sh-pfc: Fix GPIO to IRQ mapping
Subsystems (4)
cover sh-pfc: Fix GPIO to IRQ mapping
Hello,
This patch set fixes conversion from a GPIO number to an IRQ number in the sh-pfc driver in the device tree case.
Patches 01/10 to 05/10 clean up the driver to prepare it for the core fix in patch 06/10. As explained in patch 06/10 commit message, on non-DT platforms IRQ controllers associated with the GPIOs have a fixed IRQ base value known at compile time. The sh-pfc driver translates GPIO number to IRQ numbers using a hardcoded table. This mechanism breaks on DT platforms, as the IRQ base values are dynamic in that case.
The patch set fixes this by specifying IRQs associated with GPIOs in IRQ resources, populated automatically from the device tree. When IRQ resources are specified the driver requires one IRQ resource per GPIO able to generate an interrupt, and uses the translation table to compute the IRQ resource offset instead of the IRQ number.
Patches 07/10 to 09/10 then add PFC interrupts in DT for the SH73A0, R8A73A4 and R9A7740 SoCs. The only missing platform here is SH7372 due to its very uncomplete DT support. All other Renesas ARM platforms use a standaline GPIO driver and are thus not affected.
Patch 10/10 finally adds a gpio-keys device to the Armadillo board DTS, allowing the whole patch set to the tested.
The sh-pfc changes need to be merged first. Linus, could you please provide a stable branch with the first 6 patches when you will apply them to your tree ?
Laurent Pinchart (10):
sh-pfc: Turn unsigned indices into unsigned int
sh-pfc: Terminate gpios array by -1
sh-pfc: sh73a0: Add missing IRQ15
sh-pfc: sh73a0: Sort IRQ entries by IRQ number
sh-pfc: Rename sh_pfc window field to windows
sh-pfc: Support GPIO to IRQ mapping specified IRQ resources
ARM: shmobile: r8a73a4: Specify PFC interrupts in DT
ARM: shmobile: r8a7740: Specify PFC interrupts in DT
ARM: shmobile: sh73a0: Specify PFC interrupts in DT
ARM: shmobile: armadillo: dts: Add gpio-keys device
.../bindings/pinctrl/renesas,pfc-pinctrl.txt | 14 ++++ arch/arm/boot/dts/r8a73a4.dtsi | 16 +++++ .../boot/dts/r8a7740-armadillo800eva-reference.dts | 29 +++++++++ arch/arm/boot/dts/r8a7740.dtsi | 9 +++ arch/arm/boot/dts/sh73a0.dtsi | 9 +++ drivers/pinctrl/sh-pfc/core.c | 76 ++++++++++++++++------ drivers/pinctrl/sh-pfc/core.h | 4 +- drivers/pinctrl/sh-pfc/gpio.c | 24 +++++-- drivers/pinctrl/sh-pfc/pfc-r8a73a4.c | 4 +- drivers/pinctrl/sh-pfc/pfc-r8a7740.c | 2 +- drivers/pinctrl/sh-pfc/pfc-sh7372.c | 2 +- drivers/pinctrl/sh-pfc/pfc-sh73a0.c | 54 +++++++-------- drivers/pinctrl/sh-pfc/sh_pfc.h | 4 +- 13 files changed, 188 insertions(+), 59 deletions(-)
Signature · 3 lines
Regards, Laurent Pinchart
PATCH 1/10 sh-pfc: Turn unsigned indices into unsigned int
Some indices take positive values only, make them unsigned.
--- drivers/pinctrl/sh-pfc/core.c | 8 ++++---- drivers/pinctrl/sh-pfc/gpio.c | 2 +- 2 files changed, 5 insertions(+), 5 deletions(-)
drivers/pinctrl/sh-pfc/core.c +4 -4
@@ -29,7 +29,7 @@static int sh_pfc_ioremap (struct sh_pfc *pfc,struct platform_device *pdev) {struct resource *res; -int k; +unsigned int k;if (pdev->num_resources ==0 )return -EINVAL ;
@@ -147,7 +147,7 @@ static void sh_pfc_config_reg_helper(struct sh_pfc *pfc,unsigned long *maskp, unsigned long *posp) { - int k; + unsigned int k; *mapped_regp =sh_pfc_phys_to_virt (pfc, crp->reg);
@@ -196,7 +196,7 @@ static int sh_pfc_get_config_reg(struct sh_pfc *pfc, u16 enum_id, {const struct pinmux_cfg_reg *config_reg;unsigned long r_width, f_width, curr_width, ncomb; -int k, m, n, pos, bit_pos; +unsigned int k, m, n, pos, bit_pos; k =0 ;while (1 ) {
@@ -238,7 +238,7 @@ static int sh_pfc_mark_to_enum(struct sh_pfc *pfc, u16 mark, int pos, u16 *enum_idp) {const u16 *data = pfc->info->gpio_data; -int k; +unsigned int k;if (pos) { *enum_idp = data[pos +1 ];
drivers/pinctrl/sh-pfc/gpio.c +1 -1
@@ -204,7 +204,7 @@ static void gpio_pin_set(struct gpio_chip *gc, unsigned offset, int value)static int gpio_pin_to_irq (struct gpio_chip *gc,unsigned offset) {struct sh_pfc *pfc =gpio_to_pfc (gc); -int i, k; +unsigned int i, k;for (i =0 ; i < pfc->info->gpio_irq_size; i++) {unsigned short *gpios = pfc->info->gpio_irq[i].gpios;
Signature · 1 line
1.8.3.2
PATCH 1/10 sh-pfc: Turn unsigned indices into unsigned int
Quoted 3 lines · Laurent Pinchart
Some indices take positive values only, make them unsigned.
Signed-off-by Laurent Pinchart <[email protected]>
Patch applied.
Yours,
Linus Walleij
PATCH 2/10 sh-pfc: Terminate gpios array by -1
In reply to Laurent Pinchart0 is a valid GPIO value, use -1 to terminate the gpios array in IRQ lists.
