Skip to contentSkip to conversation

mmc: host: sdhci-esdhc-imx: let usr to config the wakeup for gpio cd pin through sysfs

PATCH In review [email protected] 3 participants 8 messages Oct 17, 2023 – Oct 20, 2023

Subsystems (4)
lore
Share & tools
lore

[email protected]

author maintainer
#0 lore raw

PATCH 1/1 mmc: host: sdhci-esdhc-imx: let usr to config the wakeup for gpio cd pin through sysfs

From: Haibo Chen <[email protected]>

Currently default enable the gpio cd pin wakeup, this will waste power after system suspend if usr do not need this cd pin wakeup feature. Now let usr to config the wakeup for gpio cd pin through sysfs.

---
 drivers/mmc/host/sdhci-esdhc-imx.c | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)
drivers/mmc/host/sdhci-esdhc-imx.c +10 -4
index 40a6e2f8145a..2e46648344ba 100644
@@ -1797,9 +1797,12 @@ static int sdhci_esdhc_imx_probe(struct platform_device *pdev)
 	/*
 	 * Setup the wakeup capability here, let user to decide
 	 * whether need to enable this wakeup through sysfs interface.
+	 * First check the SDIO device, second check the gpio CD pin.
 	 */
-	if ((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
-			(host->mmc->pm_caps & MMC_PM_WAKE_SDIO_IRQ))
+	if (((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
+			(host->mmc->pm_caps & MMC_PM_WAKE_SDIO_IRQ)) ||
+	    ((host->mmc->caps & MMC_CAP_CD_WAKE) &&
+			 host->mmc->slot.cd_irq >= 0))
 		device_set_wakeup_capable(&pdev->dev, true);
 
 	pm_runtime_set_active(&pdev->dev);
@@ -1878,7 +1881,8 @@ static int sdhci_esdhc_suspend(struct device *dev)
 	if (ret)
 		return ret;
 
-	ret = mmc_gpio_set_cd_wake(host->mmc, true);
+	if (device_may_wakeup(dev))
+		ret = mmc_gpio_set_cd_wake(host->mmc, true);
 
 	return ret;
 }
@@ -1901,8 +1905,10 @@ static int sdhci_esdhc_resume(struct device *dev)
 
 	if (host->mmc->caps2 & MMC_CAP2_CQE)
 		ret = cqhci_resume(host->mmc);
+	if (ret)
+		return ret;
 
-	if (!ret)
+	if (device_may_wakeup(dev))
 		ret = mmc_gpio_set_cd_wake(host->mmc, false);
 
 	return ret;
Signature · 1 line
2.34.1
#1 lore raw
Quoted 20 lines · On Tue, 17 Oct 2023 at 09:26, <[email protected]> wrote:

From: Haibo Chen <[email protected]>

Currently default enable the gpio cd pin wakeup, this will waste power after system suspend if usr do not need this cd pin wakeup feature. Now let usr to config the wakeup for gpio cd pin through sysfs.

---
 drivers/mmc/host/sdhci-esdhc-imx.c | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)
drivers/mmc/host/sdhci-esdhc-imx.c +0 -0
index 40a6e2f8145a..2e46648344ba 100644
@@ -1797,9 +1797,12 @@ static int sdhci_esdhc_imx_probe(struct platform_device *pdev)
        /*
         * Setup the wakeup capability here, let user to decide

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct platform_device *pdev) · line 1806

Quoted 10 lines
drivers/mmc/host/sdhci-esdhc-imx.c +5 -2
         * whether need to enable this wakeup through sysfs interface.
+        * First check the SDIO device, second check the gpio CD pin.
         */
-       if ((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
-                       (host->mmc->pm_caps & MMC_PM_WAKE_SDIO_IRQ))
+       if (((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
+                       (host->mmc->pm_caps & MMC_PM_WAKE_SDIO_IRQ)) ||
+           ((host->mmc->caps & MMC_CAP_CD_WAKE) &&
+                        host->mmc->slot.cd_irq >= 0))
                device_set_wakeup_capable(&pdev->dev, true);

If the wakeup is GPIO based, it doesn't mean that the device is wakeup capable, so this is wrong.

