On Thu, 2011-04-21 at 14:19 -0400, Ben Gardiner wrote:
This patch series is comprised of three bugfixes and one cleanup that were performed during prototyping of McASP operation in codec clock- master frame-slave mode.
First we noticed that the check of the number of tdm slots requested by platform data was always returning true -- unrelated to CMB/CFS mode.
Then a cleanup: the PDIR values set are currently based on magic numbers and there are available bitfield definitions. Not strictly needed but it makes the changes introduced in the last patch simpler to read.
It was found that the hardware parameters assigned when codec clock-master frame-slave is requested were incorrect. This change is required for correct CBM/CFS operation.
Finally, the direction of the pins is corrected to reflect the implications of codec clock-master frame-slave -- i.e. mcasp clock- input frame-output. This change is also required for correct operation.
The combination was tested with a logic analyzer and the hrtimer pwm device from Bill Gatliff's PWM framework [1] on a da850evm with hardware modifications to access the McASP lines.
[1] http://article.gmane.org/gmane.linux.kernel.embedded/3486/
Ben Gardiner (4): davinci-mcasp: correct tdm_slots limit davinci-mcasp: use bitfield definitions for PDIR davinci-mcasp: fix _CBM_CFS hw_params davinci-mcasp: fix _CBM_CFS pin directions
sound/soc/davinci/davinci-mcasp.c | 19 ++++++++++++------- 1 files changed, 12 insertions(+), 7 deletions(-)
All
Acked-by: Liam Girdwood lrg@ti.com