* [PATCH v1 1/1] gpio: mlxbf3: Support shutdown() function
@ 2024-06-11 14:38 Asmaa Mnebhi
2024-06-11 15:12 ` Andy Shevchenko
0 siblings, 1 reply; 2+ messages in thread
From: Asmaa Mnebhi @ 2024-06-11 14:38 UTC (permalink / raw)
To: andy.shevchenko, linus.walleij, bgolaszewski, linux-gpio
Cc: Asmaa Mnebhi, David Thompson
During Linux graceful reboot, the GPIO interrupts are not disabled.
Since the drivers are not removed during graceful reboot,
the logic to call mlxbf3_gpio_irq_disable() is not triggered.
Interrupts that remain enabled can cause issues on subsequent boots.
For example, the mlxbf-gige driver contains PHY logic to bring up the link.
If the gpio-mlxbf3 driver loads first, the mlxbf-gige driver
will use a GPIO interrupt to bring up the link.
Otherwise, it will use polling.
The next time Linux boots and loads the drivers in this order, we encounter the issue:
- mlxbf-gige loads first. The gige driver uses polling while the GPIO10 interrupt
is still enabled from previous boot so if the interrupt triggers, there is nothing
to clear it.
- gpio-mlxbf3 loads.
- i2c-mlxbf loads. The interrupt doesn't trigger for I2C because it is shared with the gpio
interrupt line which was not cleared.
The solution is to add a shutdown function to the gpio driver to clear and disable
all interrupts. Also clear the interrupt after disabling it in mlxbf3_gpio_irq_disable().
Fixes: 38a700efc510 ("gpio: mlxbf3: Add gpio driver support")
Signed-off-by: Asmaa Mnebhi <asmaa@nvidia.com>
Reviewed-by: David Thompson <davthompson@nvidia.com>
---
drivers/gpio/gpio-mlxbf3.c | 14 ++++++++++++++
1 file changed, 14 insertions(+)
diff --git a/drivers/gpio/gpio-mlxbf3.c b/drivers/gpio/gpio-mlxbf3.c
index d5906d419b0a..40bd93e3d695 100644
--- a/drivers/gpio/gpio-mlxbf3.c
+++ b/drivers/gpio/gpio-mlxbf3.c
@@ -39,6 +39,8 @@
#define MLXBF_GPIO_CAUSE_OR_EVTEN0 0x14
#define MLXBF_GPIO_CAUSE_OR_CLRCAUSE 0x18
+#define MLXBF_GPIO_CLR_ALL_INTS 0xFFFFFFFF
+
struct mlxbf3_gpio_context {
struct gpio_chip gc;
@@ -82,6 +84,8 @@ static void mlxbf3_gpio_irq_disable(struct irq_data *irqd)
val = readl(gs->gpio_cause_io + MLXBF_GPIO_CAUSE_OR_EVTEN0);
val &= ~BIT(offset);
writel(val, gs->gpio_cause_io + MLXBF_GPIO_CAUSE_OR_EVTEN0);
+
+ writel(BIT(offset), gs->gpio_cause_io + MLXBF_GPIO_CAUSE_OR_CLRCAUSE);
raw_spin_unlock_irqrestore(&gs->gc.bgpio_lock, flags);
gpiochip_disable_irq(gc, offset);
@@ -253,6 +257,15 @@ static int mlxbf3_gpio_probe(struct platform_device *pdev)
return 0;
}
+static void mlxbf3_gpio_shutdown(struct platform_device *pdev)
+{
+ struct mlxbf3_gpio_context *gs = dev_get_drvdata(&pdev->dev);
+
+ /* Disable and clear all interrupts */
+ writel(0, gs->gpio_cause_io + MLXBF_GPIO_CAUSE_OR_EVTEN0);
+ writel(MLXBF_GPIO_CLR_ALL_INTS, gs->gpio_cause_io + MLXBF_GPIO_CAUSE_OR_CLRCAUSE);
+}
+
static const struct acpi_device_id mlxbf3_gpio_acpi_match[] = {
{ "MLNXBF33", 0 },
{}
@@ -265,6 +278,7 @@ static struct platform_driver mlxbf3_gpio_driver = {
.acpi_match_table = mlxbf3_gpio_acpi_match,
},
.probe = mlxbf3_gpio_probe,
+ .shutdown = mlxbf3_gpio_shutdown,
};
module_platform_driver(mlxbf3_gpio_driver);
--
2.30.1
^ permalink raw reply related [flat|nested] 2+ messages in thread
* Re: [PATCH v1 1/1] gpio: mlxbf3: Support shutdown() function
2024-06-11 14:38 [PATCH v1 1/1] gpio: mlxbf3: Support shutdown() function Asmaa Mnebhi
@ 2024-06-11 15:12 ` Andy Shevchenko
0 siblings, 0 replies; 2+ messages in thread
From: Andy Shevchenko @ 2024-06-11 15:12 UTC (permalink / raw)
To: Asmaa Mnebhi; +Cc: linus.walleij, bgolaszewski, linux-gpio, David Thompson
On Tue, Jun 11, 2024 at 5:38 PM Asmaa Mnebhi <asmaa@nvidia.com> wrote:
>
> During Linux graceful reboot, the GPIO interrupts are not disabled.
> Since the drivers are not removed during graceful reboot,
> the logic to call mlxbf3_gpio_irq_disable() is not triggered.
> Interrupts that remain enabled can cause issues on subsequent boots.
>
> For example, the mlxbf-gige driver contains PHY logic to bring up the link.
> If the gpio-mlxbf3 driver loads first, the mlxbf-gige driver
> will use a GPIO interrupt to bring up the link.
> Otherwise, it will use polling.
> The next time Linux boots and loads the drivers in this order, we encounter the issue:
> - mlxbf-gige loads first. The gige driver uses polling while the GPIO10 interrupt
> is still enabled from previous boot so if the interrupt triggers, there is nothing
> to clear it.
> - gpio-mlxbf3 loads.
> - i2c-mlxbf loads. The interrupt doesn't trigger for I2C because it is shared with the gpio
GPIO
> interrupt line which was not cleared.
> The solution is to add a shutdown function to the gpio driver to clear and disable
GPIO
> all interrupts. Also clear the interrupt after disabling it in mlxbf3_gpio_irq_disable().
...
> +#define MLXBF_GPIO_CLR_ALL_INTS 0xFFFFFFFF
GENMASK()
...
> +static void mlxbf3_gpio_shutdown(struct platform_device *pdev)
> +{
> + struct mlxbf3_gpio_context *gs = dev_get_drvdata(&pdev->dev);
platform_get_drvdata()
> + /* Disable and clear all interrupts */
> + writel(0, gs->gpio_cause_io + MLXBF_GPIO_CAUSE_OR_EVTEN0);
> + writel(MLXBF_GPIO_CLR_ALL_INTS, gs->gpio_cause_io + MLXBF_GPIO_CAUSE_OR_CLRCAUSE);
> +}
--
With Best Regards,
Andy Shevchenko
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2024-06-11 15:12 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2024-06-11 14:38 [PATCH v1 1/1] gpio: mlxbf3: Support shutdown() function Asmaa Mnebhi
2024-06-11 15:12 ` Andy Shevchenko
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox