-6ce34ec4c700
> > Connection successful
> > [NEW] Primary Service (Handle 0x0000)
> > /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0008
> > 00001801-0000-1000-8000-00805f9b34fb
> > Generic Attribute Profile
> > ...
> > [NEW] Descriptor (Handle 0x0000)
> > /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0065/char0066/desc0069
> > 00002901-0000-1000-8000-00805f9b34fb
> > Characteristic User Description
> > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001800-0000-1000-8000-00805f9b34fb
> > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001801-0000-1000-8000-00805f9b34fb
> > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000180a-0000-1000-8000-00805f9b34fb
> > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001844-0000-1000-8000-00805f9b34fb
> > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000184e-0000-1000-8000-00805f9b34fb
> > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001850-0000-1000-8000-00805f9b34fb
> > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 03b80e5a-ede8-4b33-a751-6ce34ec4c700
> > [CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: yes
> > [CHG] Device XX:XX:XX:XX:XX:XX Paired: yes
> > [xxx]# endpoint.list
> > [xxx]# disconnect
> > Attempting to disconnect from XX:XX:XX:XX:XX:XX
> > [CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: no
> > Successful disconnected
> > [CHG] Device XX:XX:XX:XX:XX:XX Connected: no
> > [bluetooth]# connect XX:XX:XX:XX:XX:XX
> > Attempting to connect to XX:XX:XX:XX:XX:XX
> > [CHG] Device XX:XX:XX:XX:XX:XX Connected: yes
> > Connection successful
> > [NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0
> > [NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0
> > [NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
> > [NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
> > [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
> > [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
> > [CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: yes
> > [xxx]# endpoint.list
> > Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0
> > Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0
> > [xxx]#
> >
> > Endpoints and transports failed to appear on the first connect to the
> > BAP server, since GATT discovery was not yet completed when the BAP code
> > tried to discover the endpoints. Second connection succeeded.
> > Occasionally, it also happens that endpoints appear but transports do
> > not, but this is hard to reproduce.
> >
> > With this patch:
> >
> > $ sudo systemctl restart bluetooth
> > $ bluetoothctl
> > ...
> > [bluetooth]# connect XX:XX:XX:XX:XX:XX
> > Attempting to connect to XX:XX:XX:XX:XX:XX
> > [CHG] Device XX:XX:XX:XX:XX:XX Connected: yes
> > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000184e-0000-1000-8000-00805f9b34fb
> > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001850-0000-1000-8000-00805f9b34fb
> > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 03b80e5a-ede8-4b33-a751-6ce34ec4c700
> > Connection successful
> > [NEW] Primary Service (Handle 0x0000)
> > /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0008
> > 00001801-0000-1000-8000-00805f9b34fb
> > Generic Attribute Profile
> > ...
> > [NEW] Descriptor (Handle 0x0000)
> > /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0065/char0066/desc0069
> > 00002901-0000-1000-8000-00805f9b34fb
> > Characteristic User Description
> > [CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: yes
> > [CHG] Device XX:XX:XX:XX:XX:XX Paired: yes
> > [NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0
> > [NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0
> > [NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
> > [NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
> > [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
> > [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
> > [xxx]#
> >
> > Now the first connection works properly.
