Re: [PATCH 2/2] i3c: master: Add driver for MediaTek IP

[Boris Brezillon]

On Tue, 11 Jun 2019 20:25:22 +0800
Qii Wang <qii.wang@xxxxxxxxxxxx> wrote:

> On Tue, 2019-06-04 at 20:26 +0800, Qii Wang wrote:
> > On Tue, 2019-06-04 at 09:58 +0200, Boris Brezillon wrote:  
> > > On Mon, 3 Jun 2019 11:51:03 +0800
> > > Qii Wang <qii.wang@xxxxxxxxxxxx> wrote:
> > > 
> > >   
> > > > +static int mtk_i3c_master_probe(struct platform_device *pdev)
> > > > +{
> > > > +	struct device *dev = &pdev->dev;
> > > > +	struct mtk_i3c_master *master;
> > > > +	struct resource *res;
> > > > +	int ret, irqnr;
> > > > +
> > > > +	master = devm_kzalloc(dev, sizeof(*master), GFP_KERNEL);
> > > > +	if (!master)
> > > > +		return -ENOMEM;
> > > > +
> > > > +	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "main");
> > > > +	master->regs = devm_ioremap_resource(dev, res);
> > > > +	if (IS_ERR(master->regs))
> > > > +		return PTR_ERR(master->regs);
> > > > +
> > > > +	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "dma");
> > > > +	master->dma_regs = devm_ioremap_resource(dev, res);
> > > > +	if (IS_ERR(master->dma_regs))
> > > > +		return PTR_ERR(master->dma_regs);
> > > > +
> > > > +	irqnr = platform_get_irq(pdev, 0);
> > > > +	if (irqnr < 0)
> > > > +		return irqnr;
> > > > +
> > > > +	ret = devm_request_irq(dev, irqnr, mtk_i3c_master_irq,
> > > > +			       IRQF_TRIGGER_NONE, DRV_NAME, master);
> > > > +	if (ret < 0) {
> > > > +		dev_err(dev, "Request I3C IRQ %d fail\n", irqnr);
> > > > +		return ret;
> > > > +	}
> > > > +
> > > > +	ret = of_property_read_u32(pdev->dev.of_node, "clock-div",
> > > > +				   &master->clk_src_div);  
> > > 
> > > You say in one comment that this clock divider is fixed in HW but might
> > > change on a per-SoC basis. If that's the case, you should get rid of
> > > this clock-div prop and attach the divider to the compatible (using an
> > > mtk_i3c_master_variant struct that contains a divider field).
> > >   
> > 
> > ok, I will attach the divider to the compatible.
> >   
> I have rechecked your comment, maybe I have misunderstood what you mean.
> "clock-div" changes according to i2c source clock, different project may
> change i2c source clock, The previous dt-binding may be misleading, I
> will modify it.

Is it fixed or configurable? Maybe it should be modeled as a clk
driver. What's for sure is that we shouldn't have this divider defined
in the DT.