Re: [PATCH v2 2/2] mfd: intel_quark_i2c_gpio: enable MSI interrupt

[Lee Jones]

On Wed, 24 Mar 2021, Andy Shevchenko wrote:

> On Wed, Mar 24, 2021 at 10:29:31AM +0000, Lee Jones wrote:
> > On Tue, 23 Mar 2021, Andy Shevchenko wrote:
> > 
> > > Allow interrupts to be MSI if supported by hardware.
> > > 
> > > Signed-off-by: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>
> > > ---
> > > v2: new patch
> > >  drivers/mfd/intel_quark_i2c_gpio.c | 21 +++++++++++++++------
> > >  1 file changed, 15 insertions(+), 6 deletions(-)
> > > 
> > > diff --git a/drivers/mfd/intel_quark_i2c_gpio.c b/drivers/mfd/intel_quark_i2c_gpio.c
> > > index 52728a963c17..16ce9cb3aa2f 100644
> > > --- a/drivers/mfd/intel_quark_i2c_gpio.c
> > > +++ b/drivers/mfd/intel_quark_i2c_gpio.c
> > > @@ -169,8 +169,8 @@ static int intel_quark_i2c_setup(struct pci_dev *pdev, struct mfd_cell *cell)
> > >  	res[INTEL_QUARK_IORES_MEM].end =
> > >  		pci_resource_end(pdev, MFD_I2C_BAR);
> > >  
> > > -	res[INTEL_QUARK_IORES_IRQ].start = pdev->irq;
> > > -	res[INTEL_QUARK_IORES_IRQ].end = pdev->irq;
> > > +	res[INTEL_QUARK_IORES_IRQ].start = pci_irq_vector(pdev, 0);
> > > +	res[INTEL_QUARK_IORES_IRQ].end = pci_irq_vector(pdev, 0);
> > >  
> > >  	pdata = devm_kzalloc(dev, sizeof(*pdata), GFP_KERNEL);
> > >  	if (!pdata)
> > > @@ -217,7 +217,7 @@ static int intel_quark_gpio_setup(struct pci_dev *pdev, struct mfd_cell *cell)
> > >  	pdata->properties->idx		= 0;
> > >  	pdata->properties->ngpio	= INTEL_QUARK_MFD_NGPIO;
> > >  	pdata->properties->gpio_base	= INTEL_QUARK_MFD_GPIO_BASE;
> > > -	pdata->properties->irq[0]	= pdev->irq;
> > > +	pdata->properties->irq[0]	= pci_irq_vector(pdev, 0);
> > >  
> > >  	cell->platform_data = pdata;
> > >  	cell->pdata_size = sizeof(*pdata);
> > > @@ -245,22 +245,30 @@ static int intel_quark_mfd_probe(struct pci_dev *pdev,
> > >  	if (ret)
> > >  		return ret;
> > >  
> > > +	pci_set_master(pdev);
> > > +
> > > +	ret = pci_alloc_irq_vectors(pdev, 1, 1, PCI_IRQ_ALL_TYPES);
> > 
> > Is there any way these magic number can be defined or sizeof()'ed?
> 
> Grep for it in the kernel, it's rarely defined.

I already did.  It is sometimes defined, other times not.

Also, past acceptance does not guarantee ideal/correct usage.

> The semantic is min-max range and having two defines (*) here for these seems
> to me as an utter overkill.
> 
> Of course, if you insist I may do it.
> 
> *) since value is the same, we might have one definition, but it will be even
>    more confusion to have it as a min and max at the same time.

It's just tricky to decypher for people who do not know the API, which
is most people, myself included.  For APIs like usleep_range() et al.,
obviously this makes no sense at all.

What defines a vector?

-- 
Lee Jones [李琼斯]
Senior Technical Lead - Developer Services
Linaro.org │ Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog