Skip to contentSkip to conversation

imx: power-domain: enable parent power domain

PATCH In review Patrick Wildt 2 participants 6 messages Jan 28, 2023 – Jan 29, 2023 Arm i.MX and Layerscape

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 imx: power-domain: enable parent power domain

The PCIe power domains are dependant on each other, which is why the device tree makes both PCIe controllers reference the PCIe1 power domain, which then depends on the PCIe2 power domain.

Enabling the parent power domain used to be part of the driver, but got partially lost in the rewrite. Add the enable call back to be able to power up PCIe2.

---
 drivers/power/domain/imx8m-power-domain.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/power/domain/imx8m-power-domain.c +3 -0
index 145f6ec0cd..d8e9ce3291 100644
@@ -330,6 +330,9 @@ static int imx8m_power_domain_on(struct power_domain *power_domain)
 	u32 pgc;
 	int ret;
 
+	if (pdata->has_pd)
+		power_domain_on(&pdata->pd);
+
 	if (pdata->clk.count) {
 		ret = clk_enable_bulk(&pdata->clk);
 		if (ret) {
Signature · 1 line
2.39.1
#1 lore raw
Quoted 18 lines · Patrick Wildt

The PCIe power domains are dependant on each other, which is why the device tree makes both PCIe controllers reference the PCIe1 power domain, which then depends on the PCIe2 power domain.

Enabling the parent power domain used to be part of the driver, but got partially lost in the rewrite. Add the enable call back to be able to power up PCIe2.

---
  drivers/power/domain/imx8m-power-domain.c | 3 +++
  1 file changed, 3 insertions(+)
drivers/power/domain/imx8m-power-domain.c +0 -0
index 145f6ec0cd..d8e9ce3291 100644

drivers/power/domain/imx8m-power-domain.c static int imx8m_power_domain_on(struct power_domain *power_domain) · line 338

Quoted 9 lines
drivers/power/domain/imx8m-power-domain.c +3 -0
@@ -330,6 +330,9 @@ static int imx8m_power_domain_on(struct power_domain *power_domain)
  	u32 pgc;
  	int ret;
 
+	if (pdata->has_pd)
+		power_domain_on(&pdata->pd);
+
  	if (pdata->clk.count) {
  		ret = clk_enable_bulk(&pdata->clk);
  		if (ret) {

One problem with this patch is that it does not turn the power domain back OFF in imx8m_power_domain_off().

However, the driver should not have to care about that. It is the uclass job to turn on any prerequisite power domains, which I believe happens in:

drivers/power/domain/power-domain-uclass.c
dev_power_domain_ctrl()

However, I suspect the code fails to recurse through more than one level of power domains and therefore doesn't enable the power domains all the way up the power domain tree. I also think this would be the fix -- recurse in dev_power_domain_ctrl() if there are upstream domains to enable, enable them in that recursive call, and then enable the current power domain using power_domain_on() .

Can you take a closer look at the uclass and the way it enables (or fail to) the upstream domains instead ?

#2 lore raw

Am Sun, Jan 29, 2023 at 02:46:36AM +0100 schrieb Marek Vasut:

Quoted 14 lines · Patrick Wildt
Quoted 13 lines

The PCIe power domains are dependant on each other, which is why the device tree makes both PCIe controllers reference the PCIe1 power domain, which then depends on the PCIe2 power domain.

Enabling the parent power domain used to be part of the driver, but got partially lost in the rewrite. Add the enable call back to be able to power up PCIe2.

---
  drivers/power/domain/imx8m-power-domain.c | 3 +++
  1 file changed, 3 insertions(+)
Quoted 16 lines

drivers/power/domain/imx8m-power-domain.c static int imx8m_power_domain_on(struct power_domain *power_domain) · line 337

Quoted 13 lines
drivers/power/domain/imx8m-power-domain.c +3 -0
index 145f6ec0cd..d8e9ce3291 100644
@@ -330,6 +330,9 @@ static int imx8m_power_domain_on(struct power_domain *power_domain)
  	u32 pgc;
  	int ret;
+	if (pdata->has_pd)
+		power_domain_on(&pdata->pd);
+
  	if (pdata->clk.count) {
  		ret = clk_enable_bulk(&pdata->clk);
  		if (ret) {

One problem with this patch is that it does not turn the power domain back OFF in imx8m_power_domain_off().

