[PATCH v21 3/4] i2c: ast2600: Add controller driver for new register layout

[Ryan Chen]

> Subject: Re: [PATCH v21 3/4] i2c: ast2600: Add controller driver for new
> register layout
> 
> On 27/10/2025 07:12, Ryan Chen wrote:
> > Add i2c-ast2600 new register mode driver to support AST2600 i2c new
> > register mode. This i2c-ast2600 new driver and the legacy i2c-aspeed
> > driver both match the same compatible string "aspeed,ast2600-i2c-bus"
> > because they target the same I2C controller IP on AST2600. However,
> > AST2600 SoCs may configure
> 
> Where did you document ABI compatibility of new driver with old DTS, which
> do not have required properties and uses old reg?
> 
> I don't see that. I actually clearly see:
> 
> > +	i2c_bus = devm_kzalloc(dev, sizeof(*i2c_bus), GFP_KERNEL);
> > +	if (!i2c_bus)
> > +		return -ENOMEM;
> > +
> > +	i2c_bus->reg_base = devm_platform_ioremap_resource(pdev, 0);
> > +	if (IS_ERR(i2c_bus->reg_base))
> > +		return PTR_ERR(i2c_bus->reg_base);
> > +
> > +	rst = devm_reset_control_get_shared_deasserted(dev, NULL);
> > +	if (IS_ERR(rst))
> > +		return dev_err_probe(dev, PTR_ERR(rst), "Missing reset ctrl\n");
> > +
> > +	i2c_bus->global_regs =
> > +		syscon_regmap_lookup_by_phandle(dev_of_node(dev),
> "aspeed,global-regs");
> > +	if (IS_ERR(i2c_bus->global_regs))
> > +		return PTR_ERR(i2c_bus->global_regs);
> 
> 
> ABI break. No explanation of that ABI break, not even mentioning it.
> 
> There is no reason to break the ABI. Your new driver *must* support both
> interfaces and both DTB. You *cannot* (please listen carefully, because in the
> past we wasted a lot of time discussing simple statements) rely on legacy
> driver binding to old DTB, because it is the same compatible.
> 
> You need to rewrite all this to handle both DTBs in backwards compatible way.
> 
> NAK
I think I understand your point about ABI break.
I will try to merge i2c-ast260.c with legacy i2c-aspeed.c
Add new file i2c-aspeed-core.c to do legacy probe and i2c-ast2600 probe.

	if (of_device_is_compatible(dev_of_node(dev), "aspeed,ast2600-i2c-bus") &&
	    of_parse_phandle(dev_of_node(dev), "aspeed,global-regs", 0)) {
		ret = ast2600_i2c_probe(pdev);
	} else {
		ret = aspeed_i2c_probe(pdev);
	}

That would not have ABI break.

> 
> Best regards,
> Krzysztof