drivers/gpio/gpio-omap.c | 26 ++++++++++++++++++++++++++ 1 file changed, 26 insertions(+)
From: Bin Liu <b-liu@ti.com>
Add omap_gpio_disable_irq and omap_gpio_enable_irq
calls in gpio-omap.
Currently, kernel cannot disable gpio interrupts in
case of a irq storm, so add omap_gpio_disable/enable_irq
so that interrupts can be disabled/enabled.
Signed-off-by: Bin Liu <b-liu@ti.com>
[Judith: Add commit message]
Signed-off-by: Judith Mendez <jm@ti.com>
---
Changes since v2:
- Fix email header and commit message
Changes since v1 RESEND:
- split patch from series [0]
- Add disable/enable calls without wrapper functions
[0] https://lore.kernel.org/linux-omap/20241011173356.870883-1-jm@ti.com/
Tested on am335x BeagleBone Black
---
drivers/gpio/gpio-omap.c | 26 ++++++++++++++++++++++++++
1 file changed, 26 insertions(+)
diff --git a/drivers/gpio/gpio-omap.c b/drivers/gpio/gpio-omap.c
index 76d5d87e9681e..137aabada26f9 100644
--- a/drivers/gpio/gpio-omap.c
+++ b/drivers/gpio/gpio-omap.c
@@ -711,6 +711,28 @@ static void omap_gpio_unmask_irq(struct irq_data *d)
raw_spin_unlock_irqrestore(&bank->lock, flags);
}
+static void omap_gpio_disable_irq(struct irq_data *d)
+{
+ struct gpio_bank *bank = omap_irq_data_get_bank(d);
+ unsigned int offset = d->hwirq;
+ unsigned long flags;
+
+ raw_spin_lock_irqsave(&bank->lock, flags);
+ omap_set_gpio_irqenable(bank, offset, 0);
+ raw_spin_unlock_irqrestore(&bank->lock, flags);
+}
+
+static void omap_gpio_enable_irq(struct irq_data *d)
+{
+ struct gpio_bank *bank = omap_irq_data_get_bank(d);
+ unsigned int offset = d->hwirq;
+ unsigned long flags;
+
+ raw_spin_lock_irqsave(&bank->lock, flags);
+ omap_set_gpio_irqenable(bank, offset, 1);
+ raw_spin_unlock_irqrestore(&bank->lock, flags);
+}
+
static void omap_gpio_irq_print_chip(struct irq_data *d, struct seq_file *p)
{
struct gpio_bank *bank = omap_irq_data_get_bank(d);
@@ -723,6 +745,8 @@ static const struct irq_chip omap_gpio_irq_chip = {
.irq_shutdown = omap_gpio_irq_shutdown,
.irq_mask = omap_gpio_mask_irq,
.irq_unmask = omap_gpio_unmask_irq,
+ .irq_disable = omap_gpio_disable_irq,
+ .irq_enable = omap_gpio_enable_irq,
.irq_set_type = omap_gpio_irq_type,
.irq_set_wake = omap_gpio_wake_enable,
.irq_bus_lock = omap_gpio_irq_bus_lock,
@@ -737,6 +761,8 @@ static const struct irq_chip omap_gpio_irq_chip_nowake = {
.irq_shutdown = omap_gpio_irq_shutdown,
.irq_mask = omap_gpio_mask_irq,
.irq_unmask = omap_gpio_unmask_irq,
+ .irq_disable = omap_gpio_disable_irq,
+ .irq_enable = omap_gpio_enable_irq,
.irq_set_type = omap_gpio_irq_type,
.irq_bus_lock = omap_gpio_irq_bus_lock,
.irq_bus_sync_unlock = gpio_irq_bus_sync_unlock,
--
2.47.0Judith Mendez <jm@ti.com> writes: > From: Bin Liu <b-liu@ti.com> > > Add omap_gpio_disable_irq and omap_gpio_enable_irq > calls in gpio-omap. > > Currently, kernel cannot disable gpio interrupts in > case of a irq storm, so add omap_gpio_disable/enable_irq > so that interrupts can be disabled/enabled. > > Signed-off-by: Bin Liu <b-liu@ti.com> > [Judith: Add commit message] > Signed-off-by: Judith Mendez <jm@ti.com> Reviewed-by: Kevin Hilman <khilman@baylibre.com>
Hi Judith, Judith Mendez <jm@ti.com> writes: > From: Bin Liu <b-liu@ti.com> > > Add omap_gpio_disable_irq and omap_gpio_enable_irq > calls in gpio-omap. > > Currently, kernel cannot disable gpio interrupts in > case of a irq storm, so add omap_gpio_disable/enable_irq > so that interrupts can be disabled/enabled. > > Signed-off-by: Bin Liu <b-liu@ti.com> > [Judith: Add commit message] > Signed-off-by: Judith Mendez <jm@ti.com> Thanks for this patch. Can you give a bit more context on the problem(s) this solves and on which SoCs/platforms it was developed/validated? Thanks, Kevin
