On Wed, 24 Oct 2018 09:44:01 +0200, Timo Wischer wrote:
On 10/24/18 09:06, Takashi Iwai wrote:
On Mon, 22 Oct 2018 09:19:28 +0200, Timo Wischer wrote:
On 10/18/18 20:57, Takashi Iwai wrote:
But how can it be at the first place? (352 353) is already empty as the frames. The time could be kept in this representation, but the frames must be integer.
Which order of calls did it result in so?
We know that some order of calls make the selection impossible like the above, especially when both time and bytes/frames are mixed.
I have used the following ALSA configuration:
pcm.test_rate { type rate slave.pcm hw:gmdcard slave.rate 48000 }
And which driver is gmdcard?
Takashi
The driver is called Generic Machine Driver (unfortunately we have it not yet upstreamed) and it is using the Renesas RCar platform driver (https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/soun...)
But it is also reproducible with hw:Loopback with a slightly different rules negotiation but the same result at the end:
If it can be reproduced with the loopback driver, it makes easier to debug. But you modified something? Then it has to be clarified at first; i.e. let others reproduce your problem. Otherwise we have no merit to debug it in our side :)
thanks,
Takashi