Hi Luiz,
Sorry for the delay, will send the new patch soon.
On 9/20/2024 3:53 AM, Luiz Augusto von Dentz wrote:
> Hi Cheng,
>
> On Mon, Aug 26, 2024 at 10:30 AM Cheng Jiang <quic_chejiang@xxxxxxxxxxx> wrote:
>>
>> Hi Luiz,
>>
>> Thank you for your feedback. Please check the comment inline.
>>
>> On 8/26/2024 9:43 PM, Luiz Augusto von Dentz wrote:
>>> Hi Cheng,
>>>
>>> On Mon, Aug 26, 2024 at 5:01 AM Cheng Jiang <quic_chejiang@xxxxxxxxxxx> wrote:
>>>>
>>>> If the remote device drops the connection before DUT confirm the
>>>> service authorization, the DUT still must wait for service
>>>> authorization timeout before processing future request.
>>>>
>>>> Cancel the service authorization request when connection is dropped.
>>>> ---
>>>> src/adapter.c | 20 ++++++++++++++++++++
>>>> 1 file changed, 20 insertions(+)
>>>>
>>>> diff --git a/src/adapter.c b/src/adapter.c
>>>> index 245de4456..3ad000425 100644
>>>> --- a/src/adapter.c
>>>> +++ b/src/adapter.c
>>>> @@ -8502,6 +8502,25 @@ static void disconnect_notify(struct btd_device *dev, uint8_t reason)
>>>> }
>>>> }
>>>>
>>>> +static void adapter_cancel_service_auth(struct btd_adapter *adapter,
>>>> + struct btd_device *device)
>>>> +{
>>>> + struct service_auth *auth;
>>>> + GList *l;
>>>> +
>>>> + auth = NULL;
>>>> + for (l = adapter->auths->head; l != NULL; l = l->next) {
>>>> + auth = l->data;
>>>> + if (auth->device == device)
>>>> + break;
>>>> + }
>>>> + if (auth != NULL) {
>>>> + DBG("Cancel pending auth: %s", auth->uuid);
>>>> + g_queue_remove(auth->adapter->auths, auth);
>>>> + service_auth_cancel(auth);
>>>> + }
>>>> +}
>>>> +
>>>> static void dev_disconnected(struct btd_adapter *adapter,
>>>> const struct mgmt_addr_info *addr,
>>>> uint8_t reason)
>>>> @@ -8516,6 +8535,7 @@ static void dev_disconnected(struct btd_adapter *adapter,
>>>> device = btd_adapter_find_device(adapter, &addr->bdaddr, addr->type);
>>>> if (device) {
>>>> adapter_remove_connection(adapter, device, addr->type);
>>>> + adapter_cancel_service_auth(adapter, device);
>>>
>>> This shall probably be placed together with
>>> device_cancel_authentication in adapter_remove_connection but we need
>>> to track in what bearer the authorization is for and remove all
>>> authorization requests like in btd_adapter_remove_device:
>> Yes. It can be put in device_cancel_authentication, but the condition to call
>> this function in adapter_remove_connection need change. device_is_authenticating
>> is only true when request, notify or confirm pincode/passkey.
>> As for bearer, the service authorize is only happened on BDADDR_BREDR. So it
>> can be called when the address type is BDADDR_BREDR in dev_disconnected.
>>>
>>> l = adapter->auths->head;
>>> while (l != NULL) {
>>> struct service_auth *auth = l->data;
>>> GList *next = g_list_next(l);
>>>
>>> if (auth->device != dev) {
>>> l = next;
>>> continue;
>>> }
>>>
>>> g_queue_delete_link(adapter->auths, l);
>>> l = next;
>>>
>>> service_auth_cancel(auth);
>>> }
>>>
>>> I'd probably move the above code to a helper function so it can be
>>> called from different places, as for doing this on a per bearer basis
>>> would probably need to add bearer information to btd_service but I
>>> guess that can be treated separately.
>> Yes, it's great. Need use the above code to cancel all pending authorize.
>
> Looks like you never send another version of this one, are you still
> planning on having this fixed?
>
>>>
>>>
>>>> disconnect_notify(device, reason);
>>>> }
>>>>
>>>> --
>>>> 2.25.1
>>>>
>>>>
>>>
>>>
>>
>
>
