10 May
2012
10 May
'12
8:43 p.m.
Russell King - ARM Linux wrote:
On Thu, May 10, 2012 at 07:13:59PM +0200, Clemens Ladisch wrote:
I wrote:
This (untested) patch tries to avoid too many busy looping.
Oops, off-by-one error (avail == avail_min should not wait).
This is better - it now sits around 2-14% system time. I guess the five loops is rather arbitary?
Yes. The minimum loop limit is two (to handle wraparound at the end of the buffer); more than that probably does not actually make sense.
Jassi Brar wrote:
If we really want to change the behavior of play, wouldn't we want to do that for capture(snd_pcm_lib_read1) as well ?
Indeed. And we'd want to reuse the twake calculation.
Regards, Clemens