[PATCH] pinctrl: baytrail: Enable glitch filter on some systems

[PATCH] pinctrl: baytrail: Enable glitch filter on some systems

[Hans de Goede]

On some systems, some PCB traces attached to edge triggered GpioInts
are routed in such a way that they pick up enough interference to
constantly (many times per second) trigger.

Enabling glitch-filtering fixes this. This commit introduces a DMI
list with systems which need this, but maybe we should enable this on
all Bay Trail GPIOs used as edge-triggered interrupts?

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
---
I wonder if Windows just enables glitch-filtering on all GPIOs? Can
someone perhaps dump the register at physical address 0xFED0C550
(SATAGP0 PINCONF) on a Bay Trail system using GPIO attached volume
buttons while running Windows?
---
 drivers/pinctrl/intel/pinctrl-baytrail.c | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)

diff --git a/drivers/pinctrl/intel/pinctrl-baytrail.c b/drivers/pinctrl/intel/pinctrl-baytrail.c
index 9c1ca29c60b7..21a60b5dac57 100644
--- a/drivers/pinctrl/intel/pinctrl-baytrail.c
+++ b/drivers/pinctrl/intel/pinctrl-baytrail.c
@@ -14,6 +14,7 @@
  * more details.
  */
 
+#include <linux/dmi.h>
 #include <linux/kernel.h>
 #include <linux/init.h>
 #include <linux/types.h>
@@ -46,6 +47,7 @@
 #define BYT_TRIG_POS		BIT(25)
 #define BYT_TRIG_LVL		BIT(24)
 #define BYT_DEBOUNCE_EN		BIT(20)
+#define BYT_GLITCH_FILTER_EN	BIT(19)
 #define BYT_PULL_STR_SHIFT	9
 #define BYT_PULL_STR_MASK	(3 << BYT_PULL_STR_SHIFT)
 #define BYT_PULL_STR_2K		(0 << BYT_PULL_STR_SHIFT)
@@ -205,6 +207,7 @@ struct byt_gpio {
 	const struct byt_pinctrl_soc_data *soc_data;
 	struct byt_community *communities_copy;
 	struct byt_gpio_pin_context *saved_context;
+	bool enable_glitch_filter;
 };
 
 /* SCORE pins, aka GPIOC_<pin_no> or GPIO_S0_SC[<pin_no>] */
@@ -1580,6 +1583,9 @@ static int byt_irq_type(struct irq_data *d, unsigned int type)
 	value &= ~(BYT_DIRECT_IRQ_EN | BYT_TRIG_POS | BYT_TRIG_NEG |
 		   BYT_TRIG_LVL);
 
+	if ((type & IRQ_TYPE_EDGE_BOTH) && vg->enable_glitch_filter)
+		value |= BYT_GLITCH_FILTER_EN;
+
 	writel(value, reg);
 
 	if (type & IRQ_TYPE_EDGE_BOTH)
@@ -1776,6 +1782,19 @@ static const struct acpi_device_id byt_gpio_acpi_match[] = {
 };
 MODULE_DEVICE_TABLE(acpi, byt_gpio_acpi_match);
 
+/* Some devices need the glitch filter on edge interrupts to avoid IRQ storms */
+static const struct dmi_system_id enable_glitch_filter_list[] = {
+	{
+		/* GP-electronic T701 */
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Insyde"),
+			DMI_MATCH(DMI_PRODUCT_NAME, "T701"),
+			DMI_MATCH(DMI_BIOS_VERSION, "BYT70A.YNCHENG.WIN.007"),
+		},
+	},
+	{}
+};
+
 static int byt_pinctrl_probe(struct platform_device *pdev)
 {
 	const struct byt_pinctrl_soc_data *soc_data = NULL;
@@ -1816,6 +1835,9 @@ static int byt_pinctrl_probe(struct platform_device *pdev)
 		return ret;
 	}
 
+	if (dmi_check_system(enable_glitch_filter_list))
+		vg->enable_glitch_filter = true;
+
 	vg->pctl_desc		= byt_pinctrl_desc;
 	vg->pctl_desc.name	= dev_name(&pdev->dev);
 	vg->pctl_desc.pins	= vg->soc_data->pins;
-- 
2.14.3

Re: [PATCH] pinctrl: baytrail: Enable glitch filter on some systems

[Mika Westerberg]

On Thu, Dec 28, 2017 at 11:41:28PM +0100, Hans de Goede wrote:
> On some systems, some PCB traces attached to edge triggered GpioInts
> are routed in such a way that they pick up enough interference to
> constantly (many times per second) trigger.
> 
> Enabling glitch-filtering fixes this. This commit introduces a DMI
> list with systems which need this, but maybe we should enable this on
> all Bay Trail GPIOs used as edge-triggered interrupts?

I think it makes sense to always enable it for GPIOs used as interrupts
(regardless whether it is edge or level).

Re: [PATCH] pinctrl: baytrail: Enable glitch filter on some systems

[Hans de Goede]

Hi,

On 29-12-17 11:54, Mika Westerberg wrote:
> On Thu, Dec 28, 2017 at 11:41:28PM +0100, Hans de Goede wrote:
>> On some systems, some PCB traces attached to edge triggered GpioInts
>> are routed in such a way that they pick up enough interference to
>> constantly (many times per second) trigger.
>>
>> Enabling glitch-filtering fixes this. This commit introduces a DMI
>> list with systems which need this, but maybe we should enable this on
>> all Bay Trail GPIOs used as edge-triggered interrupts?
> 
> I think it makes sense to always enable it for GPIOs used as interrupts
> (regardless whether it is edge or level).

That sounds like a good idea to me. I will post a v2 implementing this.

Regards,

Hans