Re: [RFC 1/5] soundwire: bus_type: add sdw_master_device support

[Vinod Koul <vkoul@kernel.org>]

Hi Greg,

On 23-04-20, 16:24, Greg KH wrote:
> On Mon, Apr 20, 2020 at 12:56:31PM +0530, Vinod Koul wrote:
> > Hello Bard,
> > 
> > On 17-04-20, 04:55, Bard Liao wrote:
> > > From: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
> > > 
> > > In the existing SoundWire code, Master Devices are not explicitly
> > > represented - only SoundWire Slave Devices are exposed (the use of
> > > capital letters follows the SoundWire specification conventions).
> > > 
> > > The SoundWire Master Device provides the clock, synchronization
> > > information and command/control channels. When multiple links are
> > > supported, a Controller may expose more than one Master Device; they
> > > are typically embedded inside a larger audio cluster (be it in an
> > > SOC/chipset or an external audio codec), and we need to describe it
> > > using the Linux device and driver model.  This will allow for
> > > configuration functions to account for external dependencies such as
> > > power rails, clock sources or wake-up mechanisms. This transition will
> > > also allow for better sysfs support without the reference count issues
> > > mentioned in the initial reviews.
> > 
> > Well the primary reason for doing sdw_master_device for creating a
> > adding sysfs representation.
> 
> -ENOPARSE :(

Oops, sorry!

> > It *also* helps some vendors due to
> > inherent model should not be constructed as the primary approach for the
> > sdw_master_device.
> 
> No, the PRIMARY reason is "it is the correct thing to do".  It's how to
> tie into the driver model correctly, without it, crazy things happen as
> we have seen.

I agree it is *the* right this to do!

> > > In this patch, we convert the existing code to use an explicit
> > > sdw_slave_type, then define a sdw_master_device structure.
> > 
> > Please split that up, we should do the conversions required first and
> > then do addition of new things.
> 
> Can you really do that in two different steps?

Looking at it, the move of existing types first and then adding the new
type

> 
> > > +struct device_type sdw_master_type = {
> > > +	.name =		"soundwire_master",
> > > +	.release =	sdw_master_device_release,
> > > +};
> > > +
> > > +/**
> > > + * sdw_master_device_add() - create a Linux Master Device representation.
> > > + * @parent: the parent Linux device (e.g. a PCI device)
> > > + * @fwnode: the parent fwnode (e.g. an ACPI companion device to the parent)
> > > + * @link_ops: link-specific ops (optional)
> > > + * @link_id: link index as defined by MIPI DisCo specification
> > > + * @pdata: private data (e.g. register base, offsets, platform quirks, etc).
> > > + *
> > > + * The link_ops argument can be NULL, it is only used when link-specific
> > > + * initializations and power-management are required.
> > > + */
> > > +struct sdw_master_device
> > > +*sdw_master_device_add(struct device *parent,
> > > +		       struct fwnode_handle *fwnode,
> > > +		       struct sdw_link_ops *link_ops,
> > > +		       int link_id,
> > > +		       void *pdata)
> > > +{
> > > +	struct sdw_master_device *md;
> > > +	int ret;
> > > +
> > > +	md = kzalloc(sizeof(*md), GFP_KERNEL);
> > > +	if (!md)
> > > +		return ERR_PTR(-ENOMEM);
> > > +
> > > +	md->link_id = link_id;
> > > +	md->pdata = pdata;
> > > +	md->link_ops = link_ops;
> > > +
> > > +	md->dev.parent = parent;
> > > +	md->dev.fwnode = fwnode;
> > > +	md->dev.bus = &sdw_bus_type;
> > > +	md->dev.type = &sdw_master_type;
> > > +	md->dev.dma_mask = md->dev.parent->dma_mask;
> > > +	dev_set_name(&md->dev, "sdw-master-%d", md->link_id);
> > > +
> > > +	if (link_ops && link_ops->driver) {
> > > +		/*
> > > +		 * A driver is only needed for ASoC integration (need
> > > +		 * driver->name) and for link-specific power management
> > > +		 * w/ a pm_dev_ops structure.
> > 
> > That is not true for everyone, it is only true for Intel, pls call that
> > out as well...
> 
> Why is it not true for everyone?  How else do you get the pm stuff back
> to your hardware?

The rest of the world would do using the real controller device. For
example the soundwire controller on Qualcomm devices is enumerated as a
DT device and is using these...

If Intel had a standalone controller or enumerated as individual
functions, it would have been a PCI device and would manage as such

Thanks
-- 
~Vinod