--- drivers/pinctrl/sh-pfc/gpio.c | 4 ++-- drivers/pinctrl/sh-pfc/sh_pfc.h | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-)
drivers/pinctrl/sh-pfc/gpio.c +2 -2
@@ -207,9 +207,9 @@ static int gpio_pin_to_irq(struct gpio_chip *gc, unsigned offset)unsigned int i, k;for (i =0 ; i < pfc->info->gpio_irq_size; i++) { -unsigned short *gpios = pfc->info->gpio_irq[i].gpios; +short *gpios = pfc->info->gpio_irq[i].gpios; -for (k =0 ; gpios[k]; k++) { +for (k =0 ; gpios[k] >=0 ; k++) {if (gpios[k] == offset)return pfc->info->gpio_irq[i].irq; }
drivers/pinctrl/sh-pfc/sh_pfc.h +2 -2
@@ -94,11 +94,11 @@ struct pinmux_data_reg {struct pinmux_irq {int irq; -unsigned short *gpios; +short *gpios; }; #define PINMUX_IRQ(irq_nr, ids...) \ - { .irq = irq_nr, .gpios = (unsigned short []) { ids, 0 } } \ + { .irq = irq_nr, .gpios = (short []) { ids, -1 } }struct pinmux_range {u16 begin;
Signature · 1 line
1.8.3.2
PATCH 2/10 sh-pfc: Terminate gpios array by -1
Quoted 4 lines · Laurent Pinchart
0 is a valid GPIO value, use -1 to terminate the gpios array in IRQ lists.
Signed-off-by Laurent Pinchart <[email protected]>
Patch applied.
Another solution to the same problem: terminate with USHRT_MAX and for (k = 0; gpios[k] != USHRT_MAX; k++) {} but this works too...
Yours,
Linus Walleij
PATCH 3/10 sh-pfc: sh73a0: Add missing IRQ15
In reply to Laurent PinchartThe external IRQ15 input multiplexed on GPIO 0 is missing. Add it.
--- drivers/pinctrl/sh-pfc/pfc-sh73a0.c | 1 + 1 file changed, 1 insertion(+)
drivers/pinctrl/sh-pfc/pfc-sh73a0.c +1 -0
@@ -3661,6 +3661,7 @@ static const struct pinmux_data_reg pinmux_data_regs[] = { };static const struct pinmux_irq pinmux_irqs[] = { +PINMUX_IRQ (irq_pin (15 ),0 ),PINMUX_IRQ (irq_pin (19 ),9 ),PINMUX_IRQ (irq_pin (1 ),10 ),PINMUX_IRQ (irq_pin (0 ),11 ),
Signature · 1 line
1.8.3.2
PATCH 3/10 sh-pfc: sh73a0: Add missing IRQ15
Quoted 3 lines · Laurent Pinchart
The external IRQ15 input multiplexed on GPIO 0 is missing. Add it.
Signed-off-by Laurent Pinchart <[email protected]>
Patch applied.
Yours,
Linus Walleij
PATCH 4/10 sh-pfc: sh73a0: Sort IRQ entries by IRQ number
In reply to Laurent PinchartThis makes catching duplicate entries easier. Merge the two IRQ9 entries found after sorting.
--- drivers/pinctrl/sh-pfc/pfc-sh73a0.c | 47 ++++++++++++++++++------------------- 1 file changed, 23 insertions(+), 24 deletions(-)
drivers/pinctrl/sh-pfc/pfc-sh73a0.c +23 -24
@@ -3661,39 +3661,38 @@ static const struct pinmux_data_reg pinmux_data_regs[] = { };static const struct pinmux_irq pinmux_irqs[] = { -PINMUX_IRQ (irq_pin (15 ),0 ), -PINMUX_IRQ (irq_pin (19 ),9 ), -PINMUX_IRQ (irq_pin (1 ),10 ),PINMUX_IRQ (irq_pin (0 ),11 ), +PINMUX_IRQ (irq_pin (1 ),10 ), +PINMUX_IRQ (irq_pin (2 ),149 ), +PINMUX_IRQ (irq_pin (3 ),224 ), +PINMUX_IRQ (irq_pin (4 ),159 ), +PINMUX_IRQ (irq_pin (5 ),227 ), +PINMUX_IRQ (irq_pin (6 ),147 ), +PINMUX_IRQ (irq_pin (7 ),150 ), +PINMUX_IRQ (irq_pin (8 ),223 ), +PINMUX_IRQ (irq_pin (9 ),56 ,308 ), +PINMUX_IRQ (irq_pin (10 ),54 ), +PINMUX_IRQ (irq_pin (11 ),238 ), +PINMUX_IRQ (irq_pin (12 ),156 ), +PINMUX_IRQ (irq_pin (13 ),239 ), +PINMUX_IRQ (irq_pin (14 ),251 ), +PINMUX_IRQ (irq_pin (15 ),0 ), +PINMUX_IRQ (irq_pin (16 ),249 ), +PINMUX_IRQ (irq_pin (17 ),234 ),PINMUX_IRQ (irq_pin (18 ),13 ), +PINMUX_IRQ (irq_pin (19 ),9 ),PINMUX_IRQ (irq_pin (20 ),14 ),PINMUX_IRQ (irq_pin (21 ),15 ), -PINMUX_IRQ (irq_pin (31 ),26 ), -PINMUX_IRQ (irq_pin (30 ),27 ), -PINMUX_IRQ (irq_pin (29 ),28 ),PINMUX_IRQ (irq_pin (22 ),40 ),PINMUX_IRQ (irq_pin (23 ),53 ), -PINMUX_IRQ (irq_pin (10 ),54 ), -PINMUX_IRQ (irq_pin (9 ),56 ), +PINMUX_IRQ (irq_pin (24 ),118 ), +PINMUX_IRQ (irq_pin (25 ),164 ),PINMUX_IRQ (irq_pin (26 ),115 ),PINMUX_IRQ (irq_pin (27 ),116 ),PINMUX_IRQ (irq_pin (28 ),117 ), -PINMUX_IRQ (irq_pin (24 ),118 ), -PINMUX_IRQ (irq_pin (6 ),147 ), -PINMUX_IRQ (irq_pin (2 ),149 ), -PINMUX_IRQ (irq_pin (7 ),150 ), -PINMUX_IRQ (irq_pin (12 ),156 ), -PINMUX_IRQ (irq_pin (4 ),159 ), -PINMUX_IRQ (irq_pin (25 ),164 ), -PINMUX_IRQ (irq_pin (8 ),223 ), -PINMUX_IRQ (irq_pin (3 ),224 ), -PINMUX_IRQ (irq_pin (5 ),227 ), -PINMUX_IRQ (irq_pin (17 ),234 ), -PINMUX_IRQ (irq_pin (11 ),238 ), -PINMUX_IRQ (irq_pin (13 ),239 ), -PINMUX_IRQ (irq_pin (16 ),249 ), -PINMUX_IRQ (irq_pin (14 ),251 ), -PINMUX_IRQ (irq_pin (9 ),308 ), +PINMUX_IRQ (irq_pin (29 ),28 ), +PINMUX_IRQ (irq_pin (30 ),27 ), +PINMUX_IRQ (irq_pin (31 ),26 ), };/* -----------------------------------------------------------------------------
Signature · 1 line
1.8.3.2
PATCH 4/10 sh-pfc: sh73a0: Sort IRQ entries by IRQ number
Quoted 4 lines · Laurent Pinchart
This makes catching duplicate entries easier. Merge the two IRQ9 entries found after sorting.
