Skip to contentSkip to conversation

net: stmmac: Support gpio high-level reset for devices requiring it

PATCH In review Lizhe 3 participants 10 messages Jul 8, 2025 – Jul 9, 2025 5 messages not in the archive

Subsystems (4)
lore
Share & tools
lore

Lizhe

author
#0 lore raw

PATCH 1/1 net: stmmac: Support gpio high-level reset for devices requiring it

some devices only reset when the GPIO is at a high level, but the current function lacks support for such devices. add high-level reset functionality to the function to support devices that require high-level triggering for reset

---
 drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c +6 -2
index 836f2848dfeb..cb989e6d7eac 100644
@@ -458,6 +458,7 @@ int stmmac_mdio_reset(struct mii_bus *bus)
 
 #ifdef CONFIG_OF
 	if (priv->device->of_node) {
+		int active_low = 0;
 		struct gpio_desc *reset_gpio;
 		u32 delays[3] = { 0, 0, 0 };
 
@@ -467,6 +468,9 @@ int stmmac_mdio_reset(struct mii_bus *bus)
 		if (IS_ERR(reset_gpio))
 			return PTR_ERR(reset_gpio);
 
+		if (reset_gpio)
+			active_low = gpiod_is_active_low(reset_gpio);
+
 		device_property_read_u32_array(priv->device,
 					       "snps,reset-delays-us",
 					       delays, ARRAY_SIZE(delays));
@@ -474,11 +478,11 @@ int stmmac_mdio_reset(struct mii_bus *bus)
 		if (delays[0])
 			msleep(DIV_ROUND_UP(delays[0], 1000));
 
-		gpiod_set_value_cansleep(reset_gpio, 1);
+		gpiod_set_value_cansleep(reset_gpio, active_low ? 1 : 0);
 		if (delays[1])
 			msleep(DIV_ROUND_UP(delays[1], 1000));
 
-		gpiod_set_value_cansleep(reset_gpio, 0);
+		gpiod_set_value_cansleep(reset_gpio, active_low ? 0 : 1);
 		if (delays[2])
 			msleep(DIV_ROUND_UP(delays[2], 1000));
 	}
Signature · 1 line
2.17.1
#1 lore raw
Quoted 4 lines · Lizhe

some devices only reset when the GPIO is at a high level, but the current function lacks support for such devices. add high-level reset functionality to the function to support devices that require high-level triggering for reset

You can probably specify this in DT:

reset-gpios = <&qe_pio_e 18 GPIO_ACTIVE_LOW>;

The gpio core will then flip the meaning of

gpiod_set_value_cansleep(reset_gpio, 1);

such that a value of 1 will become low, 0 will become high.

Andrew

A message that is not in the archive (sent off-list or not yet received): [email protected]

#3 lore raw
Quoted 1 line · lizhe

+ gpio_state = gpiod_get_value_can_sleep(reset_gpio);

Use gpiod_get_raw_value_cansleep(). The normal get/set return the active/inactive state, whereas the _raw get/set return the physical state.

Signature · 2 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

A message that is not in the archive (sent off-list or not yet received): [email protected]

#5 lore raw
Quoted 18 lines · lizhe

Hi,

after replacing with this function, the function returns 0, meaning the gpio is still at a low voltage level.

(quoted diff) +5 -0
+              int gpio_state = -1;
 
 
                if (delays[2])
                        msleep(DIV_ROUND_UP(delays[2], 1000));
+
+               gpio_state = gpiod_get_raw_value_cansleep(reset_gpio);
+               pr_info("gpio_state: %d\n", gpio_state);
+               pr_info("gpio_state: %d\n", gpio_state);
 
 gpio-111 (                    |snps,reset          ) out lo
Quoted 2 lines
[    3.899319] gpio_state: 0
[    3.899324] gpio_state: 0

How have you declared the snps,reset-gpio property in DT?

Signature · 2 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

A message that is not in the archive (sent off-list or not yet received): [email protected]

#7 lore raw
Quoted 6 lines · lizhe

Hi,

    i have already declared it in the DTS.

    &gmac {
            snps,reset-gpio = <&gpio3 RK_PB7 GPIO_ACTIVE_HIGH>;
    };

Active high means that the reset is in effect when the output is at the logic '1' state. So, gpiod_get_value*() will return the same as gpiod_get_raw_value*().

Active low means that the reset is in effect when the output is at the logic '0' state, and in that case the return value of gpiod_get_value*() will return true when the reset is active (at logic '0' state) whereas gpiod_get_raw_value*() will return zero.

Signature · 2 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

A message that is not in the archive (sent off-list or not yet received): [email protected]

#9 lore raw
Quoted 1 line · lizhe

Hi,

Please don't top post.

Quoted 5 lines

snps, reset-gpio = <&gpioX RK_PXX GPIO_ACTIVE_HIGH>;
All of them correctly parse the GPIO pin's state are described in the DTS

Quoted 24 lines

Thx !

Lizhe

Quoted 16 lines · Russell King (Oracle)
Quoted 6 lines · lizhe

Hi,

    i have already declared it in the DTS.

    &gmac {
            snps,reset-gpio = <&gpio3 RK_PB7 GPIO_ACTIVE_HIGH>;
    };

Active high means that the reset is in effect when the output is at the logic '1' state. So, gpiod_get_value*() will return the same as gpiod_get_raw_value*().

Active low means that the reset is in effect when the output is at the logic '0' state, and in that case the return value of gpiod_get_value*() will return true when the reset is active (at logic '0' state) whereas gpiod_get_raw_value*() will return zero.

