Hi Marcel,

On 12/08/2011 03:13 PM, Marcel Holtmann wrote:
> Hi Philippe,
>
>>   gatchat/gatchat.c |   26 ++++++++++++++++++++++++++
>>   gatchat/gatchat.h |    4 ++++
>>   2 files changed, 30 insertions(+), 0 deletions(-)
>>
>> diff --git a/gatchat/gatchat.c b/gatchat/gatchat.c
>> index 7a0ef35..00e5fa8 100644
>> --- a/gatchat/gatchat.c
>> +++ b/gatchat/gatchat.c
>> @@ -35,6 +35,7 @@
>>   #include "ringbuffer.h"
>>   #include "gatchat.h"
>>   #include "gatio.h"
>> +#include "gathdlc.h"
>>
>>   /* #define WRITE_SCHEDULER_DEBUG 1 */
>>
>> @@ -110,6 +111,7 @@ struct _GAtChat {
>>   	struct at_chat *parent;
>>   	guint group;
>>   	GAtChat *slave;
>> +	GAtHDLC	*diag_monitor;
>>   };
>
> what is this for? I don't see this needed at all. You are not crossing
> atom driver implementations anyway.

Here, the idea is to link one AT capable port to one Diagnostic port.
Commonly, we are specifying only one AT capable channel (GAtChat) when 
creating an atom.
For cdma_netreg, I need both (AT and QCDM). Indeed, with some EV-DO 
capable hardware, the result of some AT commands (like AT+CSS?) may be 
wrong. That's why, I introduced the QCDM support. To retrieve the 
GAtHDLC from the driver implementation, I chose to use a solution 
similar to 'g_at_chat_get_slave'.

Regards,

Philippe.