On Tue, Apr 28, 2009 at 08:08:06AM +0200, Takashi Iwai wrote:
Mark Brown wrote:
On Mon, Apr 27, 2009 at 05:09:27PM +0200, Jaroslav Kysela wrote:
Appearently, FIFO is too big. I'm working on some changes to take account
Could you expand a bit on what qualifies as "too big" here? I've not been following the thread here or the changes that caused problems.
It's bigger than the pre-defined tolerance of jiffies check in PCM core code.
Hrm, OK. That may create problems for some embedded systems which can use very large FIFOs (well, FIFO equivalents) to allow the processor to go to sleep when doing things like MP3 playback. I don't think any of those have audio support in mainline yet, though.
The new PCM core checks the returned hw_ptr together with jiffies delta whether it's a sane value. When FIFO is used with the ring buffer, the hardware tends to report back earlier before the expected timing.
But (from other messages in the thread) it's possible to disable these checks if required?