> >
> > On first reading, the spec does not seem to clearly comment if ASEs and
> > PACs could be removed/added by the server while client is connected. If
> > that's allowed, then we'd need some bigger changes here. Regardless,
> > waiting for GATT client ready before first scan seems good also in this
> > case.
> >
> > src/shared/bap.c | 93 +++++++++++++++++++++++++++++++-----------------
> > 1 file changed, 60 insertions(+), 33 deletions(-)
> >
> > diff --git a/src/shared/bap.c b/src/shared/bap.c
> > index 22f2e6714..2d676d96f 100644
> > --- a/src/shared/bap.c
> > +++ b/src/shared/bap.c
> > @@ -158,6 +158,7 @@ struct bt_bap {
> > struct bt_att *att;
> > struct bt_bap_req *req;
> > unsigned int cp_id;
> > + unsigned int client_ready_id;
> >
> > unsigned int process_id;
> > unsigned int disconn_id;
> > @@ -3733,41 +3734,10 @@ static void bap_attach_att(struct bt_bap *bap, struct bt_att *att)
> > bap, NULL);
> > }
> >
> > -bool bt_bap_attach(struct bt_bap *bap, struct bt_gatt_client *client)
> > +static void bap_attach_finish(struct bt_bap *bap)
> > {
> > bt_uuid_t uuid;
> >
> > - if (queue_find(sessions, NULL, bap)) {
> > - /* If instance already been set but there is no client proceed
> > - * to clone it otherwise considered it already attached.
> > - */
> > - if (client && !bap->client)
> > - goto clone;
> > - return true;
> > - }
> > -
> > - if (!sessions)
> > - sessions = queue_new();
> > -
> > - queue_push_tail(sessions, bap);
> > -
> > - queue_foreach(bap_cbs, bap_attached, bap);
> > -
> > - if (!client) {
> > - bap_attach_att(bap, bap->att);
> > - return true;
> > - }
> > -
> > - if (bap->client)
> > - return false;
> > -
> > -clone:
> > - bap->client = bt_gatt_client_clone(client);
> > - if (!bap->client)
> > - return false;
> > -
> > - bap_attach_att(bap, bt_gatt_client_get_att(client));
> > -
> > if (bap->rdb->pacs) {
> > uint16_t value_handle;
> > struct bt_pacs *pacs = bap->rdb->pacs;
> > @@ -3798,7 +3768,7 @@ clone:
> >
> > bap_notify_ready(bap);
> >
> > - return true;
> > + return;
> > }
> >
> > bt_uuid16_create(&uuid, PACS_UUID);
> > @@ -3806,6 +3776,57 @@ clone:
> >
> > bt_uuid16_create(&uuid, ASCS_UUID);
> > gatt_db_foreach_service(bap->rdb->db, &uuid, foreach_ascs_service, bap);
> > +}
> > +
> > +static void bap_attach_ready_cb(bool success, uint8_t att_ecode,
> > + void *user_data)
> > +{
> > + struct bt_bap *bap = user_data;
> > +
> > + bap->client_ready_id = 0;
> > +
> > + bap_attach_finish(bap);
> > +}
> > +
> > +bool bt_bap_attach(struct bt_bap *bap, struct bt_gatt_client *client)
> > +{
> > + if (queue_find(sessions, NULL, bap)) {
> > + /* If instance already been set but there is no client proceed
> > + * to clone it otherwise considered it already attached.
> > + */
> > + if (client && !bap->client)
> > + goto clone;
> > + return true;
> > + }
> > +
> > + if (!sessions)
> > + sessions = queue_new();
> > +
> > + queue_push_tail(sessions, bap);
> > +
> > + queue_foreach(bap_cbs, bap_attached, bap);
> > +
> > + if (!client) {
> > + bap_attach_att(bap, bap->att);
> > + return true;
> > + }
> > +
> > + if (bap->client)
> > + return false;
> > +
> > +clone:
> > + bap->client = bt_gatt_client_clone(client);
> > + if (!bap->client)
> > + return false;
> > +
> > + bap_attach_att(bap, bt_gatt_client_get_att(bap->client));
> > +
> > + if (bt_gatt_client_is_ready(bap->client)) {
> > + bap_attach_finish(bap);
> > + } else {
> > + bap->client_ready_id = bt_gatt_client_ready_register(
> > + bap->client, bap_attach_ready_cb, bap, NULL);
> > + }
> >
> > return true;
> > }
> > @@ -3824,6 +3845,12 @@ void bt_bap_detach(struct bt_bap *bap)
> > if (!queue_remove(sessions, bap))
> > return;
> >
> > + if (bap->client_ready_id) {
> > + bt_gatt_client_ready_unregister(bap->client,
> > + bap->client_ready_id);
> > + bap->client_ready_id = 0;
> > + }
> > +
> > bt_gatt_client_unref(bap->client);
> > bap->client = NULL;
> >
> > --
> > 2.39.1
> >
>
>

