Re: DSA using cpsw and lan9303

[Vladimir Oltean <olteanv@gmail.com>]

On Wed, Feb 16, 2022 at 02:23:24PM +0000, Måns Rullgård wrote:
> Vladimir Oltean <olteanv@gmail.com> writes:
> 
> > On Wed, Feb 16, 2022 at 01:17:47PM +0000, Måns Rullgård wrote:
> >> > Some complaints about accessing the CPU port as dsa_to_port(chip->ds, 0),
> >> > but it's not the first place in this driver where that is done.
> >> 
> >> What would be the proper way to do it?
> >
> > Generally speaking:
> >
> > 	struct dsa_port *cpu_dp;
> >
> > 	dsa_switch_for_each_cpu_port(cpu_dp, ds)
> > 		break;
> >
> > 	// use cpu_dp
> >
> > If your code runs after dsa_tree_setup_default_cpu(), which contains the
> > "DSA: tree %d has no CPU port\n" check, you don't even need to check
> > whether cpu_dp was found or not - it surely was. Everything that runs
> > after dsa_register_switch() has completed successfully - for example the
> > DSA ->setup() method - qualifies here.
> 
> In this particular driver, the setup function contains this:
> 
> 	/* Make sure that port 0 is the cpu port */
> 	if (!dsa_is_cpu_port(ds, 0)) {
> 		dev_err(chip->dev, "port 0 is not the CPU port\n");
> 		return -EINVAL;
> 	}
> 
> I take this to mean that port 0 is guaranteed to be the cpu port.  Of
> course, it can't hurt to be thorough just in case that check is ever
> removed.

Yes, I saw that, and I said that there are other places in the driver
that assume port 0 is the CPU port. Although I don't know why that is,
if the switch can only operate like that, etc. I just pointed out how it
would be preferable to get a hold of the CPU port in a regular DSA
driver without any special constraints.