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

[Koul, Vinod]

> 
> > +	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.
Will fix this

> > +
> > +	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...
No as the DSP FW returns the frame value for the pointer

> 
> > +	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
Will fix

> 
> > +MODULE_LICENSE("GPL v2");
> > +MODULE_ALIAS("mid-platform");
> 
> Same issue as with the CODEC driver here.
Okay

> > +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?
These function are implemented in platform_lib.c and called in mid_platform.c
 
> > +enum mid_drv_status {
> > +	INIT = 1,
> > +	STARTED,
> > +	RUNNING,
> > +	PAUSED,
> > +	DROPPED,
> > +};
> 
> Namespacing.
Will fix

~Vinod