[alsa-devel] [PATCH 2/4] ASoC sst: Add mid platform driver

[Mark Brown]

On Thu, Dec 30, 2010 at 04:42:53PM +0530, Vinod Koul wrote:

> +	case SNDRV_PCM_TRIGGER_START:
> +		pr_debug("sst: Trigger Start\n");
> +		ret_val = stream->sstdrv_ops->control_set(
> +						SST_SND_START, &str_id);
> +		if (ret_val)
> +			return ret_val;
> +		stream->stream_status = RUNNING;
> +		stream->stream_info.mad_substream = substream;
> +		break;

You've no locking in the rest of the driver protecting the variables
you're updating here.

> +
> +	stream = substream->runtime->private_data;
> +	if (stream->stream_status == INIT)
> +		return 0;
> +	str_info = &stream->stream_info;
> +	ret_val = stream->sstdrv_ops->control_set(
> +				SST_SND_BUFFER_POINTER, &str_info);
> +	if (ret_val) {
> +		pr_err("sst: error code = 0x%x\n", ret_val);
> +		return ret_val;
> +	}
> +
> +	return stream->stream_info.buffer_ptr;

I suspect you need to return bytes_to_frames() of this...

> +	pr_debug("intelmid_platform_probe called\n");
> +	ret = snd_soc_register_platform(&pdev->dev, &intelmid_soc_platform_drv);
> +	if (ret) {
> +		pr_err("registering soc platform failed\n");
> +		return ret;
> +	}
> +	return snd_soc_register_dais(&pdev->dev,
> +				mfld_dai, ARRAY_SIZE(mfld_dai));

You're not unwinding the platform register if the DAI registration
fail.s

> +MODULE_LICENSE("GPL v2");
> +MODULE_ALIAS("mid-platform");

Same issue as with the CODEC driver here.

> +int snd_intelmad_alloc_stream(struct snd_pcm_substream *substream);
> +int snd_intelmad_init_stream(struct snd_pcm_substream *substream);

Why are these exposed here?

> +enum mid_drv_status {
> +	INIT = 1,
> +	STARTED,
> +	RUNNING,
> +	PAUSED,
> +	DROPPED,
> +};

Namespacing.