That's not true. That function already turns the power domain off, it just never enables them in the first place. The enabling was lost when the code was rewritten to not use smcc but the disabling is still there:

static int imx8m_power_domain_off(struct power_domain *power_domain)

{
	(...)
	if (pdata->has_pd)
		power_domain_off(&pdata->pd);

	return 0;
	(...)
}

What might be missing though is error handling to turn the power domain back off when enabling clocks/power up fails. In case this needs to be fixed elsewhere, then maybe we need to *remove* the call to power_domain_ off().

Quoted 13 lines

However, the driver should not have to care about that. It is the uclass job to turn on any prerequisite power domains, which I believe happens in:

drivers/power/domain/power-domain-uclass.c
dev_power_domain_ctrl()

However, I suspect the code fails to recurse through more than one level of power domains and therefore doesn't enable the power domains all the way up the power domain tree. I also think this would be the fix -- recurse in dev_power_domain_ctrl() if there are upstream domains to enable, enable them in that recursive call, and then enable the current power domain using power_domain_on() .

Quoted 2 lines

Can you take a closer look at the uclass and the way it enables (or fail to) the upstream domains instead ?

I guess one issue is that I was calling power_domain_on() directly instead of dev_power_domain_on() in my PCIe driver. And maybe I shouldn't even need to have the PCIe driver call that one function anyway, maybe uclass is supposed to turn on my PCIe controllers power domain automatically.

As you mentioned, there's no "parent handling" in dev_power_domain_ctrl(). Something like this could work, but I'm not sure that's better. Also I can't judge if there are any other side-effects of this. Still feels a bit like a layer violation, but maybe it's the right thing to do.

	for (i = 0; i < count; i++) {
		ret = power_domain_get_by_index(dev, &pd, i);
		if (ret)
			return ret;
		if (on) {
			ret = dev_power_domain_on(pd.dev);
			if (!ret)
				ret = power_domain_on(&pd);
		} else {
			ret = power_domain_off(&pd);
			if (!ret)
				ret = dev_power_domain_off(pd.dev);
		}
	}

I'll give this a go after some sleep.

Cheers,
Patrick

#3 lore raw
Quoted 37 lines · Patrick Wildt

Am Sun, Jan 29, 2023 at 02:46:36AM +0100 schrieb Marek Vasut:

Quoted 14 lines · Patrick Wildt
Quoted 13 lines

The PCIe power domains are dependant on each other, which is why the device tree makes both PCIe controllers reference the PCIe1 power domain, which then depends on the PCIe2 power domain.

Enabling the parent power domain used to be part of the driver, but got partially lost in the rewrite. Add the enable call back to be able to power up PCIe2.

---
   drivers/power/domain/imx8m-power-domain.c | 3 +++
   1 file changed, 3 insertions(+)
Quoted 16 lines

drivers/power/domain/imx8m-power-domain.c static int imx8m_power_domain_on(struct power_domain *power_domain) · line 337

Quoted 13 lines
drivers/power/domain/imx8m-power-domain.c +3 -0
index 145f6ec0cd..d8e9ce3291 100644
@@ -330,6 +330,9 @@ static int imx8m_power_domain_on(struct power_domain *power_domain)
   	u32 pgc;
   	int ret;
+	if (pdata->has_pd)
+		power_domain_on(&pdata->pd);
+
   	if (pdata->clk.count) {
   		ret = clk_enable_bulk(&pdata->clk);
   		if (ret) {

One problem with this patch is that it does not turn the power domain back OFF in imx8m_power_domain_off().

That's not true. That function already turns the power domain off, it just never enables them in the first place. The enabling was lost when the code was rewritten to not use smcc but the disabling is still there:

static int imx8m_power_domain_off(struct power_domain *power_domain)

Quoted 8 lines
{
	(...)
	if (pdata->has_pd)
		power_domain_off(&pdata->pd);

	return 0;
	(...)
}

Doh, sorry, I did miss that one. Indeed, the imx power domain driver is not symmetrical now.