Signed-off-by Laurent Pinchart <[email protected]>
Patch applied.
Yours,
Linus Walleij
PATCH 5/10 sh-pfc: Rename sh_pfc window field to windows
In reply to Laurent PinchartThere's more than one window, name the field windows.
--- drivers/pinctrl/sh-pfc/core.c | 19 ++++++++++--------- drivers/pinctrl/sh-pfc/core.h | 2 +- drivers/pinctrl/sh-pfc/gpio.c | 4 ++-- drivers/pinctrl/sh-pfc/pfc-r8a73a4.c | 4 ++-- drivers/pinctrl/sh-pfc/pfc-r8a7740.c | 2 +- drivers/pinctrl/sh-pfc/pfc-sh7372.c | 2 +- drivers/pinctrl/sh-pfc/pfc-sh73a0.c | 8 ++++---- 7 files changed, 21 insertions(+), 20 deletions(-)
drivers/pinctrl/sh-pfc/core.c +10 -9
@@ -34,20 +34,21 @@ static int sh_pfc_ioremap(struct sh_pfc *pfc, struct platform_device *pdev)if (pdev->num_resources ==0 )return -EINVAL ; - pfc->window =devm_kzalloc (pfc->dev, pdev->num_resources * -sizeof (*pfc->window),GFP_NOWAIT ); -if (!pfc->window) + pfc->windows =devm_kzalloc (pfc->dev, pdev->num_resources * +sizeof (*pfc->windows),GFP_NOWAIT ); +if (!pfc->windows)return -ENOMEM ; pfc->num_windows = pdev->num_resources;for (k =0 , res = pdev->resource; k < pdev->num_resources; k++, res++) {WARN_ON (resource_type (res) !=IORESOURCE_MEM ); - pfc->window[k].phys = res->start; - pfc->window[k].size =resource_size (res); - pfc->window[k].virt =devm_ioremap_nocache (pfc->dev, res->start, -resource_size (res)); -if (!pfc->window[k].virt) + pfc->windows[k].phys = res->start; + pfc->windows[k].size =resource_size (res); + pfc->windows[k].virt = +devm_ioremap_nocache (pfc->dev, res->start, +resource_size (res)); +if (!pfc->windows[k].virt)return -ENOMEM ; }
@@ -62,7 +63,7 @@ static void __iomem *sh_pfc_phys_to_virt(struct sh_pfc *pfc,/* scan through physical windows and convert address */ for (i =0 ; i < pfc->num_windows; i++) { - window = pfc->window + i; + window = pfc->windows + i;if (address < window->phys)continue ;
drivers/pinctrl/sh-pfc/core.h +1 -1
@@ -37,7 +37,7 @@ struct sh_pfc {spinlock_t lock;unsigned int num_windows; -struct sh_pfc_window *window; +struct sh_pfc_window *windows;struct sh_pfc_pin_range *ranges;unsigned int nr_ranges;
drivers/pinctrl/sh-pfc/gpio.c +2 -2
@@ -347,7 +347,7 @@ int sh_pfc_register_gpiochip(struct sh_pfc *pfc)* GPIOs. */ for (i =0 ; i < pfc->num_windows; ++i) { -struct sh_pfc_window *window = &pfc->window[i]; +struct sh_pfc_window *window = &pfc->windows[i];if (pfc->info->data_regs[0 ].reg >= window->phys && pfc->info->data_regs[0 ].reg < window->phys + window->size)
@@ -358,7 +358,7 @@ int sh_pfc_register_gpiochip(struct sh_pfc *pfc)return 0 ;/* Register the real GPIOs chip. */ - chip =sh_pfc_add_gpiochip (pfc, gpio_pin_setup, &pfc->window[i]); + chip =sh_pfc_add_gpiochip (pfc, gpio_pin_setup, &pfc->windows[i]);if (IS_ERR (chip))return PTR_ERR (chip);
drivers/pinctrl/sh-pfc/pfc-r8a73a4.c +2 -2
@@ -2691,7 +2691,7 @@ static unsigned int r8a73a4_pinmux_get_bias(struct sh_pfc *pfc, {void __iomem *addr; - addr = pfc->window->virt + r8a73a4_portcr_offsets[pin >>5 ] + pin; + addr = pfc->windows->virt + r8a73a4_portcr_offsets[pin >>5 ] + pin;switch (ioread8 (addr) &PORTCR_PULMD_MASK ) {case PORTCR_PULMD_UP :
@@ -2710,7 +2710,7 @@ static void r8a73a4_pinmux_set_bias(struct sh_pfc *pfc, unsigned int pin,void __iomem *addr;u32 value; - addr = pfc->window->virt + r8a73a4_portcr_offsets[pin >>5 ] + pin; + addr = pfc->windows->virt + r8a73a4_portcr_offsets[pin >>5 ] + pin; value =ioread8 (addr) & ~PORTCR_PULMD_MASK ;switch (bias) {
drivers/pinctrl/sh-pfc/pfc-r8a7740.c +1 -1
@@ -3721,7 +3721,7 @@ static void __iomem *r8a7740_pinmux_portcr(struct sh_pfc *pfc, unsigned int pin) &r8a7740_portcr_offsets[i];if (i <= group->end_pin) -return pfc->window->virt + group->offset + pin; +return pfc->windows->virt + group->offset + pin; }return NULL ;