In principle this must be managed through the GPIO irqchip driver instead.

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_suspend(struct device *dev) · line 1884

Quoted 8 lines
drivers/mmc/host/sdhci-esdhc-imx.c +1 -1
        pm_runtime_set_active(&pdev->dev);
@@ -1878,7 +1881,8 @@ static int sdhci_esdhc_suspend(struct device *dev)
        if (ret)
                return ret;
 
-       ret = mmc_gpio_set_cd_wake(host->mmc, true);
+       if (device_may_wakeup(dev))

As I indicated above. It's not really the device that corresponds to the mmc controller that can wake up the system, but rather the gpio irqchip.

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_resume(struct device *dev) · line 1911

Quoted 13 lines
drivers/mmc/host/sdhci-esdhc-imx.c +4 -1
+               ret = mmc_gpio_set_cd_wake(host->mmc, true);
 
        return ret;
 }
@@ -1901,8 +1905,10 @@ static int sdhci_esdhc_resume(struct device *dev)
 
        if (host->mmc->caps2 & MMC_CAP2_CQE)
                ret = cqhci_resume(host->mmc);
+       if (ret)
+               return ret;
 
-       if (!ret)
+       if (device_may_wakeup(dev))

Ditto.

Quoted 6 lines
drivers/mmc/host/sdhci-esdhc-imx.c +0 -1
                ret = mmc_gpio_set_cd_wake(host->mmc, false);
 
        return ret;
--

2.34.1

Kind regards
Uffe

author maintainer
#2 lore raw
Quoted 41 lines

-----Original Message-----
From: Ulf Hansson <[email protected]>
Sent: 2023年10月17日 19:35
To: Bough Chen <[email protected]>
Cc: [email protected]; [email protected]; dl-linux-imx <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: [PATCH] mmc: host: sdhci-esdhc-imx: let usr to config the wakeup for gpio cd pin through sysfs

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct

Quoted 18 lines · On Tue, 17 Oct 2023 at 09:26, <[email protected]> wrote:

From: Haibo Chen <[email protected]>

Currently default enable the gpio cd pin wakeup, this will waste power after system suspend if usr do not need this cd pin wakeup feature. Now let usr to config the wakeup for gpio cd pin through sysfs.

---
 drivers/mmc/host/sdhci-esdhc-imx.c | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)
drivers/mmc/host/sdhci-esdhc-imx.c +0 -0
index 40a6e2f8145a..2e46648344ba 100644
@@ -1797,9 +1797,12 @@ static int sdhci_esdhc_imx_probe(struct

platform_device *pdev)

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct platform_device *pdev)

Quoted 7 lines
drivers/mmc/host/sdhci-esdhc-imx.c +1 -2
        /*
         * Setup the wakeup capability here, let user to decide
         * whether need to enable this wakeup through sysfs interface.
+        * First check the SDIO device, second check the gpio CD pin.
         */
-       if ((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
-                       (host->mmc->pm_caps &

MMC_PM_WAKE_SDIO_IRQ))

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct · line 1803

Quoted 2 lines
drivers/mmc/host/sdhci-esdhc-imx.c +2 -0
+       if (((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
+                       (host->mmc->pm_caps &

MMC_PM_WAKE_SDIO_IRQ)) ||

Quoted 8 lines

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct platform_device *pdev) · line 1806

Quoted 3 lines
drivers/mmc/host/sdhci-esdhc-imx.c +2 -0
+           ((host->mmc->caps & MMC_CAP_CD_WAKE) &&
+                        host->mmc->slot.cd_irq >= 0))
                device_set_wakeup_capable(&pdev->dev, true);

If the wakeup is GPIO based, it doesn't mean that the device is wakeup capable, so this is wrong.