Did you read what Russell said? If so, why are you using GPIO_ACTIVE_HIGH?

Andrew

#10 lore raw
In reply to a message not in the archive
Quoted 36 lines · lizhe

Hi,

if i add the following code to this function, the gpio outputs a high level

without this code, it outputs a low level,
the function currently drivers the reset GPIO to a low state, failing to account

for devices requiring an active-high reset.

i invited our hardware engineer to measure the voltage level on this GPIO pin,

without adding this code, the voltage at this GPIO pin remains at 0V

+ int current_value;

+ keep_high = device_property_read_bool(priv->device,

+ "snps,reset-keep-high");

+ if (keep_high) {

+ gpiod_set_value_cansleep(reset_gpio, 1);

+ current_value = gpiod_get_value_cansleep(reset_gpio);

Quoted 9 lines

+ pr_info("current_value: %d\n", current_value);

+ }

in the RK3588 system, i am using ,there are many DTS node configured link this:

snps, reset-gpio = <&gpioX RK_PXX GPIO_ACTIVE_HIGH>;
All of them correctly parse the GPIO pin's state are described in the DTS

I'm wondering at this point whether the problem here is one of mis-understanding the engineering terminology. Look at the below using a fixed-width font:

Active-high reset: _____/^^^^^^^^\____

Active-low reset: ^^^^^\________/^^^^

                        | reset  |
			|asserted|

So, an active high reset needs to be logic low in order for the device to function. An active low reset needs to be logic high for the device to function.

You seem to be wanting to tell the kernel that you have an active high reset, and expect it to be logic high when you want it to be active. That is *not* an active high reset.

Signature · 2 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!
#11 lore raw
In reply to a message not in the archive
Quoted 10 lines · lizhe

Hi, Andrew

Thx,

i conducted an experiment, and no matter whether i configured it as

GPIO_ACTIVE_LOW or GPIO_ACTIVE_HIGH in the dts, the resulting

GPIO pin state was 0, indicating a low level.

You need to keep digging and understand why.

Scatter some printk() in the gpio core code. Does it recognise the option in DT? Is GPIO_ACTIVE_LOW, FLAG_ACTIVE_LOW being set?

Put some prints into the actual GPIO driver, what is passed to it.

There is something interesting in gpio.txt:

Most controllers are specifying a generic flag bitfield in the last cell, so for these, use the macros defined in include/dt-bindings/gpio/gpio.h whenever possible:

It says 'Most'. Is the GPIO controller you are using not actually doing this?

Andrew

#12 lore raw
In reply to Lizhe
Quoted 36 lines · Lizhe

some devices only reset when the GPIO is at a high level, but the current function lacks support for such devices. add high-level reset functionality to the function to support devices that require high-level triggering for reset

---
 drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c +4 -0
index 836f2848dfeb..cb989e6d7eac 100644
@@ -458,6 +458,7 @@ int stmmac_mdio_reset(struct mii_bus *bus)
 
 #ifdef CONFIG_OF
 	if (priv->device->of_node) {
+		int active_low = 0;
 		struct gpio_desc *reset_gpio;
 		u32 delays[3] = { 0, 0, 0 };
 
@@ -467,6 +468,9 @@ int stmmac_mdio_reset(struct mii_bus *bus)
 		if (IS_ERR(reset_gpio))
 			return PTR_ERR(reset_gpio);
 
+		if (reset_gpio)
+			active_low = gpiod_is_active_low(reset_gpio);
+
 		device_property_read_u32_array(priv->device,
 					       "snps,reset-delays-us",
 					       delays, ARRAY_SIZE(delays));
@@ -474,11 +478,11 @@ int stmmac_mdio_reset(struct mii_bus *bus)
 		if (delays[0])
 			msleep(DIV_ROUND_UP(delays[0], 1000));
 

drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c int stmmac_mdio_reset(struct mii_bus *bus) · line 488

Quoted 10 lines
drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c +2 -2
-		gpiod_set_value_cansleep(reset_gpio, 1);
+		gpiod_set_value_cansleep(reset_gpio, active_low ? 1 : 0);
 		if (delays[1])
 			msleep(DIV_ROUND_UP(delays[1], 1000));
 
-		gpiod_set_value_cansleep(reset_gpio, 0);
+		gpiod_set_value_cansleep(reset_gpio, active_low ? 0 : 1);
 		if (delays[2])
 			msleep(DIV_ROUND_UP(delays[2], 1000));
 	}

NAK. Not required. The GPIO layer can cope with active-high and active-low signals declared in firmware without needing driver modification. Use the right data in the firmware and you don't need to patch.

/* Bit 0 express polarity */
#define GPIO_ACTIVE_HIGH 0
#define GPIO_ACTIVE_LOW 1

Signature · 2 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

A message that is not in the archive (sent off-list or not yet received): [email protected]

#14 lore raw
Quoted 9 lines · lizhe

Hi,

Thx !

i conducted an experiment, and no matter whether i configured it as GPIO_ACTIVE_LOW or GPIO_ACTIVE_HIGH in the dts, the resulting GPIO pin state was 0, indicating a low level.

Quoted 5 lines

if (delays[2])

    msleep(DIV_ROUND_UP(DELAYS[2], 1000));

+ gpio_state = gpiod_get_value_can_sleep(reset_gpio);
+ pr_info("gpio_state: %d\n", gpio_state);

Use gpiod_get_raw_value_cansleep().

Signature · 2 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!