drivers/pinctrl/sh-pfc/pfc-sh7372.c +1 -1
@@ -2585,7 +2585,7 @@ static void __iomem *sh7372_pinmux_portcr(struct sh_pfc *pfc, unsigned int pin) &sh7372_portcr_offsets[i];if (i <= group->end_pin) -return pfc->window->virt + group->offset + pin; +return pfc->windows->virt + group->offset + pin; }return NULL ;
drivers/pinctrl/sh-pfc/pfc-sh73a0.c +4 -4
@@ -3702,7 +3702,7 @@ static const struct pinmux_irq pinmux_irqs[] = {static void sh73a0_vccq_mc0_endisable (struct regulator_dev *reg,bool enable) {struct sh_pfc *pfc = reg->reg_data; -void __iomem *addr = pfc->window[1 ].virt +4 ; +void __iomem *addr = pfc->windows[1 ].virt +4 ;unsigned long flags;u32 value;
@@ -3735,7 +3735,7 @@ static int sh73a0_vccq_mc0_disable(struct regulator_dev *reg)static int sh73a0_vccq_mc0_is_enabled (struct regulator_dev *reg) {struct sh_pfc *pfc = reg->reg_data; -void __iomem *addr = pfc->window[1 ].virt +4 ; +void __iomem *addr = pfc->windows[1 ].virt +4 ;unsigned long flags;u32 value;
@@ -3794,7 +3794,7 @@ static const unsigned int sh73a0_portcr_offsets[] = {static unsigned int sh73a0_pinmux_get_bias (struct sh_pfc *pfc,unsigned int pin) { -void __iomem *addr = pfc->window->virt +void __iomem *addr = pfc->windows->virt + sh73a0_portcr_offsets[pin >>5 ] + pin;u32 value =ioread8 (addr) & PORTnCR_PULMD_MASK;
@@ -3812,7 +3812,7 @@ static unsigned int sh73a0_pinmux_get_bias(struct sh_pfc *pfc, unsigned int pin)static void sh73a0_pinmux_set_bias (struct sh_pfc *pfc,unsigned int pin,unsigned int bias) { -void __iomem *addr = pfc->window->virt +void __iomem *addr = pfc->windows->virt + sh73a0_portcr_offsets[pin >>5 ] + pin;u32 value =ioread8 (addr) & ~PORTnCR_PULMD_MASK;
Signature · 1 line
1.8.3.2
PATCH 5/10 sh-pfc: Rename sh_pfc window field to windows
Quoted 3 lines · Laurent Pinchart
There's more than one window, name the field windows.
Signed-off-by Laurent Pinchart <[email protected]>
Patch applied.
Yours,
Linus Walleij
PATCH 6/10 sh-pfc: Support GPIO to IRQ mapping specified IRQ resources
In reply to Laurent PinchartOn non-DT platforms IRQ controllers associated with the GPIOs have a fixed IRQ base value known at compile time. The sh-pfc driver translates GPIO number to IRQ numbers using a hardcoded table. This mechanism breaks on DT platforms, as the IRQ base values are dynamic in that case.
Fix this by specifying IRQs associated with GPIOs in IRQ resources, populated automatically from the device tree. When IRQ resources are specified the driver requires one IRQ resource per GPIO able to generate an interrupt, and uses the translation table to compute the IRQ resource offset instead of the IRQ number.
--- .../bindings/pinctrl/renesas,pfc-pinctrl.txt | 14 +++++ drivers/pinctrl/sh-pfc/core.c | 69 +++++++++++++++++----- drivers/pinctrl/sh-pfc/core.h | 2 + drivers/pinctrl/sh-pfc/gpio.c | 14 ++++- 4 files changed, 82 insertions(+), 17 deletions(-)
Documentation/devicetree/bindings/pinctrl/renesas,pfc-pinctrl.txt +14 -0
@@ -26,6 +26,11 @@ Optional properties: - #gpio-range-cells: Mandatory when the PFC doesn't handle GPIO, forbidden otherwise. Should be 3. + - interrupts-extended: Specify the interrupts associated with external + IRQ pins. This property is mandatory when the PFC handles GPIOs and + forbidden otherwise. When specified, it must contain one interrupt per + external IRQ, sorted by external IRQ number. + The PFC node also acts as a container for pin configuration nodes. Please refer to pinctrl-bindings.txt in this directory for the definition of the term "pin configuration node" and for the common pinctrl bindings used by client devices.
