On 12/18/2012 06:55 AM, Jean-Christophe PLAGNIOL-VILLARD wrote: > On 20:47 Mon 17 Dec , Wolfgang Grandegger wrote: >> On 12/17/2012 07:02 PM, Roland Stigge wrote: >>> On 12/17/2012 06:37 PM, Wolfgang Grandegger wrote: >>>> /* Do synchronous data output with a single write access */ >>>> __raw_writel(~mask, pio + PIO_OWDR); >>>> __raw_writel(mask, pio + PIO_OWER); >>>> __raw_writel(val, pio + PIO_ODSR); >>>> >>>> For caching we would need a storage. Not sure if it's worth compared to >>>> a context switch into the kernel. >>> >>> Block GPIO is not only for you in userspace. ;-) You can also implement >>> efficient n-bit bus I/O in kernel drivers, n-bit-banging. :-) So not >>> always context switches involved. >> >> OK, what do you think about the following untested patch: >> >> From b44cad16cbbca84715dffd4cb5268497216add25 Mon Sep 17 00:00:00 2001 >> From: Wolfgang Grandegger <w...@grandegger.com> >> Date: Mon, 3 Dec 2012 08:31:55 +0100 >> Subject: [PATCH 1/2] gpio: add GPIO block callback functions for AT91 >> >> Signed-off-by: Wolfgang Grandegger <w...@grandegger.com> >> --- >> arch/arm/mach-at91/gpio.c | 29 +++++++++++++++++++++++++++++ >> 1 file changed, 29 insertions(+) >> >> diff --git a/arch/arm/mach-at91/gpio.c b/arch/arm/mach-at91/gpio.c >> index be42cf0..cf6bd45 100644 >> --- a/arch/arm/mach-at91/gpio.c >> +++ b/arch/arm/mach-at91/gpio.c >> @@ -42,13 +42,16 @@ struct at91_gpio_chip { >> void __iomem *regbase; /* PIO bank virtual address */ >> struct clk *clock; /* associated clock */ >> struct irq_domain *domain; /* associated irq domain */ >> + unsigned long mask_shadow; /* synchronous data output */ >> }; >> >> #define to_at91_gpio_chip(c) container_of(c, struct at91_gpio_chip, chip) >> >> static void at91_gpiolib_dbg_show(struct seq_file *s, struct gpio_chip >> *chip); >> static void at91_gpiolib_set(struct gpio_chip *chip, unsigned offset, int >> val); >> +static void at91_gpiolib_set_block(struct gpio_chip *chip, unsigned long >> mask, unsigned long val); >> static int at91_gpiolib_get(struct gpio_chip *chip, unsigned offset); >> +static unsigned long at91_gpiolib_get_block(struct gpio_chip *chip, >> unsigned long mask); >> static int at91_gpiolib_direction_output(struct gpio_chip *chip, >> unsigned offset, int val); >> static int at91_gpiolib_direction_input(struct gpio_chip *chip, >> @@ -62,7 +65,9 @@ static int at91_gpiolib_to_irq(struct gpio_chip *chip, >> unsigned offset); >> .direction_input = at91_gpiolib_direction_input, \ >> .direction_output = at91_gpiolib_direction_output, \ >> .get = at91_gpiolib_get, \ >> + .get_block = at91_gpiolib_get_block, \ >> .set = at91_gpiolib_set, \ >> + .set_block = at91_gpiolib_set_block, \ >> .dbg_show = at91_gpiolib_dbg_show, \ >> .to_irq = at91_gpiolib_to_irq, \ >> .ngpio = nr_gpio, \ >> @@ -896,6 +901,16 @@ static int at91_gpiolib_get(struct gpio_chip *chip, >> unsigned offset) >> return (pdsr & mask) != 0; >> } >> >> +static unsigned long at91_gpiolib_get_block(struct gpio_chip *chip, >> unsigned long mask) >> +{ >> + struct at91_gpio_chip *at91_gpio = to_at91_gpio_chip(chip); >> + void __iomem *pio = at91_gpio->regbase; >> + u32 pdsr; >> + >> + pdsr = __raw_readl(pio + PIO_PDSR); >> + return pdsr & mask; >> +} >> + >> static void at91_gpiolib_set(struct gpio_chip *chip, unsigned offset, int >> val) >> { >> struct at91_gpio_chip *at91_gpio = to_at91_gpio_chip(chip); >> @@ -905,6 +920,20 @@ static void at91_gpiolib_set(struct gpio_chip *chip, >> unsigned offset, int val) >> __raw_writel(mask, pio + (val ? PIO_SODR : PIO_CODR)); >> } >> >> +static void at91_gpiolib_set_block(struct gpio_chip *chip, unsigned long >> mask, unsigned long val) >> +{ >> + struct at91_gpio_chip *at91_gpio = to_at91_gpio_chip(chip); >> + void __iomem *pio = at91_gpio->regbase; >> + >> + /* Do synchronous data output with a single write access */ >> + if (mask != at91_gpio->mask_shadow) { >> + at91_gpio->mask_shadow = mask; >> + __raw_writel(~mask, pio + PIO_OWDR); >> + __raw_writel(mask, pio + PIO_OWER); >> + } >> + __raw_writel(val, pio + PIO_ODSR); >> +} > this driver is only for old at91 platfrom if you touch at91 you need to update > the pinctrl too
Well, the patch is for the hardware I have at hand and I can test. There are many other GPIO hardware interfaces which could be enhanced with block gpio. Roland only did it for the interfaces in driver/gpio. Also, I think, an ACK for this patch series would be nice before we continue. Wolfgang. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/