On Tue, Feb 3, 2009 at 6:35 PM, Mark Brown <broonie@xxxxxxxxxxxxx> wrote:
> On Tue, Feb 03, 2009 at 06:13:22PM +0100, pHilipp Zabel wrote:
>
>> - priv->dev.ssp = ssp_request(dai->id, "SoC audio");
>> + priv->dev.ssp = ssp_request(dai->id + 1, "SoC audio");
>> if (priv->dev.ssp == NULL) {
>> ret = -ENODEV;
>> goto err_priv;
>
>> But this isn't right either. Because if we request the correct SSP
>> port here, the subsequent request of the same SSP port in ssp_init
>> (called from ops->startup) fails.
>> How to fix?
>
> Hrm. ssp_init() does a ssp_request() already. The request on probe is
> just redundant. Eric was muttering about refactoring the API some more
> but never got round to it.
Hm, hm.. Ripping ssp_init/exit's innards out of them (minus the
ssp/irq_request parts), like this, seems to work:
--------------------------- sound/soc/pxa/pxa-ssp.c ---------------------------
index 1687e18..c56a07c 100644
@@ -21,6 +21,8 @@
#include <linux/clk.h>
#include <linux/io.h>
+#include <asm/irq.h>
+
#include <sound/core.h>
#include <sound/pcm.h>
#include <sound/initval.h>
@@ -221,9 +223,9 @@ static int pxa_ssp_startup(struct
snd_pcm_substream *substream,
int ret = 0;
if (!cpu_dai->active) {
- ret = ssp_init(&priv->dev, cpu_dai->id + 1, SSP_NO_IRQ);
- if (ret < 0)
- return ret;
+ priv->dev.port = cpu_dai->id + 1;
+ priv->dev.irq = NO_IRQ;
+ clk_enable(priv->dev.ssp->clk);
ssp_disable(&priv->dev);
}
return ret;
@@ -238,7 +240,7 @@ static void pxa_ssp_shutdown(struct
snd_pcm_substream *substream,
if (!cpu_dai->active) {
ssp_disable(&priv->dev);
- ssp_exit(&priv->dev);
+ clk_disable(priv->dev.ssp->clk);
}
}
_______________________________________________
Alsa-devel mailing list
Alsa-devel@xxxxxxxxxxxxxxxx
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