@@ -103,6 +108,15 @@ Example 1: SH73A0 (SH-Mobile AG5) pin controller node <0xe605801c 0x1c>; gpio-controller; #gpio-cells = <2>; + interrupts-extended = + <&irqpin0 0 0>, <&irqpin0 1 0>, <&irqpin0 2 0>, <&irqpin0 3 0>, + <&irqpin0 4 0>, <&irqpin0 5 0>, <&irqpin0 6 0>, <&irqpin0 7 0>, + <&irqpin1 0 0>, <&irqpin1 1 0>, <&irqpin1 2 0>, <&irqpin1 3 0>, + <&irqpin1 4 0>, <&irqpin1 5 0>, <&irqpin1 6 0>, <&irqpin1 7 0>, + <&irqpin2 0 0>, <&irqpin2 1 0>, <&irqpin2 2 0>, <&irqpin2 3 0>, + <&irqpin2 4 0>, <&irqpin2 5 0>, <&irqpin2 6 0>, <&irqpin2 7 0>, + <&irqpin3 0 0>, <&irqpin3 1 0>, <&irqpin3 2 0>, <&irqpin3 3 0>, + <&irqpin3 4 0>, <&irqpin3 5 0>, <&irqpin3 6 0>, <&irqpin3 7 0>; }; Example 2: A GPIO LED node that references a GPIO
drivers/pinctrl/sh-pfc/core.c +53 -16
90 lines not on this page · raw message
drivers/pinctrl/sh-pfc/core.h +2 -0
@@ -38,6 +38,8 @@ struct sh_pfc {unsigned int num_windows;struct sh_pfc_window *windows; +unsigned int num_irqs; +unsigned int *irqs;struct sh_pfc_pin_range *ranges;unsigned int nr_ranges;
drivers/pinctrl/sh-pfc/gpio.c +13 -1
@@ -211,11 +211,17 @@ static int gpio_pin_to_irq(struct gpio_chip *gc, unsigned offset)for (k =0 ; gpios[k] >=0 ; k++) {if (gpios[k] == offset) -return pfc->info->gpio_irq[i].irq; + goto found; } }return -ENOSYS ; + +found: +if (pfc->num_irqs) +return pfc->irqs[i]; +else +return pfc->info->gpio_irq[i].irq; }static int gpio_pin_setup (struct sh_pfc_chip *chip)
@@ -357,6 +363,12 @@ int sh_pfc_register_gpiochip(struct sh_pfc *pfc)if (i == pfc->num_windows)return 0 ; +/* If we have IRQ resources make sure their number is correct. */ +if (pfc->num_irqs && pfc->num_irqs != pfc->info->gpio_irq_size) { +dev_err (pfc->dev,"invalid number of IRQ resources\n" ); +return -EINVAL ; + } +/* Register the real GPIOs chip. */ chip =sh_pfc_add_gpiochip (pfc, gpio_pin_setup, &pfc->windows[i]);if (IS_ERR (chip))
Signature · 1 line
1.8.3.2
PATCH 6/10 sh-pfc: Support GPIO to IRQ mapping specified IRQ resources
Quoted 5 lines · Laurent Pinchart
On non-DT platforms IRQ controllers associated with the GPIOs have a fixed IRQ base value known at compile time. The sh-pfc driver translates GPIO number to IRQ numbers using a hardcoded table. This mechanism breaks on DT platforms, as the IRQ base values are dynamic in that case.
Quoted 8 lines
Fix this by specifying IRQs associated with GPIOs in IRQ resources, populated automatically from the device tree. When IRQ resources are specified the driver requires one IRQ resource per GPIO able to generate an interrupt, and uses the translation table to compute the IRQ resource offset instead of the IRQ number.
Cc devicetree at vger.kernel.orgSigned-off-by Laurent Pinchart <[email protected]>
Make things much better and only use standard bindings so patch applied.
Yours,
Linus Walleij
PATCH 7/10 ARM: shmobile: r8a73a4: Specify PFC interrupts in DT
In reply to Laurent Pinchart--- arch/arm/boot/dts/r8a73a4.dtsi | 16 ++++++++++++++++ 1 file changed, 16 insertions(+)
arch/arm/boot/dts/r8a73a4.dtsi +16 -0
@@ -288,6 +288,22 @@ reg = <0 0xe6050000 0 0x9000>; gpio-controller; #gpio-cells = <2>; + interrupts-extended = + <&irqc0 0 0>, <&irqc0 1 0>, <&irqc0 2 0>, <&irqc0 3 0>, + <&irqc0 4 0>, <&irqc0 5 0>, <&irqc0 6 0>, <&irqc0 7 0>, + <&irqc0 8 0>, <&irqc0 9 0>, <&irqc0 10 0>, <&irqc0 11 0>, + <&irqc0 12 0>, <&irqc0 13 0>, <&irqc0 14 0>, <&irqc0 15 0>, + <&irqc0 16 0>, <&irqc0 17 0>, <&irqc0 18 0>, <&irqc0 19 0>, + <&irqc0 20 0>, <&irqc0 21 0>, <&irqc0 22 0>, <&irqc0 23 0>, + <&irqc0 24 0>, <&irqc0 25 0>, <&irqc0 26 0>, <&irqc0 27 0>, + <&irqc0 28 0>, <&irqc0 29 0>, <&irqc0 30 0>, <&irqc0 31 0>, + <&irqc1 0 0>, <&irqc1 1 0>, <&irqc1 2 0>, <&irqc1 3 0>, + <&irqc1 4 0>, <&irqc1 5 0>, <&irqc1 6 0>, <&irqc1 7 0>, + <&irqc1 8 0>, <&irqc1 9 0>, <&irqc1 10 0>, <&irqc1 11 0>, + <&irqc1 12 0>, <&irqc1 13 0>, <&irqc1 14 0>, <&irqc1 15 0>, + <&irqc1 16 0>, <&irqc1 17 0>, <&irqc1 18 0>, <&irqc1 19 0>, + <&irqc1 20 0>, <&irqc1 21 0>, <&irqc1 22 0>, <&irqc1 23 0>, + <&irqc1 24 0>, <&irqc1 25 0>; }; sdhi0: sd at ee100000 {
Signature · 1 line
1.8.3.2
PATCH 7/10 ARM: shmobile: r8a73a4: Specify PFC interrupts in DT
Quoted 1 line · Laurent Pinchart
Signed-off-by Laurent Pinchart <[email protected]>
for patches 7 thru 10.
I expect these to go in through the shmobile tree.
