On Thu, Jan 10, 2019 at 03:06:10PM +0000, Srinivas Kandagatla wrote:
This all looks good apart from a couple of small things that should be
easy to fix:
> diff --git a/sound/soc/codecs/Kconfig b/sound/soc/codecs/Kconfig
> index 62bdb7e333b8..1c4904940621 100644
> --- a/sound/soc/codecs/Kconfig
> +++ b/sound/soc/codecs/Kconfig
> @@ -1100,6 +1100,12 @@ config SND_SOC_UDA1380
> tristate
> depends on I2C
>
> +config SND_SOC_WCD9335
> + tristate "WCD9335 Codec"
> + depends on SLIMBUS
> + select REGMAP_SLIMBUS
> + tristate
> +
> config SND_SOC_WL1273
> tristate
>
You should add this to SND_SOC_ALL_CODECS.
> +static irqreturn_t wcd9335_slimbus_irq(int irq, void *data)
> +{
> + struct wcd9335_codec *wcd = data;
> + unsigned long status = 0;
> + int i, j, port_id;
> + unsigned int val, int_val = 0;
> + bool tx;
> + unsigned short reg = 0;
> +
> + for (i = WCD9335_SLIM_PGD_PORT_INT_STATUS_RX_0, j = 0;
> + i <= WCD9335_SLIM_PGD_PORT_INT_STATUS_TX_1; i++, j++) {
> + regmap_read(wcd->if_regmap, i, &val);
> + status |= ((u32)val << (8 * j));
> + }
> +
> + for_each_set_bit(j, &status, 32) {
> + tx = (j >= 16 ? true : false);
> + port_id = (tx ? j - 16 : j);
...
> + return IRQ_HANDLED;
> +}
This will report that it handled the interrupt unconditionally, this
means that if the interrupt fires in error or there's some bug the
interrupt will never be acknowledged (it looks like the chip requires
acks?) but the interrupt core won't be able to notice and handle the
problem as effectively as it could've done. It's better to report
IRQ_NONE if nothing as handled so that the core can do error handling.
Attachment:
signature.asc
Description: PGP signature
