Re: [PATCH v3 1/2] SPI: Add SPI driver for Sunplus SP7021

From: Andy Shevchenko
Date: Fri Nov 26 2021 - 06:17:01 EST


On Fri, Nov 26, 2021 at 9:49 AM Lh Kuo 郭力豪 <lh.Kuo@xxxxxxxxxxx> wrote:

(Uncommented is okay)

...

> > > ret = devm_request_irq(dev, pspim->mas_irq, sp7021_spi_mas_irq
> > > , IRQF_TRIGGER_RISING,
> > > pdev->name, pspim);
> >
> > Ugly indentation.
> >
>
> Amended as follows, is it okay?
>
> ret = devm_request_irq(dev, pspim->mas_irq, sp7021_spi_mas_irq
> , IRQF_TRIGGER_RISING, pdev->name, pspim);
> if (ret)
> return ret;

Still not okay. Have you seen this style somewhere in the kernel?
Hint: something is really wrong with comma's location.

...

> > > pm_runtime_enable(dev);
> > >
> > > ret = devm_spi_register_controller(dev, ctlr);
> >
> > You can't mix non-devm with devm APIs. Either all non-devm, or devm followed by non-devm.

> I don't understand so I need to change to spi_register_controller(ctlr)? why?

I haven't told you that. What I'm saying is this:
1) all calls are devm_*() - OK!
2) all calls are non-devm_*() OK!
3) devm_*() followed by non-devm_*() OK!
4) non-devm_*() call followed by devm_*() call NOT okay!

You need to fulfil your homework (see plenty of the examples in the
Linux kernel source tree on how to proceed).

> I modified the remove-function as follows. I think devm_spi_register_controller(dev, ctlr); should be no problem in the probe funciton.

It has ordering issues. That's why 4) above is not okay.

> static int sp7021_spi_controller_remove(struct platform_device *pdev)
> {
> struct spi_controller *ctlr = dev_get_drvdata(&pdev->dev);
> struct sp7021_spi_ctlr *pspim = spi_master_get_devdata(ctlr);
>
> pm_runtime_disable(&pdev->dev);
> pm_runtime_set_suspended(&pdev->dev);
> reset_control_assert(pspim->rstc);
> clk_disable_unprepare(pspim->spi_clk);
>
> return 0;
> }

--
With Best Regards,
Andy Shevchenko