Yours,
Linus Walleij
PATCH 8/10 ARM: shmobile: r8a7740: Specify PFC interrupts in DT
In reply to Laurent Pinchart--- arch/arm/boot/dts/r8a7740.dtsi | 9 +++++++++ 1 file changed, 9 insertions(+)
arch/arm/boot/dts/r8a7740.dtsi +9 -0
@@ -155,6 +155,15 @@ <0xe605800c 0x20>; gpio-controller; #gpio-cells = <2>; + interrupts-extended = + <&irqpin0 0 0>, <&irqpin0 1 0>, <&irqpin0 2 0>, <&irqpin0 3 0>, + <&irqpin0 4 0>, <&irqpin0 5 0>, <&irqpin0 6 0>, <&irqpin0 7 0>, + <&irqpin1 0 0>, <&irqpin1 1 0>, <&irqpin1 2 0>, <&irqpin1 3 0>, + <&irqpin1 4 0>, <&irqpin1 5 0>, <&irqpin1 6 0>, <&irqpin1 7 0>, + <&irqpin2 0 0>, <&irqpin2 1 0>, <&irqpin2 2 0>, <&irqpin2 3 0>, + <&irqpin2 4 0>, <&irqpin2 5 0>, <&irqpin2 6 0>, <&irqpin2 7 0>, + <&irqpin3 0 0>, <&irqpin3 1 0>, <&irqpin3 2 0>, <&irqpin3 3 0>, + <&irqpin3 4 0>, <&irqpin3 5 0>, <&irqpin3 6 0>, <&irqpin3 7 0>; }; tpu: pwm at e6600000 {
Signature · 1 line
1.8.3.2
PATCH 9/10 ARM: shmobile: sh73a0: Specify PFC interrupts in DT
In reply to Laurent Pinchart--- arch/arm/boot/dts/sh73a0.dtsi | 9 +++++++++ 1 file changed, 9 insertions(+)
arch/arm/boot/dts/sh73a0.dtsi +9 -0
@@ -242,5 +242,14 @@ <0xe605801c 0x1c>; gpio-controller; #gpio-cells = <2>; + interrupts-extended = + <&irqpin0 0 0>, <&irqpin0 1 0>, <&irqpin0 2 0>, <&irqpin0 3 0>, + <&irqpin0 4 0>, <&irqpin0 5 0>, <&irqpin0 6 0>, <&irqpin0 7 0>, + <&irqpin1 0 0>, <&irqpin1 1 0>, <&irqpin1 2 0>, <&irqpin1 3 0>, + <&irqpin1 4 0>, <&irqpin1 5 0>, <&irqpin1 6 0>, <&irqpin1 7 0>, + <&irqpin2 0 0>, <&irqpin2 1 0>, <&irqpin2 2 0>, <&irqpin2 3 0>, + <&irqpin2 4 0>, <&irqpin2 5 0>, <&irqpin2 6 0>, <&irqpin2 7 0>, + <&irqpin3 0 0>, <&irqpin3 1 0>, <&irqpin3 2 0>, <&irqpin3 3 0>, + <&irqpin3 4 0>, <&irqpin3 5 0>, <&irqpin3 6 0>, <&irqpin3 7 0>; }; };
Signature · 1 line
1.8.3.2
PATCH 10/10 ARM: shmobile: armadillo: dts: Add gpio-keys device
In reply to Laurent PinchartThe board had 4 buttons connected to GPIOs, add a corresponding gpio-keys device.
--- .../boot/dts/r8a7740-armadillo800eva-reference.dts | 29 ++++++++++++++++++++++ 1 file changed, 29 insertions(+)
arch/arm/boot/dts/r8a7740-armadillo800eva-reference.dts +29 -0
@@ -63,6 +63,35 @@ enable-active-high; }; + gpio-keys { + compatible = "gpio-keys"; + + power-key { + gpios = <&pfc 99 GPIO_ACTIVE_LOW>; + linux,code = <116>; + label = "SW3"; + gpio-key,wakeup; + }; + + back-key { + gpios = <&pfc 100 GPIO_ACTIVE_LOW>; + linux,code = <158>; + label = "SW4"; + }; + + menu-key { + gpios = <&pfc 97 GPIO_ACTIVE_LOW>; + linux,code = <139>; + label = "SW5"; + }; + + home-key { + gpios = <&pfc 98 GPIO_ACTIVE_LOW>; + linux,code = <102>; + label = "SW6"; + }; + }; + leds { compatible = "gpio-leds"; led1 {
Signature · 1 line
1.8.3.2
cover sh-pfc: Fix GPIO to IRQ mapping
In reply to Laurent PinchartQuoted 29 lines · Laurent Pinchart
Hello,
This patch set fixes conversion from a GPIO number to an IRQ number in the sh-pfc driver in the device tree case.
Patches 01/10 to 05/10 clean up the driver to prepare it for the core fix in patch 06/10. As explained in patch 06/10 commit message, on non-DT platforms IRQ controllers associated with the GPIOs have a fixed IRQ base value known at compile time. The sh-pfc driver translates GPIO number to IRQ numbers using a hardcoded table. This mechanism breaks on DT platforms, as the IRQ base values are dynamic in that case.
The patch set fixes this by specifying IRQs associated with GPIOs in IRQ resources, populated automatically from the device tree. When IRQ resources are specified the driver requires one IRQ resource per GPIO able to generate an interrupt, and uses the translation table to compute the IRQ resource offset instead of the IRQ number.
Patches 07/10 to 09/10 then add PFC interrupts in DT for the SH73A0, R8A73A4 and R9A7740 SoCs. The only missing platform here is SH7372 due to its very uncomplete DT support. All other Renesas ARM platforms use a standaline GPIO driver and are thus not affected.
Patch 10/10 finally adds a gpio-keys device to the Armadillo board DTS, allowing the whole patch set to the tested.
The sh-pfc changes need to be merged first. Linus, could you please provide a stable branch with the first 6 patches when you will apply them to your tree ?
Quoted 11 lines
Laurent Pinchart (10):
sh-pfc: Turn unsigned indices into unsigned int
sh-pfc: Terminate gpios array by -1
sh-pfc: sh73a0: Add missing IRQ15
sh-pfc: sh73a0: Sort IRQ entries by IRQ number
sh-pfc: Renamesh_pfcwindow field to windows
sh-pfc: Support GPIO to IRQ mapping specified IRQ resources
ARM: shmobile: r8a73a4: Specify PFC interrupts in DT
ARM: shmobile: r8a7740: Specify PFC interrupts in DT
ARM: shmobile: sh73a0: Specify PFC interrupts in DT
ARM: shmobile: armadillo: dts: Add gpio-keys device
Looking great, nice to see that this is getting resolved! For the entire series:
cover sh-pfc: Fix GPIO to IRQ mapping
In reply to Laurent PinchartQuoted 18 lines · Laurent Pinchart
Hello,
This patch set fixes conversion from a GPIO number to an IRQ number in the sh-pfc driver in the device tree case.