Quoted 18 lines

What might be missing though is error handling to turn the power domain back off when enabling clocks/power up fails. In case this needs to be fixed elsewhere, then maybe we need to *remove* the call to power_domain_ off().

Quoted 13 lines

However, the driver should not have to care about that. It is the uclass job to turn on any prerequisite power domains, which I believe happens in:

drivers/power/domain/power-domain-uclass.c
dev_power_domain_ctrl()

However, I suspect the code fails to recurse through more than one level of power domains and therefore doesn't enable the power domains all the way up the power domain tree. I also think this would be the fix -- recurse in dev_power_domain_ctrl() if there are upstream domains to enable, enable them in that recursive call, and then enable the current power domain using power_domain_on() .

Quoted 5 lines
Quoted 2 lines

Can you take a closer look at the uclass and the way it enables (or fail to) the upstream domains instead ?

I guess one issue is that I was calling power_domain_on() directly instead of dev_power_domain_on() in my PCIe driver.

That sounds like a good idea.

Quoted 3 lines

And maybe I shouldn't even need to
have the PCIe driver call that one function anyway, maybe uclass is supposed to turn on my PCIe controllers power domain automatically.

device_probe() does call dev_power_domain_on(), so that sounds like a good idea too.

Quoted 4 lines

As you mentioned, there's no "parent handling" in dev_power_domain_ctrl(). Something like this could work, but I'm not sure that's better. Also I can't judge if there are any other side-effects of this. Still feels a bit like a layer violation

How so ?

Quoted 17 lines

, but maybe it's the right thing to do.

	for (i = 0; i < count; i++) {
		ret = power_domain_get_by_index(dev, &pd, i);
		if (ret)
			return ret;
		if (on) {
			ret = dev_power_domain_on(pd.dev);
			if (!ret)
				ret = power_domain_on(&pd);
		} else {
			ret = power_domain_off(&pd);
			if (!ret)
				ret = dev_power_domain_off(pd.dev);
		}
	}
Quoted 1 line

I'll give this a go after some sleep.

I think this makes sense -- the domain should enable all its upstream domains first, before enabling itself.

It might also make sense to remove the aforementioned power_domain_off() from imx8m-power-domain.c once the dev_power_domain_ctrl() is fixed.

#4 lore raw
Quoted 107 lines · Marek Vasut
Quoted 37 lines · Patrick Wildt

Am Sun, Jan 29, 2023 at 02:46:36AM +0100 schrieb Marek Vasut:

Quoted 14 lines · Patrick Wildt
Quoted 13 lines

The PCIe power domains are dependant on each other, which is why the device tree makes both PCIe controllers reference the PCIe1 power domain, which then depends on the PCIe2 power domain.

Enabling the parent power domain used to be part of the driver, but got partially lost in the rewrite. Add the enable call back to be able to power up PCIe2.

---
   drivers/power/domain/imx8m-power-domain.c | 3 +++
   1 file changed, 3 insertions(+)
Quoted 16 lines

drivers/power/domain/imx8m-power-domain.c static int imx8m_power_domain_on(struct power_domain *power_domain) · line 337

Quoted 13 lines
drivers/power/domain/imx8m-power-domain.c +3 -0
index 145f6ec0cd..d8e9ce3291 100644
@@ -330,6 +330,9 @@ static int imx8m_power_domain_on(struct power_domain *power_domain)
   	u32 pgc;
   	int ret;
+	if (pdata->has_pd)
+		power_domain_on(&pdata->pd);
+
   	if (pdata->clk.count) {
   		ret = clk_enable_bulk(&pdata->clk);
   		if (ret) {

One problem with this patch is that it does not turn the power domain back OFF in imx8m_power_domain_off().

That's not true. That function already turns the power domain off, it just never enables them in the first place. The enabling was lost when the code was rewritten to not use smcc but the disabling is still there:

static int imx8m_power_domain_off(struct power_domain *power_domain)

Quoted 8 lines
{
	(...)
	if (pdata->has_pd)
		power_domain_off(&pdata->pd);

	return 0;
	(...)
}

Doh, sorry, I did miss that one. Indeed, the imx power domain driver is not symmetrical now.

