I'm sorry but I correct what I said:
Actually the target device seems to update the max payload size after host connects to CMP output plug because its initial value is zero. Then we already allocate isochronous resources and isochronous packet to host system. I don't think it better to reallocate them after the connecting.
We don't allocate buffer for isochronous packet yet when connecting CMP output plug. Then we just gain isochronous resource.
Anyway I don't think it better to reallocate isochronous resource.
Regards
Takashi Sakamoto
(Apr 29 2013 18:25), Takashi Sakamoto wrote:
Clemens,
The specification was written for the general case where the device that creates the connection might be different from the transmitting and receiving devices. In that case, the payload size is needed for correct bandwidth allocation.
I'm sure that the general case.
In our case, the Linux PC does not have plugs that could be accessed by anybody else.
I'm sure that host system (Linux PC) doesn't have no plugs.
(And I'm not sure if the value set by the Fireworks firmware is correct enough so that we'd want to use it.)
With my Fireworks device, it always report 0, it means 1024 bytes for max payload size for output plug. But I'm not sure that it always report the same value. Especially at 192.0 kHz, the payload size is over 1024 bytes but my device supports up to 96.0 kHz...
Actually the target device seems to update the max payload size after host connects to CMP output plug because its initial value is zero. Then we already allocate isochronous resources and isochronous packet to host system. I don't think it better to reallocate them after the connecting.
It's reasonable to ignore the payload field in any processing.
Regards
Takashi Sakamoto
(Apr 29 2013 15:30), Clemens Ladisch wrote:
Takashi Sakamoto wrote:
Then I have a question about handling payload field. In specification, "The payload field shall specify the maximum number of quadlets that may be transmitted in a single isochronous packet for this plug" but actually cheking this field seems to make no sense because there is no checking process in the procedure. I think I can do nothing for payload field.
The specification was written for the general case where the device that creates the connection might be different from the transmitting and receiving devices. In that case, the payload size is needed for correct bandwidth allocation.
In our case, the Linux PC does not have plugs that could be accessed by anybody else. (And I'm not sure if the value set by the Fireworks firmware is correct enough so that we'd want to use it.)
Regards, Clemens