Re: [net-next: PATCH 1/3] net: mvmdio: add ACPI support

From: Marcin Wojtas <mw@semihalf.com>
To: Andy Shevchenko <andy.shevchenko@gmail.com>
Cc: Andrew Lunn <andrew@lunn.ch>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"netdev@vger.kernel.org" <netdev@vger.kernel.org>,
	"davem@davemloft.net" <davem@davemloft.net>,
	"kuba@kernel.org" <kuba@kernel.org>,
	"linux@armlinux.org.uk" <linux@armlinux.org.uk>,
	"jaz@semihalf.com" <jaz@semihalf.com>,
	"gjb@semihalf.com" <gjb@semihalf.com>,
	"upstream@semihalf.com" <upstream@semihalf.com>,
	"Samer.El-Haj-Mahmoud@arm.com" <Samer.El-Haj-Mahmoud@arm.com>,
	"jon@solid-run.com" <jon@solid-run.com>
Subject: Re: [net-next: PATCH 1/3] net: mvmdio: add ACPI support
Date: Tue, 15 Jun 2021 17:09:12 +0200	[thread overview]
Message-ID: <CAPv3WKeubNaxpv442d57bEqA1ZtPcTXOswcsuCsregW_2Akdww@mail.gmail.com> (raw)
In-Reply-To: <CAHp75VdMsYJMCwH2o14e7nJBTj6A38dkcZJ+0WQfnW=keOyoAg@mail.gmail.com>

Hi,

niedz., 13 cze 2021 o 22:08 Andy Shevchenko
<andy.shevchenko@gmail.com> napisał(a):
>
>
>
> On Sunday, June 13, 2021, Andrew Lunn <andrew@lunn.ch> wrote:
>>
>> > @@ -336,7 +338,7 @@ static int orion_mdio_probe(struct platform_device *pdev)
>> >                       dev_warn(&pdev->dev,
>> >                                "unsupported number of clocks, limiting to the first "
>> >                                __stringify(ARRAY_SIZE(dev->clk)) "\n");
>> > -     } else {
>> > +     } else if (!has_acpi_companion(&pdev->dev)) {
>> >               dev->clk[0] = clk_get(&pdev->dev, NULL);
>> >               if (PTR_ERR(dev->clk[0]) == -EPROBE_DEFER) {
>> >                       ret = -EPROBE_DEFER;
>>
>> Is this needed? As you said, there are no clocks when ACPI is used, So
>> doesn't clk_get() return -ENODEV? Since this is not EPRODE_DEFER, it
>> keeps going. The clk_prepare_enable() won't be called.
>

Indeed, I'll double check if it works and will keep the if {} else {} intact.

>
>
> The better approach is to switch to devm_get_clk_optional() as I have done in several drivers, IIRC recently in mvpp2
>
>
> https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git/commit/?id=cf3399b731d36bc780803ff63e4d480a1efa33ac
>

Yes, this would be a nice improvement, however the
devm_get_clk_optional requires clock name (type char *) - mvmdio uses
raw indexes, so this helper unfortunately seems to be not applicable.

>>
>> > -     ret = of_mdiobus_register(bus, pdev->dev.of_node);
>> > +     if (pdev->dev.of_node)
>> > +             ret = of_mdiobus_register(bus, pdev->dev.of_node);
>> > +     else if (is_acpi_node(pdev->dev.fwnode))
>> > +             ret = acpi_mdiobus_register(bus, pdev->dev.fwnode);
>> > +     else
>> > +             ret = -EINVAL;
>> >       if (ret < 0) {
>> >               dev_err(&pdev->dev, "Cannot register MDIO bus (%d)\n", ret);
>> >               goto out_mdio;
>> > @@ -383,6 +390,9 @@ static int orion_mdio_probe(struct platform_device *pdev)
>> >       if (dev->err_interrupt > 0)
>> >               writel(0, dev->regs + MVMDIO_ERR_INT_MASK);
>> >
>> > +     if (has_acpi_companion(&pdev->dev))
>> > +             return ret;
>> > +
>>
>> I think this can also be removed for the same reason.
>>
>> We should try to avoid adding has_acpi_companion() and
>> !pdev->dev.of_node whenever we can. It makes the driver code too much
>> of a maze.

Clock routines silently accept NULL pointers, so it will be safe to
drop this addition in v2.

Best regards,
Marcin