2023-02-13 20:40:35

by Luiz Augusto von Dentz

Subject: Re: [PATCH BlueZ] shared/bap: wait for GATT client ready before ASCS/PACS discovery

Hi Pauli,

On Sun, Feb 12, 2023 at 1:50 PM Pauli Virtanen <[email protected]> wrote:
>
> Hi Luiz,
>
> su, 2023-02-12 kello 12:45 -0800, Luiz Augusto von Dentz kirjoitti:
> > Hi Pauli,
> >
> > On Sun, Feb 12, 2023 at 10:41 AM Pauli Virtanen <[email protected]> wrote:
> > >
> > > BAP client fails to find endpoints if GATT services are not resolved
> > > when the GATT client is attached to the session. This condition
> > > almost always occurs on the first BAP connection to a device between
> > > bluetoothd restarts, and may occur also otherwise.
> >
> > Hmm, this shouldn't happen if you have caching enabled, did you
> > disable it for some reason?
>
> Caching has not been disabled. The files in /var/lib/bluetooth/*/cache
> also get updated so presumably it is working.
>
> > > The BAP client code discovers ASCS/PACS services and characteristics at
> > > client attach time.
> > >
> > > Fix the connection issue by postponing PAC/ASE discovery until the
> > > GATT client is ready, if it was not ready at attach time.
> >
> > This should probably be done by the core, it should check if the
> > services are cached then it can call accept, otherwise it needs to
> > wait for the client to be ready, perhaps there is something not quite
> > right with the way to probe services as afaik if we don't have any
> > cache the driver shouldn't have been called to begin with.
>
> bap_accept gets called from
>
> #0 bap_accept (service=0x604000005f50) at profiles/audio/bap.c:1304
> #1 0x0000000000631755 in service_accept (service=0x604000005f50, initiator=true) at src/service.c:203
> #2 0x0000000000668561 in device_accept_gatt_profiles (device=0x617000000080) at src/device.c:3739
> #3 0x0000000000676128 in gatt_client_init (device=0x617000000080) at src/device.c:5161
> #4 0x00000000006782e5 in device_attach_att (dev=0x617000000080, io=0x60c00000b2c0) at src/device.c:5331
> #5 0x00000000006788e5 in att_connect_cb (io=0x60c00000b2c0, gerr=0x0, user_data=0x617000000080) at src/device.c:5378
> #6 0x000000000053de23 in connect_cb (io=0x60c00000b2c0, cond=G_IO_OUT, user_data=0x60300001f8d0) at btio/btio.c:229
> #7 0x00007f4955d16cbf in g_main_context_dispatch () from target:/lib64/libglib-2.0.so.0
> #8 0x00007f4955d6c598 in g_main_context_iterate.constprop () from target:/lib64/libglib-2.0.so.0
> #9 0x00007f4955d1628f in g_main_loop_run () from target:/lib64/libglib-2.0.so.0
> #10 0x00000000007df9ca in mainloop_run () at src/shared/mainloop-glib.c:66
> #11 0x00000000007e075a in mainloop_run_with_signal (func=0x558597 <signal_callback>, user_data=0x0) at src/shared/mainloop-notify.c:188
> #12 0x0000000000558e97 in main (argc=1, argv=0x7ffd399f7788) at src/main.c:1304
>
> which does not wait for GATT client ready. The comment above that
>
> 5157│ /*
> 5158│ * Notify notify existing service about the new connection so they can
> 5159│ * react to notifications while discovering services
> 5160│ */
>
> looks like this is intentional. If it's like this, then services would
> need to explicitly wait.
>
> Didn't look yet how come it knows the gatt service is there then,
> however I guess it only looks for the PACS service uuid, as I think
> nothing tells device.c also ASCS would be needed.

We do load the database from cache in case there is one:

https://github.com/bluez/bluez/blob/master/src/device.c#L5328

Or are you experiencing some problem where PACS is cached but ASCS isn't?

