2012-04-27 20:53:58

by Mike Brudevold

[permalink] [raw]
Subject: TP/SIG/SMG/BI-02-C failure

Hi,

I am experiencing a failure on TP/SIG/SMG/BI-02-C "To verify that the
IUT (ACP) is able to discard a corrupted stream discover command
received by the INT." The test claims that "No message is sent to the
upper layers and no response is generated." and the problem I have is
that the disconnect_timeout function is sending its own DISCOVER_CMD
after the timeout. This apparently is invalid.

For those who are more familiar with the position of
"remove_disconnect_timer(session);" in session_cb, does it make sense
to move that outside the AVDTP_MSG_TYPE_COMMAND block so that even the
invalid discover command (general reject) turns off the timer?

Mike


2012-04-29 19:36:38

by Mike Brudevold

[permalink] [raw]
Subject: Re: TP/SIG/SMG/BI-02-C failure

Hi Johan,

>> > For those who are more familiar with the position of
>> > "remove_disconnect_timer(session);" in session_cb, does it make sense
>> > to move that outside the AVDTP_MSG_TYPE_COMMAND block so that even the
>> > invalid discover command (general reject) turns off the timer?
>>
>> The timer is only removed if there are no streams present
>
> Sorry, that should say "if there *are* streams present":
>
> ? ? ? ? ? ? ? ?if (session->streams && session->dc_timer)
> ? ? ? ? ? ? ? ? ? ? ? ?remove_disconnect_timer(session);

The BQE has accepted it as is, now that he's had enough time to look at it.

Thanks,
Mike

2012-04-28 10:13:11

by Johan Hedberg

[permalink] [raw]
Subject: Re: TP/SIG/SMG/BI-02-C failure

Hi,

On Sat, Apr 28, 2012, Johan Hedberg wrote:
> > For those who are more familiar with the position of
> > "remove_disconnect_timer(session);" in session_cb, does it make sense
> > to move that outside the AVDTP_MSG_TYPE_COMMAND block so that even the
> > invalid discover command (general reject) turns off the timer?
>
> The timer is only removed if there are no streams present

Sorry, that should say "if there *are* streams present":

if (session->streams && session->dc_timer)
remove_disconnect_timer(session);

Johan

2012-04-28 09:56:17

by Johan Hedberg

[permalink] [raw]
Subject: Re: TP/SIG/SMG/BI-02-C failure

Hi Mike,

On Fri, Apr 27, 2012, Mike wrote:
> I am experiencing a failure on TP/SIG/SMG/BI-02-C "To verify that the
> IUT (ACP) is able to discard a corrupted stream discover command
> received by the INT." The test claims that "No message is sent to the
> upper layers and no response is generated." and the problem I have is
> that the disconnect_timeout function is sending its own DISCOVER_CMD
> after the timeout. This apparently is invalid.

The test spec talks about a response, not about a new command. So from
that perspective I'd say we're still compliant and that the test vector
itself might be wrongly define (not looking at the message type sent by
us).

> For those who are more familiar with the position of
> "remove_disconnect_timer(session);" in session_cb, does it make sense
> to move that outside the AVDTP_MSG_TYPE_COMMAND block so that even the
> invalid discover command (general reject) turns off the timer?

The timer is only removed if there are no streams present, so I don't
see how moving it outside of the if branch would help. However as I said
this is still looking like a test vector issue to me.

Johan