Re: [alsa-devel] [PATCH v4 04/15] soundwire: Add MIPI DisCo property helpers

[Pierre-Louis Bossart]

On 12/3/17 10:52 AM, Vinod Koul wrote:
> On Fri, Dec 01, 2017 at 04:49:01PM -0600, Pierre-Louis Bossart wrote:
>
> > > +int sdw_master_read_prop(struct sdw_bus *bus)
> > > +{
> > > +	struct sdw_master_prop *prop = &bus->prop;
> > > +	struct fwnode_handle *link;
> > > +	unsigned int count = 0;
> > > +	char name[32];
> > > +	int nval, i;
> > > +
> > > +	device_property_read_u32(bus->dev,
> > > +			"mipi-sdw-sw-interface-revision", &prop->revision);
> > > +	device_property_read_u32(bus->dev, "mipi-sdw-master-count", &count);
> > > +
> > > +	/* Find link handle */
> > > +	snprintf(name, sizeof(name),
> > > +			"mipi-sdw-link-%d-subproperties", bus->link_id);
> >
> > if you follow the DisCo spec, this property is at the controller level,
> > isn't there a confusion between controller/master here, and consequently are
> > we reading the same things multiple times or using the wrong bus parameter?
>
> Not sure I follow, this one is for a specific master ie a specfic link.
> we need to read respective master thru mipi-sdw-link-N-subproperties
>
> > If I look at intel_probe(), there is a clear reference to a link_id, and
> > then you set the pointer to this read_prop which reads the number of links,
> > which looks like the wrong order. You can't assign a link ID before knowing
> > how many links there are - or you may be unable to detect issues.
>
> Sorry I dont follow this part. FWIW, when master driver is enumerated it
> know the link_id value and then sets the read_prop and then these are read.
>
> Here we are reading "a specific link property" with the knowledge of link_id
> value...

the sdw_master-count is at the controller level, and the linkid has to 
be < master_count.

The fact that you are reading this property for each master instance is 
the problem.

> > > +	fwnode_property_read_u32(link, "mipi-sdw-default-frame-rate",
> > > +				&prop->default_frame_rate);
> > > +	fwnode_property_read_u32(link, "mipi-sdw-default-frame-row-size",
> > > +				&prop->default_row);
> > > +	fwnode_property_read_u32(link, "mipi-sdw-default-frame-col-size",
> >
> > This is fine, just wondering if we should warnings if the values make no
> > sense, e.g. the DisCo spec states in Note1 page 15 that the values are
> > interrelated.
>
> I think we discussed in past and that would kind of form the firmware
> validation. We check all the values to see if firmware gave us sane values..
>
> > > +	/*
> > > +	 * Based on each DPn port, get source and sink dpn properties.
> > > +	 * Also, some ports can operate as both source or sink.
> > > +	 */
> > > +
> > > +	/* Allocate memory for set bits in port lists */
> > > +	nval = hweight32(prop->source_ports);
> > > +	num_of_ports += nval;
> >
> > this and...
> >
> > > +	prop->src_dpn_prop = devm_kcalloc(&slave->dev, nval,
> > > +				sizeof(*prop->src_dpn_prop), GFP_KERNEL);
> > > +	if (!prop->src_dpn_prop)
> > > +		return -ENOMEM;
> > > +
> > > +	/* Read dpn properties for source port(s) */
> > > +	sdw_slave_read_dpn(slave, prop->src_dpn_prop, nval,
> > > +			prop->source_ports, "source");
> > > +
> > > +	nval = hweight32(prop->sink_ports);
> > > +	num_of_ports += nval;
> >
> > ... this is no longer needed since...
> >
> > > +	prop->sink_dpn_prop = devm_kcalloc(&slave->dev, nval,
> > > +				sizeof(*prop->sink_dpn_prop), GFP_KERNEL);
> > > +	if (!prop->sink_dpn_prop)
> > > +		return -ENOMEM;
> > > +
> > > +	/* Read dpn properties for sink port(s) */
> > > +	sdw_slave_read_dpn(slave, prop->sink_dpn_prop, nval,
> > > +			prop->sink_ports, "sink");
> > > +
> > > +	/* some ports are bidirectional so check total ports by ORing */
> > > +	nval = prop->source_ports | prop->sink_ports;
> > > +	num_of_ports = hweight32(nval) + 1; /* add 1 for DP0 */
> >
> > ... you reassign the value here. That was one earlier feedback from me but
> > you left the variable incrementation in the code.
>
> This seems to have artifact of merge conflicts as I clearly remember removing
> this, thanks for pointing will remove these..
>
> > > +/**
> > > + * enum sdw_clk_stop_mode - Clock Stop modes
> > > + * @SDW_CLK_STOP_MODE0: Slave can continue operation seamlessly on clock
> > > + * restart
> > > + * @SDW_CLK_STOP_MODE1: Slave may have entered a deeper power-saving mode,
> > > + * not capable of continuing operation seamlessly when the clock restarts
> > > + */
> > > +enum sdw_clk_stop_mode {
> > > +	SDW_CLK_STOP_MODE0 = 1,
> > > +	SDW_CLK_STOP_MODE1 = 2,
> >
> > why not 0 and 1?
>
> why not 1 and 2 :D
>
> I think it was to ensure we have a non zero value, but am not sure, will
> check though..

I don't think the value matter and you should use the same conventions 
for such enums.

>