> Best,
> Pauli
>
> >
> > > ---
> > >
> > > Notes:
> > > Typical bluetoothctl output in the failing case, without this patch:
> > >
> > > $ sudo systemctl restart bluetooth
> > > $ bluetoothctl
> > > ...
> > > [bluetooth]# connect XX:XX:XX:XX:XX:XX
> > > Attempting to connect to XX:XX:XX:XX:XX:XX
> > > [CHG] Device XX:XX:XX:XX:XX:XX Connected: yes
> > > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000184e-0000-1000-8000-00805f9b34fb
> > > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001850-0000-1000-8000-00805f9b34fb
> > > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 03b80e5a-ede8-4b33-a751-6ce34ec4c700
> > > Connection successful
> > > [NEW] Primary Service (Handle 0x0000)
> > > /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0008
> > > 00001801-0000-1000-8000-00805f9b34fb
> > > Generic Attribute Profile
> > > ...
> > > [NEW] Descriptor (Handle 0x0000)
> > > /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0065/char0066/desc0069
> > > 00002901-0000-1000-8000-00805f9b34fb
> > > Characteristic User Description
> > > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001800-0000-1000-8000-00805f9b34fb
> > > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001801-0000-1000-8000-00805f9b34fb
> > > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000180a-0000-1000-8000-00805f9b34fb
> > > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001844-0000-1000-8000-00805f9b34fb
> > > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000184e-0000-1000-8000-00805f9b34fb
> > > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001850-0000-1000-8000-00805f9b34fb
> > > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 03b80e5a-ede8-4b33-a751-6ce34ec4c700
> > > [CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: yes
> > > [CHG] Device XX:XX:XX:XX:XX:XX Paired: yes
> > > [xxx]# endpoint.list
> > > [xxx]# disconnect
> > > Attempting to disconnect from XX:XX:XX:XX:XX:XX
> > > [CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: no
> > > Successful disconnected
> > > [CHG] Device XX:XX:XX:XX:XX:XX Connected: no
> > > [bluetooth]# connect XX:XX:XX:XX:XX:XX
> > > Attempting to connect to XX:XX:XX:XX:XX:XX
> > > [CHG] Device XX:XX:XX:XX:XX:XX Connected: yes
> > > Connection successful
> > > [NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0
> > > [NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0
> > > [NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
> > > [NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
> > > [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
> > > [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
> > > [CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: yes
> > > [xxx]# endpoint.list
> > > Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0
> > > Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0
> > > [xxx]#
> > >
> > > Endpoints and transports failed to appear on the first connect to the
> > > BAP server, since GATT discovery was not yet completed when the BAP code
> > > tried to discover the endpoints. Second connection succeeded.
> > > Occasionally, it also happens that endpoints appear but transports do
> > > not, but this is hard to reproduce.
> > >
> > > With this patch:
> > >
> > > $ sudo systemctl restart bluetooth
> > > $ bluetoothctl
> > > ...
> > > [bluetooth]# connect XX:XX:XX:XX:XX:XX
> > > Attempting to connect to XX:XX:XX:XX:XX:XX
> > > [CHG] Device XX:XX:XX:XX:XX:XX Connected: yes
> > > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 0000184e-0000-1000-8000-00805f9b34fb
> > > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 00001850-0000-1000-8000-00805f9b34fb
> > > [CHG] Device XX:XX:XX:XX:XX:XX UUIDs: 03b80e5a-ede8-4b33-a751-6ce34ec4c700
> > > Connection successful
> > > [NEW] Primary Service (Handle 0x0000)
> > > /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0008
> > > 00001801-0000-1000-8000-00805f9b34fb
> > > Generic Attribute Profile
> > > ...
> > > [NEW] Descriptor (Handle 0x0000)
> > > /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/service0065/char0066/desc0069
> > > 00002901-0000-1000-8000-00805f9b34fb
> > > Characteristic User Description
> > > [CHG] Device XX:XX:XX:XX:XX:XX ServicesResolved: yes
> > > [CHG] Device XX:XX:XX:XX:XX:XX Paired: yes
> > > [NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0
> > > [NEW] Endpoint /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0
> > > [NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
> > > [NEW] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