Patches 01/10 to 05/10 clean up the driver to prepare it for the core fix in patch 06/10. As explained in patch 06/10 commit message, on non-DT platforms IRQ controllers associated with the GPIOs have a fixed IRQ base value known at compile time. The sh-pfc driver translates GPIO number to IRQ numbers using a hardcoded table. This mechanism breaks on DT platforms, as the IRQ base values are dynamic in that case.
The patch set fixes this by specifying IRQs associated with GPIOs in IRQ resources, populated automatically from the device tree. When IRQ resources are specified the driver requires one IRQ resource per GPIO able to generate an interrupt, and uses the translation table to compute the IRQ resource offset instead of the IRQ number.
Quoted 4 lines
Patches 07/10 to 09/10 then add PFC interrupts in DT for the SH73A0, R8A73A4 and R9A7740 SoCs. The only missing platform here is SH7372 due to its very uncomplete DT support. All other Renesas ARM platforms use a standaline GPIO driver and are thus not affected.
s/uncomplete/incomplete/
Quoted 5 lines
Patch 10/10 finally adds a gpio-keys device to the Armadillo board DTS, allowing the whole patch set to the tested.
The sh-pfc changes need to be merged first. Linus, could you please provide a stable branch with the first 6 patches when you will apply them to your tree ?
It seems to me that if Linus applies the pfc changes the shmobile changes could be applied independently. Am I missing and do the shmobile patches have build-dependencies on the pfc patches?
Quoted 38 lines
Laurent Pinchart (10):
sh-pfc: Turn unsigned indices into unsigned int
sh-pfc: Terminate gpios array by -1
sh-pfc: sh73a0: Add missing IRQ15
sh-pfc: sh73a0: Sort IRQ entries by IRQ number
sh-pfc: Renamesh_pfcwindow field to windows
sh-pfc: Support GPIO to IRQ mapping specified IRQ resources
ARM: shmobile: r8a73a4: Specify PFC interrupts in DT
ARM: shmobile: r8a7740: Specify PFC interrupts in DT
ARM: shmobile: sh73a0: Specify PFC interrupts in DT
ARM: shmobile: armadillo: dts: Add gpio-keys device.../bindings/pinctrl/renesas,pfc-pinctrl.txt | 14 ++++ arch/arm/boot/dts/r8a73a4.dtsi | 16 +++++ .../boot/dts/r8a7740-armadillo800eva-reference.dts | 29 +++++++++ arch/arm/boot/dts/r8a7740.dtsi | 9 +++ arch/arm/boot/dts/sh73a0.dtsi | 9 +++ drivers/pinctrl/sh-pfc/core.c | 76 ++++++++++++++++------ drivers/pinctrl/sh-pfc/core.h | 4 +- drivers/pinctrl/sh-pfc/gpio.c | 24 +++++-- drivers/pinctrl/sh-pfc/pfc-r8a73a4.c | 4 +- drivers/pinctrl/sh-pfc/pfc-r8a7740.c | 2 +- drivers/pinctrl/sh-pfc/pfc-sh7372.c | 2 +- drivers/pinctrl/sh-pfc/pfc-sh73a0.c | 54 +++++++-------- drivers/pinctrl/sh-pfc/sh_pfc.h | 4 +- 13 files changed, 188 insertions(+), 59 deletions(-)Signature · 8 lines
Regards, Laurent Pinchart -- To unsubscribe from this list: send the line "unsubscribe linux-sh" in the body of a message to majordomo at vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
cover sh-pfc: Fix GPIO to IRQ mapping
Hi Simon,
Quoted 30 lines · On Thursday 12 December 2013 22:11:39 Simon Horman wrote:
Quoted 19 lines · Laurent Pinchart
Hello,
This patch set fixes conversion from a GPIO number to an IRQ number in the sh-pfc driver in the device tree case.
Patches 01/10 to 05/10 clean up the driver to prepare it for the core fix in patch 06/10. As explained in patch 06/10 commit message, on non-DT platforms IRQ controllers associated with the GPIOs have a fixed IRQ base value known at compile time. The sh-pfc driver translates GPIO number to IRQ numbers using a hardcoded table. This mechanism breaks on DT platforms, as the IRQ base values are dynamic in that case.
The patch set fixes this by specifying IRQs associated with GPIOs in IRQ resources, populated automatically from the device tree. When IRQ resources
are specified the driver requires one IRQ resource per GPIO able to generate an interrupt, and uses the translation table to compute the IRQ resource offset instead of the IRQ number.Quoted 4 lines
Patches 07/10 to 09/10 then add PFC interrupts in DT for the SH73A0, R8A73A4 and R9A7740 SoCs. The only missing platform here is SH7372 due to its very uncomplete DT support. All other Renesas ARM platforms use a standaline GPIO driver and are thus not affected.
s/uncomplete/incomplete/
Quoted 3 lines
Patch 10/10 finally adds a gpio-keys device to the Armadillo board DTS, allowing the whole patch set to the tested.
Quoted 7 lines
Quoted 3 lines
The sh-pfc changes need to be merged first. Linus, could you please provide a stable branch with the first 6 patches when you will apply them to your tree ?
It seems to me that if Linus applies the pfc changes the shmobile changes could be applied independently. Am I missing and do the shmobile patches have build-dependencies on the pfc patches?
The DTS patches will result in the PFC platform device having IRQ resources. The sh-pfc driver currently supports MEM resources only and will crash.