In principle this must be managed through the GPIO irqchip driver instead.

Yes, I know GPIO and USDHC are different device, combine GPIO wakeup to USDHC is not that reasonable. I send this patch because I notice there is similar code in sdhci-pci-core.c, refer to sdhci_pci_suspend_host()

Seems PCI subsystem combine GPIO wakeup to PCI controller.

Anyway, I will try to implement this in GPIO irqchip driver.

Best Regards
Haibo Chen

Quoted 39 lines

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_suspend(struct device *dev) · line 1884

Quoted 9 lines
drivers/mmc/host/sdhci-esdhc-imx.c +1 -1
        pm_runtime_set_active(&pdev->dev);
@@ -1878,7 +1881,8 @@ static int sdhci_esdhc_suspend(struct device *dev)
        if (ret)
                return ret;
 
-       ret = mmc_gpio_set_cd_wake(host->mmc, true);
+       if (device_may_wakeup(dev))

As I indicated above. It's not really the device that corresponds to the mmc controller that can wake up the system, but rather the gpio irqchip.

Quoted 5 lines
drivers/mmc/host/sdhci-esdhc-imx.c +1 -0
+               ret = mmc_gpio_set_cd_wake(host->mmc, true);
 
        return ret;
 }

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_resume(struct device · line 1912

Quoted 9 lines
drivers/mmc/host/sdhci-esdhc-imx.c +3 -1
@@ -1901,8 +1905,10 @@ static int sdhci_esdhc_resume(struct device
 *dev)
 
        if (host->mmc->caps2 & MMC_CAP2_CQE)
                ret = cqhci_resume(host->mmc);
+       if (ret)
+               return ret;
 
-       if (!ret)
+       if (device_may_wakeup(dev))

Ditto.

Quoted 6 lines
drivers/mmc/host/sdhci-esdhc-imx.c +0 -0
                ret = mmc_gpio_set_cd_wake(host->mmc, false);
 
        return ret;

--
2.34.1

Kind regards
Uffe

#3 lore raw
Quoted 49 lines · Bough Chen
Quoted 49 lines

-----Original Message-----
From: Ulf Hansson <[email protected]>
Sent: 2023年10月17日 19:35
To: Bough Chen <[email protected]>
Cc: [email protected]; [email protected]; dl-linux-imx <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: [PATCH] mmc: host: sdhci-esdhc-imx: let usr to config the wakeup for gpio cd pin through sysfs

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct

Quoted 18 lines · On Tue, 17 Oct 2023 at 09:26, <[email protected]> wrote:

From: Haibo Chen <[email protected]>

Currently default enable the gpio cd pin wakeup, this will waste power after system suspend if usr do not need this cd pin wakeup feature. Now let usr to config the wakeup for gpio cd pin through sysfs.

---
 drivers/mmc/host/sdhci-esdhc-imx.c | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)
drivers/mmc/host/sdhci-esdhc-imx.c +0 -0
index 40a6e2f8145a..2e46648344ba 100644
@@ -1797,9 +1797,12 @@ static int sdhci_esdhc_imx_probe(struct

platform_device *pdev)

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct platform_device *pdev)

Quoted 7 lines
drivers/mmc/host/sdhci-esdhc-imx.c +1 -2
        /*
         * Setup the wakeup capability here, let user to decide
         * whether need to enable this wakeup through sysfs interface.
+        * First check the SDIO device, second check the gpio CD pin.
         */
-       if ((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
-                       (host->mmc->pm_caps &

MMC_PM_WAKE_SDIO_IRQ))

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct · line 1803

Quoted 2 lines
drivers/mmc/host/sdhci-esdhc-imx.c +2 -0
+       if (((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
+                       (host->mmc->pm_caps &

MMC_PM_WAKE_SDIO_IRQ)) ||

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct platform_device *pdev) · line 1806

Quoted 3 lines
drivers/mmc/host/sdhci-esdhc-imx.c +2 -0
+           ((host->mmc->caps & MMC_CAP_CD_WAKE) &&
+                        host->mmc->slot.cd_irq >= 0))
                device_set_wakeup_capable(&pdev->dev, true);

If the wakeup is GPIO based, it doesn't mean that the device is wakeup capable, so this is wrong.