> > > [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1
> > > [CHG] Transport /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_sink0/fd1 Links: /org/bluez/hci0/dev_XX_XX_XX_XX_XX_XX/pac_source0/fd0
> > > [xxx]#
> > >
> > > Now the first connection works properly.
> > >
> > > On first reading, the spec does not seem to clearly comment if ASEs and
> > > PACs could be removed/added by the server while client is connected. If
> > > that's allowed, then we'd need some bigger changes here. Regardless,
> > > waiting for GATT client ready before first scan seems good also in this
> > > case.
> > >
> > > src/shared/bap.c | 93 +++++++++++++++++++++++++++++++-----------------
> > > 1 file changed, 60 insertions(+), 33 deletions(-)
> > >
> > > diff --git a/src/shared/bap.c b/src/shared/bap.c
> > > index 22f2e6714..2d676d96f 100644
> > > --- a/src/shared/bap.c
> > > +++ b/src/shared/bap.c
> > > @@ -158,6 +158,7 @@ struct bt_bap {
> > > struct bt_att *att;
> > > struct bt_bap_req *req;
> > > unsigned int cp_id;
> > > + unsigned int client_ready_id;
> > >
> > > unsigned int process_id;
> > > unsigned int disconn_id;
> > > @@ -3733,41 +3734,10 @@ static void bap_attach_att(struct bt_bap *bap, struct bt_att *att)
> > > bap, NULL);
> > > }
> > >
> > > -bool bt_bap_attach(struct bt_bap *bap, struct bt_gatt_client *client)
> > > +static void bap_attach_finish(struct bt_bap *bap)
> > > {
> > > bt_uuid_t uuid;
> > >
> > > - if (queue_find(sessions, NULL, bap)) {
> > > - /* If instance already been set but there is no client proceed
> > > - * to clone it otherwise considered it already attached.
> > > - */
> > > - if (client && !bap->client)
> > > - goto clone;
> > > - return true;
> > > - }
> > > -
> > > - if (!sessions)
> > > - sessions = queue_new();
> > > -
> > > - queue_push_tail(sessions, bap);
> > > -
> > > - queue_foreach(bap_cbs, bap_attached, bap);
> > > -
> > > - if (!client) {
> > > - bap_attach_att(bap, bap->att);
> > > - return true;
> > > - }
> > > -
> > > - if (bap->client)
> > > - return false;
> > > -
> > > -clone:
> > > - bap->client = bt_gatt_client_clone(client);
> > > - if (!bap->client)
> > > - return false;
> > > -
> > > - bap_attach_att(bap, bt_gatt_client_get_att(client));
> > > -
> > > if (bap->rdb->pacs) {
> > > uint16_t value_handle;
> > > struct bt_pacs *pacs = bap->rdb->pacs;
> > > @@ -3798,7 +3768,7 @@ clone:
> > >
> > > bap_notify_ready(bap);
> > >
> > > - return true;
> > > + return;
> > > }
> > >
> > > bt_uuid16_create(&uuid, PACS_UUID);
> > > @@ -3806,6 +3776,57 @@ clone:
> > >
> > > bt_uuid16_create(&uuid, ASCS_UUID);
> > > gatt_db_foreach_service(bap->rdb->db, &uuid, foreach_ascs_service, bap);
> > > +}
> > > +
> > > +static void bap_attach_ready_cb(bool success, uint8_t att_ecode,
> > > + void *user_data)
> > > +{
> > > + struct bt_bap *bap = user_data;
> > > +
> > > + bap->client_ready_id = 0;
> > > +
> > > + bap_attach_finish(bap);
> > > +}
> > > +
> > > +bool bt_bap_attach(struct bt_bap *bap, struct bt_gatt_client *client)
> > > +{
> > > + if (queue_find(sessions, NULL, bap)) {
> > > + /* If instance already been set but there is no client proceed
> > > + * to clone it otherwise considered it already attached.
> > > + */
> > > + if (client && !bap->client)
> > > + goto clone;
> > > + return true;
> > > + }
> > > +
> > > + if (!sessions)
> > > + sessions = queue_new();
> > > +
> > > + queue_push_tail(sessions, bap);
> > > +
> > > + queue_foreach(bap_cbs, bap_attached, bap);
> > > +
> > > + if (!client) {
> > > + bap_attach_att(bap, bap->att);
> > > + return true;
> > > + }
> > > +
> > > + if (bap->client)
> > > + return false;
> > > +
> > > +clone:
> > > + bap->client = bt_gatt_client_clone(client);
> > > + if (!bap->client)
> > > + return false;
> > > +
> > > + bap_attach_att(bap, bt_gatt_client_get_att(bap->client));
> > > +
> > > + if (bt_gatt_client_is_ready(bap->client)) {
> > > + bap_attach_finish(bap);
> > > + } else {
> > > + bap->client_ready_id = bt_gatt_client_ready_register(
> > > + bap->client, bap_attach_ready_cb, bap, NULL);
> > > + }
> > >
> > > return true;
> > > }
> > > @@ -3824,6 +3845,12 @@ void bt_bap_detach(struct bt_bap *bap)
> > > if (!queue_remove(sessions, bap))
> > > return;
> > >
> > > + if (bap->client_ready_id) {
> > > + bt_gatt_client_ready_unregister(bap->client,
> > > + bap->client_ready_id);
> > > + bap->client_ready_id = 0;
> > > + }
> > > +
> > > bt_gatt_client_unref(bap->client);
> > > bap->client = NULL;
> > >
> > > --
> > > 2.39.1
> > >
> >
> >
>


--
Luiz Augusto von Dentz