Quoted 26 lines
Quoted 26 lines
Laurent Pinchart (10):
sh-pfc: Turn unsigned indices into unsigned int
sh-pfc: Terminate gpios array by -1
sh-pfc: sh73a0: Add missing IRQ15
sh-pfc: sh73a0: Sort IRQ entries by IRQ number
sh-pfc: Renamesh_pfcwindow field to windows
sh-pfc: Support GPIO to IRQ mapping specified IRQ resources
ARM: shmobile: r8a73a4: Specify PFC interrupts in DT
ARM: shmobile: r8a7740: Specify PFC interrupts in DT
ARM: shmobile: sh73a0: Specify PFC interrupts in DT
ARM: shmobile: armadillo: dts: Add gpio-keys device.../bindings/pinctrl/renesas,pfc-pinctrl.txt | 14 ++++ arch/arm/boot/dts/r8a73a4.dtsi | 16 +++++ .../boot/dts/r8a7740-armadillo800eva-reference.dts | 29 +++++++++ arch/arm/boot/dts/r8a7740.dtsi | 9 +++ arch/arm/boot/dts/sh73a0.dtsi | 9 +++ drivers/pinctrl/sh-pfc/core.c | 76 +++++++++++------ drivers/pinctrl/sh-pfc/core.h | 4 +- drivers/pinctrl/sh-pfc/gpio.c | 24 +++++-- drivers/pinctrl/sh-pfc/pfc-r8a73a4.c | 4 +- drivers/pinctrl/sh-pfc/pfc-r8a7740.c | 2 +- drivers/pinctrl/sh-pfc/pfc-sh7372.c | 2 +- drivers/pinctrl/sh-pfc/pfc-sh73a0.c | 54 +++++++-------- drivers/pinctrl/sh-pfc/sh_pfc.h | 4 +- 13 files changed, 188 insertions(+), 59 deletions(-)
Signature · 3 lines
Regards, Laurent Pinchart
cover sh-pfc: Fix GPIO to IRQ mapping
Quoted 37 lines · Laurent Pinchart
Hi Simon,
Quoted 34 lines · On Thursday 12 December 2013 22:11:39 Simon Horman wrote:
Quoted 19 lines · Laurent Pinchart
Hello,
This patch set fixes conversion from a GPIO number to an IRQ number in the sh-pfc driver in the device tree case.
Patches 01/10 to 05/10 clean up the driver to prepare it for the core fix in patch 06/10. As explained in patch 06/10 commit message, on non-DT platforms IRQ controllers associated with the GPIOs have a fixed IRQ base value known at compile time. The sh-pfc driver translates GPIO number to IRQ numbers using a hardcoded table. This mechanism breaks on DT platforms, as the IRQ base values are dynamic in that case.
The patch set fixes this by specifying IRQs associated with GPIOs in IRQ resources, populated automatically from the device tree. When IRQ resources
are specified the driver requires one IRQ resource per GPIO able to generate an interrupt, and uses the translation table to compute the IRQ resource offset instead of the IRQ number.Quoted 4 lines
Patches 07/10 to 09/10 then add PFC interrupts in DT for the SH73A0, R8A73A4 and R9A7740 SoCs. The only missing platform here is SH7372 due to its very uncomplete DT support. All other Renesas ARM platforms use a standaline GPIO driver and are thus not affected.
s/uncomplete/incomplete/
Quoted 6 lines
Patch 10/10 finally adds a gpio-keys device to the Armadillo board DTS, allowing the whole patch set to the tested.
The sh-pfc changes need to be merged first. Linus, could you please provide a stable branch with the first 6 patches when you will apply them to your tree ?
Quoted 6 lines
Quoted 3 lines
It seems to me that if Linus applies the pfc changes the shmobile changes could be applied independently. Am I missing and do the shmobile patches have build-dependencies on the pfc patches?
The DTS patches will result in the PFC platform device having IRQ resources. The sh-pfc driver currently supports MEM resources only and will crash.
Thanks, got it :)
cover sh-pfc: Fix GPIO to IRQ mapping
In reply to Simon HormanQuoted 3 lines · Simon Horman
It seems to me that if Linus applies the pfc changes the shmobile changes could be applied independently. Am I missing and do the shmobile patches have build-dependencies on the pfc patches?
I've applied the PFC changes to the pin control tree and will push to kernelorg shortly, then allow it to stablilze before you attempt to pull that into the SH mobile tree please, and notify me before doing so and make sure to just pick the top commit you need.
Yours,
Linus Walleij
cover sh-pfc: Fix GPIO to IRQ mapping
Quoted 10 lines · Linus Walleij
Quoted 3 lines · Simon Horman
It seems to me that if Linus applies the pfc changes the shmobile changes could be applied independently. Am I missing and do the shmobile patches have build-dependencies on the pfc patches?
I've applied the PFC changes to the pin control tree and will push to kernelorg shortly, then allow it to stablilze before you attempt to pull that into the SH mobile tree please, and notify me before doing so and make sure to just pick the top commit you need.
Thanks, I understand.
I'll re-visit this next week.
cover sh-pfc: Fix GPIO to IRQ mapping
Quoted 7 lines · Simon Horman
Quoted 6 lines · Linus Walleij
Quoted 3 lines · Simon Horman
It seems to me that if Linus applies the pfc changes the shmobile changes could be applied independently. Am I missing and do the shmobile patches have build-dependencies on the pfc patches?
Quoted 8 lines
Quoted 4 lines
I've applied the PFC changes to the pin control tree and will push to kernelorg shortly, then allow it to stablilze before you attempt to pull that into the SH mobile tree please, and notify me before doing so and make sure to just pick the top commit you need.
Thanks, I understand.
I'll re-visit this next week.
Hi Linus,
just to let you know I am planning to use70c8f01a357ac74d ("sh-pfc: Support GPIO to IRQ mapping specified IRQ resources")
cover sh-pfc: Fix GPIO to IRQ mapping
Quoted 2 lines · Simon Horman
just to let you know I am planning to use
70c8f01a357ac74d("sh-pfc: Support GPIO to IRQ mapping specified IRQ resources")
OK go ahead and pull that to your tree.
Yours,
Linus Walleij
cover sh-pfc: Fix GPIO to IRQ mapping
Quoted 6 lines · Linus Walleij
Quoted 2 lines · Simon Horman
just to let you know I am planning to use
70c8f01a357ac74d("sh-pfc: Support GPIO to IRQ mapping specified IRQ resources")OK go ahead and pull that to your tree.
Thanks, done.