Quoted 6 lines
Quoted 1 line

In principle this must be managed through the GPIO irqchip driver instead.

Yes, I know GPIO and USDHC are different device, combine GPIO wakeup to USDHC is not that reasonable. I send this patch because I notice there is similar code in sdhci-pci-core.c, refer to sdhci_pci_suspend_host()

Seems PCI subsystem combine GPIO wakeup to PCI controller.

Good point! I am not sure that is the correct way to do it either, but maybe it's fine for PCI based devices. Adrian probably knows this better.

The main problem I see with your approach, is that it may prevent a PM domain (genpd), if the USDHC device is attached to one, from being powered-off during system suspend. That would be wrong as the USDHC device doesn't need to stay powered-on to allow a GPIO to be configured as a system wakeup irq.

Quoted 2 lines

Anyway, I will try to implement this in GPIO irqchip driver.

That sounds like the right thing to do.

Although, it does mean that we won't be able to allow userspace to decide on a per GPIO pin/irq basis, if system wakeup should be enabled or disabled. As far as I know, we don't have a solution in place to support that.

[...]

Kind regards
Uffe

maintainer
#4 lore raw
Quoted 52 lines · Ulf Hansson
Quoted 51 lines · Bough Chen
Quoted 50 lines

-----Original Message-----
From: Ulf Hansson <[email protected]>
Sent: 2023年10月17日 19:35
To: Bough Chen <[email protected]>
Cc: [email protected]; [email protected]; dl-linux-imx <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: [PATCH] mmc: host: sdhci-esdhc-imx: let usr to config the wakeup for gpio cd pin through sysfs

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct

Quoted 18 lines · On Tue, 17 Oct 2023 at 09:26, <[email protected]> wrote:

From: Haibo Chen <[email protected]>

Currently default enable the gpio cd pin wakeup, this will waste power after system suspend if usr do not need this cd pin wakeup feature. Now let usr to config the wakeup for gpio cd pin through sysfs.

---
 drivers/mmc/host/sdhci-esdhc-imx.c | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)
drivers/mmc/host/sdhci-esdhc-imx.c +0 -0
index 40a6e2f8145a..2e46648344ba 100644
@@ -1797,9 +1797,12 @@ static int sdhci_esdhc_imx_probe(struct

platform_device *pdev)

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct platform_device *pdev)

Quoted 7 lines
drivers/mmc/host/sdhci-esdhc-imx.c +1 -2
        /*
         * Setup the wakeup capability here, let user to decide
         * whether need to enable this wakeup through sysfs interface.
+        * First check the SDIO device, second check the gpio CD pin.
         */
-       if ((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
-                       (host->mmc->pm_caps &

MMC_PM_WAKE_SDIO_IRQ))

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct · line 1803

Quoted 2 lines
drivers/mmc/host/sdhci-esdhc-imx.c +2 -0
+       if (((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
+                       (host->mmc->pm_caps &

MMC_PM_WAKE_SDIO_IRQ)) ||

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct platform_device *pdev) · line 1806

Quoted 3 lines
drivers/mmc/host/sdhci-esdhc-imx.c +2 -0
+           ((host->mmc->caps & MMC_CAP_CD_WAKE) &&
+                        host->mmc->slot.cd_irq >= 0))
                device_set_wakeup_capable(&pdev->dev, true);

If the wakeup is GPIO based, it doesn't mean that the device is wakeup capable, so this is wrong.

In principle this must be managed through the GPIO irqchip driver instead.

