On Thursday, October 14, 2010 2:58 PM, Ryan Mallon wrote:
On 10/15/2010 10:46 AM, H Hartley Sweeten wrote:
On Thursday, October 14, 2010 7:49 AM, Mika Westerberg wrote:
+void __init ep93xx_register_ac97(void) +{
- /*
* Make sure that the AC97 pins are not used by I2S.*/- ep93xx_devcfg_clear_bits(EP93XX_SYSCON_DEVCFG_I2SONAC97);
This is fine for now.
Eventually some thought should go into a proper iomux to prevent loading drivers with conflicting pin requirements. Currently, depending on the configuration, this could break if a user should try to use the ac97 and i2s audio at the same time (I2SONAC97 mode). But, the same problem already exists if a user tries using spi and i2s audio (I2SONSSP mode).
I initially suggested this also, since we have acquire/release functions for some other drivers, but I now think this is fine as is. It's highly unlikely that a board is going to have both I2S and AC97 audio. We can change this later if such a board shows up, no need to make the code more complicated for hypothetical boards :-).
Agree. That's why I said eventually.... ;-)
The only platforms I know of that have both audio systems are the EDB9307, EDB9312, and EDB9315 boards from Cirrus. I don't have any of these so, unless someone steps up with one to test, Mika's implementation is fine.
Regards, Hartley