Re: [PATCH] gpio: add Penwell gpio support

[Du, Alek]

On Fri, 21 May 2010 05:28:21 +0800
Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:

> On Tue, 18 May 2010 15:40:25 +0800
> "Du, Alek" <alek.du@xxxxxxxxx> wrote:
> 
> > >From 963f6e83843b0f94f8a5337def6e897ec5bb99bf Mon Sep 17 00:00:00 2001
> > From: Alek Du <alek.du@xxxxxxxxx>
> > Date: Thu, 25 Feb 2010 14:32:46 +0800
> > Subject: [PATCH] gpio: add Penwell gpio support
> > 
> > Intel Penwell chip has two 96 pins GPIO blocks, which are very similiar as
> > Intel Langwell chip GPIO block, except for pin number difference. This
> > patch expends the original Langwell GPIO driver to support Penwell's.
> > 
> 
> Has the driver been retested on Moorestown?

Yes, retested with Moorestown platform.

> 
> > -static int lnw_gpio_get(struct gpio_chip *chip, unsigned offset)
> > +static inline void __iomem *gpio_reg(struct gpio_chip *chip, unsigned offset,
> > +			enum GPIO_REG reg_type)
> >  {
> >  	struct lnw_gpio *lnw = container_of(chip, struct lnw_gpio, chip);
> > +	unsigned nreg = chip->ngpio / 32;
> >  	u8 reg = offset / 32;
> > -	void __iomem *gplr;
> > +	void __iomem *ptr;
> > +
> > +	ptr = (void __iomem *)(lnw->reg_base + reg_type * nreg * 4 + reg * 4);
> > +	return ptr;
> > +}
> 
> inlining this function was probably the wrong thing to do.  But modern
> gcc's often just ignore the `inline' and do the right thing anyway.
>

Yes, as I looked at the assembly code, the function is too big. I should remove "inline".

> 
> > -static struct pci_device_id lnw_gpio_ids[] = {
> > -	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x080f) },
> > +static struct pci_device_id lnw_gpio_ids[] = {      /* pin number */
> > +	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x080f), .driver_data = 64 },
> > +	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x081f), .driver_data = 96 },
> > +	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x081a), .driver_data = 96 },
> >  	{ 0, }
> 
> I suppose we should be using DEFINE_PCI_DEVICE_TABLE() here.
> 
>
Yes, here is the incremental patch against previous one (I got the mail said
the previous one is in mm tree now):