Quoted 8 lines
Quoted 4 lines

Yes, I know GPIO and USDHC are different device, combine GPIO wakeup to USDHC is not that reasonable. I send this patch because I notice there is similar code in sdhci-pci-core.c, refer to sdhci_pci_suspend_host()

Seems PCI subsystem combine GPIO wakeup to PCI controller.

Good point! I am not sure that is the correct way to do it either, but maybe it's fine for PCI based devices. Adrian probably knows this better.

PCI / ACPI layers handle configuration of wakeup signaling, so a device will be put to the appropriate power state. It won't be forced to be "on" by registering a wakeup source.

Quoted 21 lines

The main problem I see with your approach, is that it may prevent a PM domain (genpd), if the USDHC device is attached to one, from being powered-off during system suspend. That would be wrong as the USDHC device doesn't need to stay powered-on to allow a GPIO to be configured as a system wakeup irq.

Quoted 2 lines

Anyway, I will try to implement this in GPIO irqchip driver.

That sounds like the right thing to do.

Although, it does mean that we won't be able to allow userspace to decide on a per GPIO pin/irq basis, if system wakeup should be enabled or disabled. As far as I know, we don't have a solution in place to support that.

[...]

Kind regards
Uffe

#5 lore raw
Quoted 59 lines · Adrian Hunter
Quoted 57 lines · Ulf Hansson
Quoted 55 lines · Bough Chen
Quoted 42 lines

-----Original Message-----
From: Ulf Hansson <[email protected]>
Sent: 2023年10月17日 19:35
To: Bough Chen <[email protected]>
Cc: [email protected]; [email protected]; dl-linux-imx <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: [PATCH] mmc: host: sdhci-esdhc-imx: let usr to config the wakeup for gpio cd pin through sysfs

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct

Quoted 18 lines · On Tue, 17 Oct 2023 at 09:26, <[email protected]> wrote:

From: Haibo Chen <[email protected]>

Currently default enable the gpio cd pin wakeup, this will waste power after system suspend if usr do not need this cd pin wakeup feature. Now let usr to config the wakeup for gpio cd pin through sysfs.

---
 drivers/mmc/host/sdhci-esdhc-imx.c | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)
drivers/mmc/host/sdhci-esdhc-imx.c +0 -0
index 40a6e2f8145a..2e46648344ba 100644
@@ -1797,9 +1797,12 @@ static int sdhci_esdhc_imx_probe(struct

platform_device *pdev)

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct platform_device *pdev)

Quoted 7 lines
drivers/mmc/host/sdhci-esdhc-imx.c +1 -2
        /*
         * Setup the wakeup capability here, let user to decide
         * whether need to enable this wakeup through sysfs interface.
+        * First check the SDIO device, second check the gpio CD pin.
         */
-       if ((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
-                       (host->mmc->pm_caps &

MMC_PM_WAKE_SDIO_IRQ))

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct · line 1803

Quoted 2 lines
drivers/mmc/host/sdhci-esdhc-imx.c +2 -0
+       if (((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
+                       (host->mmc->pm_caps &

MMC_PM_WAKE_SDIO_IRQ)) ||

Quoted 8 lines

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct platform_device *pdev) · line 1806

Quoted 3 lines
drivers/mmc/host/sdhci-esdhc-imx.c +2 -0
+           ((host->mmc->caps & MMC_CAP_CD_WAKE) &&
+                        host->mmc->slot.cd_irq >= 0))
                device_set_wakeup_capable(&pdev->dev, true);

If the wakeup is GPIO based, it doesn't mean that the device is wakeup capable, so this is wrong.

In principle this must be managed through the GPIO irqchip driver instead.

Yes, I know GPIO and USDHC are different device, combine GPIO wakeup to USDHC is not that reasonable. I send this patch because I notice there is similar code in sdhci-pci-core.c, refer to sdhci_pci_suspend_host()

Seems PCI subsystem combine GPIO wakeup to PCI controller.

