Re: [alsa-devel] Sample format coversion bug?

Takashi Iwai wrote:

Well, apparently something wrong with mmap, then.

Do you mean "something is wrong with the kernel driver" or "plug really expects mmap access to the card to be available"?

Does the device really support 24bit format?

No, it doesn't - that's the whole point of the test. I expected "plug" to guess that the device supports a 16-bit format, record 16-bit samples from the hardware, multiply all recorded samples by 256, and return the result as 32-bit words (so that the result is a set of valid S24_LE samples, even though the least significant byte is always zero).

S24_LE isn't 3bytes format but packed into the lower 3 bytes in 4 byte frames.

Yes, I know.