android/handsfree.c | Possible Fix | Review

android/handsfree.c | Possible Fix | Review

[Prabhu Chawandi]

Hi,

  When we were testing handsfree we were seeing command is not being
identified and when looked size was  sent zero when we set it as below
it started working.

static const struct ipc_handler sco_handlers[] = {
        /* SCO_OP_GET_FD */
-        { bt_sco_get_fd, false, 0 }
};


static const struct ipc_handler sco_handlers[] = {
        /* SCO_OP_GET_FD */
 +       { bt_sco_get_fd, false, sizeof(struct sco_cmd_get_fd) }
};

one request -  have you documented anywhere the sequence of commands
to test each profiles supported till now. if we get the the proper
sequences we can validate much faster stabilising the code. Else we
need to spend time in identifying the sequences.

So, if you have documented anywhere request you to share,

ManyThanks,
Prabhu

Re: android/handsfree.c | Possible Fix | Review

[Szymon Janc]

Hi Prabhu,

On Wednesday 01 of April 2015 18:40:52 Prabhu Chawandi wrote:
> Hi,
> 
>   When we were testing handsfree we were seeing command is not being
> identified and when looked size was  sent zero when we set it as below
> it started working.
> 
> static const struct ipc_handler sco_handlers[] = {
>         /* SCO_OP_GET_FD */
> -        { bt_sco_get_fd, false, 0 }
> };
> 
> 
> static const struct ipc_handler sco_handlers[] = {
>         /* SCO_OP_GET_FD */
>  +       { bt_sco_get_fd, false, sizeof(struct sco_cmd_get_fd) }
> };

Yes, the fix is correct. Could you send a proper patch for it?
 
> one request -  have you documented anywhere the sequence of commands
> to test each profiles supported till now. if we get the the proper
> sequences we can validate much faster stabilising the code. Else we
> need to spend time in identifying the sequences.
> 
> So, if you have documented anywhere request you to share,

Those sequences are defined by android code (framework or multimedia) and are 
not really documented. You could find some suggestion if you look at our PTS 
testing documentation (android/pts-*.txt) where we do use haltest for tests. 

Note that if possible we prefer to use Android Framework for validation 
instead of haltest (which is mostly development tool).

-- 
BR
Szymon Janc

Re: android/handsfree.c | Possible Fix | Review

[Prabhu Chawandi]

Hi Szymon,

  Sure, I will send the patch soon, I have to pull 5.30 as I am
currently on 5.28.

with sco read, I was not able to initiate read at all, when i debug it
is always return at the same point where I had pointed a crash couple
of days ago in hal-sco,c

I get channel number allocated but I am not able to understand why I
am getting frame size as zero.. please through some light on it.

        if (!frame_size)
                return -1;  ==>

        frame_num = bytes / frame_size;
        input_frame_num = frame_num;


 Actually, we are trying on linux host, we are not using the Android framework.
May be if you have some free time (Not sure :(), please capture the
sequences, one profile at time. sorry if I am asking fore more !

manyThanks,
Prabhu



On Wed, Apr 1, 2015 at 7:41 PM, Szymon Janc <szymon.janc@tieto.com> wrote:
> Hi Prabhu,
>
> On Wednesday 01 of April 2015 18:40:52 Prabhu Chawandi wrote:
>> Hi,
>>
>>   When we were testing handsfree we were seeing command is not being
>> identified and when looked size was  sent zero when we set it as below
>> it started working.
>>
>> static const struct ipc_handler sco_handlers[] = {
>>         /* SCO_OP_GET_FD */
>> -        { bt_sco_get_fd, false, 0 }
>> };
>>
>>
>> static const struct ipc_handler sco_handlers[] = {
>>         /* SCO_OP_GET_FD */
>>  +       { bt_sco_get_fd, false, sizeof(struct sco_cmd_get_fd) }
>> };
>
> Yes, the fix is correct. Could you send a proper patch for it?
>
>> one request -  have you documented anywhere the sequence of commands
>> to test each profiles supported till now. if we get the the proper
>> sequences we can validate much faster stabilising the code. Else we
>> need to spend time in identifying the sequences.
>>
>> So, if you have documented anywhere request you to share,
>
> Those sequences are defined by android code (framework or multimedia) and are
> not really documented. You could find some suggestion if you look at our PTS
> testing documentation (android/pts-*.txt) where we do use haltest for tests.
>
> Note that if possible we prefer to use Android Framework for validation
> instead of haltest (which is mostly development tool).
>
> --
> BR
> Szymon Janc