Quoted 8 lines
Quoted 3 lines

Good point! I am not sure that is the correct way to do it either, but maybe it's fine for PCI based devices. Adrian probably knows this better.

PCI / ACPI layers handle configuration of wakeup signaling, so a device will be put to the appropriate power state. It won't be forced to be "on" by registering a wakeup source.

Okay, I see.

So you are saying that PCI bus/layers are aware of the GPIO pin/irq being used for wakeup for the device in question?

Kind regards
Uffe

maintainer
#6 lore raw
Quoted 69 lines · Ulf Hansson
Quoted 59 lines · Adrian Hunter
Quoted 57 lines · Ulf Hansson
Quoted 55 lines · Bough Chen
Quoted 42 lines

-----Original Message-----
From: Ulf Hansson <[email protected]>
Sent: 2023年10月17日 19:35
To: Bough Chen <[email protected]>
Cc: [email protected]; [email protected]; dl-linux-imx <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: [PATCH] mmc: host: sdhci-esdhc-imx: let usr to config the wakeup for gpio cd pin through sysfs

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct

Quoted 18 lines · On Tue, 17 Oct 2023 at 09:26, <[email protected]> wrote:

From: Haibo Chen <[email protected]>

Currently default enable the gpio cd pin wakeup, this will waste power after system suspend if usr do not need this cd pin wakeup feature. Now let usr to config the wakeup for gpio cd pin through sysfs.

---
 drivers/mmc/host/sdhci-esdhc-imx.c | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)
drivers/mmc/host/sdhci-esdhc-imx.c +0 -0
index 40a6e2f8145a..2e46648344ba 100644
@@ -1797,9 +1797,12 @@ static int sdhci_esdhc_imx_probe(struct

platform_device *pdev)

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct platform_device *pdev)

Quoted 7 lines
drivers/mmc/host/sdhci-esdhc-imx.c +1 -2
        /*
         * Setup the wakeup capability here, let user to decide
         * whether need to enable this wakeup through sysfs interface.
+        * First check the SDIO device, second check the gpio CD pin.
         */
-       if ((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
-                       (host->mmc->pm_caps &

MMC_PM_WAKE_SDIO_IRQ))

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct · line 1803

Quoted 2 lines
drivers/mmc/host/sdhci-esdhc-imx.c +2 -0
+       if (((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
+                       (host->mmc->pm_caps &

MMC_PM_WAKE_SDIO_IRQ)) ||

Quoted 8 lines

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct platform_device *pdev) · line 1806

Quoted 3 lines
drivers/mmc/host/sdhci-esdhc-imx.c +2 -0
+           ((host->mmc->caps & MMC_CAP_CD_WAKE) &&
+                        host->mmc->slot.cd_irq >= 0))
                device_set_wakeup_capable(&pdev->dev, true);

If the wakeup is GPIO based, it doesn't mean that the device is wakeup capable, so this is wrong.

In principle this must be managed through the GPIO irqchip driver instead.

Yes, I know GPIO and USDHC are different device, combine GPIO wakeup to USDHC is not that reasonable. I send this patch because I notice there is similar code in sdhci-pci-core.c, refer to sdhci_pci_suspend_host()

Seems PCI subsystem combine GPIO wakeup to PCI controller.

Quoted 8 lines
Quoted 3 lines

Good point! I am not sure that is the correct way to do it either, but maybe it's fine for PCI based devices. Adrian probably knows this better.

PCI / ACPI layers handle configuration of wakeup signaling, so a device will be put to the appropriate power state. It won't be forced to be "on" by registering a wakeup source.

Quoted 4 lines

Okay, I see.

So you are saying that PCI bus/layers are aware of the GPIO pin/irq being used for wakeup for the device in question?

No, it is more like PCI / ACPI take care of most things but the wakeup framework provides sysfs wakeup API. We can add that and use it to determine if wakeup is enabled (by the user) to then enable the CD GPIO IRQ wakeup. That way, the sysfs wakeup API is on the correct device, and the user doesn't need to know anything about GPIO.