Hi Kevin, On 11/1/24 9:29 AM, Kevin Hilman wrote: > Hi Judith, > > Judith Mendez <jm@ti.com> writes: > >> From: Bin Liu <b-liu@ti.com> >> >> Add omap_gpio_disable_irq and omap_gpio_enable_irq >> calls in gpio-omap. >> >> Currently, kernel cannot disable gpio interrupts in >> case of a irq storm, so add omap_gpio_disable/enable_irq >> so that interrupts can be disabled/enabled. >> >> Signed-off-by: Bin Liu <b-liu@ti.com> >> [Judith: Add commit message] >> Signed-off-by: Judith Mendez <jm@ti.com> > > Thanks for this patch. Can you give a bit more context on the > problem(s) this solves and on which SoCs/platforms it was > developed/validated? Sorry for the late response. Patch was tested/developed on am335x device BBB, If you feed a PWM signal at 200KHz frequency to GPIO, and execute gpiomon 0 12 &, Linux will be unresponsive even after CTRL+C without these 2 functions in this patch. Once this patch is applied, you can get console back after hitting CTRL+C and then proceed to kill gpiomon. ~ Judith
On Tue, Nov 05, 2024 at 12:47:58PM -0600, Judith Mendez wrote: > Hi Kevin, > > On 11/1/24 9:29 AM, Kevin Hilman wrote: > > Hi Judith, > > > > Judith Mendez <jm@ti.com> writes: > > > > > From: Bin Liu <b-liu@ti.com> > > > > > > Add omap_gpio_disable_irq and omap_gpio_enable_irq > > > calls in gpio-omap. > > > > > > Currently, kernel cannot disable gpio interrupts in > > > case of a irq storm, so add omap_gpio_disable/enable_irq > > > so that interrupts can be disabled/enabled. > > > > > > Signed-off-by: Bin Liu <b-liu@ti.com> > > > [Judith: Add commit message] > > > Signed-off-by: Judith Mendez <jm@ti.com> > > > > Thanks for this patch. Can you give a bit more context on the > > problem(s) this solves and on which SoCs/platforms it was > > developed/validated? > > Sorry for the late response. Patch was tested/developed on am335x > device BBB, If you feed a PWM signal at 200KHz frequency to > GPIO, and execute gpiomon 0 12 &, Linux will be unresponsive > even after CTRL+C without these 2 functions in this patch. Once > this patch is applied, you can get console back after hitting > CTRL+C and then proceed to kill gpiomon. In addtion to Judith's explanation, when the PWM is applied to a GPIO pin, kernel detects the interrupt storm and disables the irq, however, without these callbacks, this gpio platform driver doesn't really disable the interrupt in the gpio controller, so the interrupt storm is still happening and handled by this gpio controller driver then causes Linux unresponsive. -Bin.
On Tue, Nov 5, 2024 at 8:00 PM Bin Liu <b-liu@ti.com> wrote: > > On Tue, Nov 05, 2024 at 12:47:58PM -0600, Judith Mendez wrote: > > Hi Kevin, > > > > On 11/1/24 9:29 AM, Kevin Hilman wrote: > > > Hi Judith, > > > > > > Judith Mendez <jm@ti.com> writes: > > > > > > > From: Bin Liu <b-liu@ti.com> > > > > > > > > Add omap_gpio_disable_irq and omap_gpio_enable_irq > > > > calls in gpio-omap. > > > > > > > > Currently, kernel cannot disable gpio interrupts in > > > > case of a irq storm, so add omap_gpio_disable/enable_irq > > > > so that interrupts can be disabled/enabled. > > > > > > > > Signed-off-by: Bin Liu <b-liu@ti.com> > > > > [Judith: Add commit message] > > > > Signed-off-by: Judith Mendez <jm@ti.com> > > > > > > Thanks for this patch. Can you give a bit more context on the > > > problem(s) this solves and on which SoCs/platforms it was > > > developed/validated? > > > > Sorry for the late response. Patch was tested/developed on am335x > > device BBB, If you feed a PWM signal at 200KHz frequency to > > GPIO, and execute gpiomon 0 12 &, Linux will be unresponsive > > even after CTRL+C without these 2 functions in this patch. Once > > this patch is applied, you can get console back after hitting > > CTRL+C and then proceed to kill gpiomon. > > In addtion to Judith's explanation, when the PWM is applied to a GPIO > pin, kernel detects the interrupt storm and disables the irq, however, > without these callbacks, this gpio platform driver doesn't really > disable the interrupt in the gpio controller, so the interrupt storm is > still happening and handled by this gpio controller driver then causes > Linux unresponsive. > > -Bin. I take it that this is a fix then and should have relevant Fixes and Cc tags? Bart