#7 lore raw
Quoted 76 lines · Adrian Hunter
Quoted 69 lines · Ulf Hansson
Quoted 59 lines · Adrian Hunter
Quoted 57 lines · Ulf Hansson
Quoted 55 lines · Bough Chen
Quoted 42 lines

-----Original Message-----
From: Ulf Hansson <[email protected]>
Sent: 2023年10月17日 19:35
To: Bough Chen <[email protected]>
Cc: [email protected]; [email protected]; dl-linux-imx <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: [PATCH] mmc: host: sdhci-esdhc-imx: let usr to config the wakeup for gpio cd pin through sysfs

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct

Quoted 18 lines · On Tue, 17 Oct 2023 at 09:26, <[email protected]> wrote:

From: Haibo Chen <[email protected]>

Currently default enable the gpio cd pin wakeup, this will waste power after system suspend if usr do not need this cd pin wakeup feature. Now let usr to config the wakeup for gpio cd pin through sysfs.

---
 drivers/mmc/host/sdhci-esdhc-imx.c | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)
drivers/mmc/host/sdhci-esdhc-imx.c +0 -0
index 40a6e2f8145a..2e46648344ba 100644
@@ -1797,9 +1797,12 @@ static int sdhci_esdhc_imx_probe(struct

platform_device *pdev)

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct · line 1802

Quoted 7 lines
drivers/mmc/host/sdhci-esdhc-imx.c +1 -2
        /*
         * Setup the wakeup capability here, let user to decide
         * whether need to enable this wakeup through sysfs interface.
+        * First check the SDIO device, second check the gpio CD pin.
         */
-       if ((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
-                       (host->mmc->pm_caps &

MMC_PM_WAKE_SDIO_IRQ))

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct · line 1803

Quoted 2 lines
drivers/mmc/host/sdhci-esdhc-imx.c +2 -0
+       if (((host->mmc->pm_caps & MMC_PM_KEEP_POWER) &&
+                       (host->mmc->pm_caps &

MMC_PM_WAKE_SDIO_IRQ)) ||

Quoted 8 lines

drivers/mmc/host/sdhci-esdhc-imx.c static int sdhci_esdhc_imx_probe(struct · line 1806

Quoted 3 lines
drivers/mmc/host/sdhci-esdhc-imx.c +2 -0
+           ((host->mmc->caps & MMC_CAP_CD_WAKE) &&
+                        host->mmc->slot.cd_irq >= 0))
                device_set_wakeup_capable(&pdev->dev, true);

If the wakeup is GPIO based, it doesn't mean that the device is wakeup capable, so this is wrong.

In principle this must be managed through the GPIO irqchip driver instead.

Yes, I know GPIO and USDHC are different device, combine GPIO wakeup to USDHC is not that reasonable. I send this patch because I notice there is similar code in sdhci-pci-core.c, refer to sdhci_pci_suspend_host()

Seems PCI subsystem combine GPIO wakeup to PCI controller.

Quoted 8 lines
Quoted 3 lines

Good point! I am not sure that is the correct way to do it either, but maybe it's fine for PCI based devices. Adrian probably knows this better.

PCI / ACPI layers handle configuration of wakeup signaling, so a device will be put to the appropriate power state. It won't be forced to be "on" by registering a wakeup source.

Quoted 4 lines

Okay, I see.

So you are saying that PCI bus/layers are aware of the GPIO pin/irq being used for wakeup for the device in question?

Quoted 7 lines

No, it is more like PCI / ACPI take care of most things but the wakeup framework provides sysfs wakeup API. We can add that and use it to determine if wakeup is enabled (by the user) to then enable the CD GPIO IRQ wakeup. That way, the sysfs wakeup API is on the correct device, and the user doesn't need to know anything about GPIO.

Okay, so it's really ACPI that is needed here too. Anyway, thanks for sharing!

Kind regards
Uffe