Hi Bartosz, On 11/6/24 7:42 AM, Bartosz Golaszewski wrote: > On Tue, Nov 5, 2024 at 8:00 PM Bin Liu <b-liu@ti.com> wrote: >> >> On Tue, Nov 05, 2024 at 12:47:58PM -0600, Judith Mendez wrote: >>> Hi Kevin, >>> >>> On 11/1/24 9:29 AM, Kevin Hilman wrote: >>>> Hi Judith, >>>> >>>> Judith Mendez <jm@ti.com> writes: >>>> >>>>> From: Bin Liu <b-liu@ti.com> >>>>> >>>>> Add omap_gpio_disable_irq and omap_gpio_enable_irq >>>>> calls in gpio-omap. >>>>> >>>>> Currently, kernel cannot disable gpio interrupts in >>>>> case of a irq storm, so add omap_gpio_disable/enable_irq >>>>> so that interrupts can be disabled/enabled. >>>>> >>>>> Signed-off-by: Bin Liu <b-liu@ti.com> >>>>> [Judith: Add commit message] >>>>> Signed-off-by: Judith Mendez <jm@ti.com> >>>> >>>> Thanks for this patch. Can you give a bit more context on the >>>> problem(s) this solves and on which SoCs/platforms it was >>>> developed/validated? >>> >>> Sorry for the late response. Patch was tested/developed on am335x >>> device BBB, If you feed a PWM signal at 200KHz frequency to >>> GPIO, and execute gpiomon 0 12 &, Linux will be unresponsive >>> even after CTRL+C without these 2 functions in this patch. Once >>> this patch is applied, you can get console back after hitting >>> CTRL+C and then proceed to kill gpiomon. >> >> In addtion to Judith's explanation, when the PWM is applied to a GPIO >> pin, kernel detects the interrupt storm and disables the irq, however, >> without these callbacks, this gpio platform driver doesn't really >> disable the interrupt in the gpio controller, so the interrupt storm is >> still happening and handled by this gpio controller driver then causes >> Linux unresponsive. >> >> -Bin. > > I take it that this is a fix then and should have relevant Fixes and Cc tags? ok, will send v3 with fixes tag. ~ Judith
Bin Liu <b-liu@ti.com> writes: > On Tue, Nov 05, 2024 at 12:47:58PM -0600, Judith Mendez wrote: >> Hi Kevin, >> >> On 11/1/24 9:29 AM, Kevin Hilman wrote: >> > Hi Judith, >> > >> > Judith Mendez <jm@ti.com> writes: >> > >> > > From: Bin Liu <b-liu@ti.com> >> > > >> > > Add omap_gpio_disable_irq and omap_gpio_enable_irq >> > > calls in gpio-omap. >> > > >> > > Currently, kernel cannot disable gpio interrupts in >> > > case of a irq storm, so add omap_gpio_disable/enable_irq >> > > so that interrupts can be disabled/enabled. >> > > >> > > Signed-off-by: Bin Liu <b-liu@ti.com> >> > > [Judith: Add commit message] >> > > Signed-off-by: Judith Mendez <jm@ti.com> >> > >> > Thanks for this patch. Can you give a bit more context on the >> > problem(s) this solves and on which SoCs/platforms it was >> > developed/validated? >> >> Sorry for the late response. Patch was tested/developed on am335x >> device BBB, If you feed a PWM signal at 200KHz frequency to >> GPIO, and execute gpiomon 0 12 &, Linux will be unresponsive >> even after CTRL+C without these 2 functions in this patch. Once >> this patch is applied, you can get console back after hitting >> CTRL+C and then proceed to kill gpiomon. > > In addtion to Judith's explanation, when the PWM is applied to a GPIO > pin, kernel detects the interrupt storm and disables the irq, however, > without these callbacks, this gpio platform driver doesn't really > disable the interrupt in the gpio controller, so the interrupt storm is > still happening and handled by this gpio controller driver then causes > Linux unresponsive. Excellent! Thank you both for the detailed explanations. Kevin
© 2016 - 2024 Red Hat, Inc.