Quoted 18 lines

What might be missing though is error handling to turn the power domain back off when enabling clocks/power up fails. In case this needs to be fixed elsewhere, then maybe we need to *remove* the call to power_domain_ off().

Quoted 13 lines

However, the driver should not have to care about that. It is the uclass job to turn on any prerequisite power domains, which I believe happens in:

drivers/power/domain/power-domain-uclass.c
dev_power_domain_ctrl()

However, I suspect the code fails to recurse through more than one level of power domains and therefore doesn't enable the power domains all the way up the power domain tree. I also think this would be the fix -- recurse in dev_power_domain_ctrl() if there are upstream domains to enable, enable them in that recursive call, and then enable the current power domain using power_domain_on() .

Quoted 5 lines
Quoted 2 lines

Can you take a closer look at the uclass and the way it enables (or fail to) the upstream domains instead ?

I guess one issue is that I was calling power_domain_on() directly instead of dev_power_domain_on() in my PCIe driver.

That sounds like a good idea.

Quoted 3 lines

And maybe I shouldn't even need to
have the PCIe driver call that one function anyway, maybe uclass is supposed to turn on my PCIe controllers power domain automatically.

device_probe() does call dev_power_domain_on(), so that sounds like a good idea too.

Quoted 4 lines

As you mentioned, there's no "parent handling" in dev_power_domain_ctrl(). Something like this could work, but I'm not sure that's better. Also I can't judge if there are any other side-effects of this. Still feels a bit like a layer violation

How so ?

Quoted 17 lines

, but maybe it's the right thing to do.

	for (i = 0; i < count; i++) {
		ret = power_domain_get_by_index(dev, &pd, i);
		if (ret)
			return ret;
		if (on) {
			ret = dev_power_domain_on(pd.dev);
			if (!ret)
				ret = power_domain_on(&pd);
		} else {
			ret = power_domain_off(&pd);
			if (!ret)
				ret = dev_power_domain_off(pd.dev);
		}
	}
Quoted 7 lines
Quoted 1 line

I'll give this a go after some sleep.

I think this makes sense -- the domain should enable all its upstream domains first, before enabling itself.

It might also make sense to remove the aforementioned power_domain_off() from imx8m-power-domain.c once the dev_power_domain_ctrl() is fixed.

Just tested, this diff works for me. Should I send it as a patch to the list?

One thing that worries me is that there's no refcounting to make sure that a parent that is used twice doesn't get disabled. That might be something to follow-up on.

drivers/power/domain/power-domain-uclass.c +8 -3
index f6286c70c1..3b3caf6164 100644
@@ -138,10 +138,15 @@ static int dev_power_domain_ctrl(struct udevice *dev, bool on)
 		ret = power_domain_get_by_index(dev, &pd, i);
 		if (ret)
 			return ret;
-		if (on)
-			ret = power_domain_on(&pd);
-		else
+		if (on) {
+			ret = dev_power_domain_on(pd.dev);
+			if (!ret)
+				ret = power_domain_on(&pd);
+		} else {
 			ret = power_domain_off(&pd);
+			if (!ret)
+				ret = dev_power_domain_off(pd.dev);
+		}
 	}
 
 	/*
#5 lore raw
Quoted 110 lines · Patrick Wildt
Quoted 109 lines · Marek Vasut
Quoted 37 lines · Patrick Wildt

Am Sun, Jan 29, 2023 at 02:46:36AM +0100 schrieb Marek Vasut:

Quoted 14 lines · Patrick Wildt
Quoted 13 lines

The PCIe power domains are dependant on each other, which is why the device tree makes both PCIe controllers reference the PCIe1 power domain, which then depends on the PCIe2 power domain.

Enabling the parent power domain used to be part of the driver, but got partially lost in the rewrite. Add the enable call back to be able to power up PCIe2.

---
    drivers/power/domain/imx8m-power-domain.c | 3 +++
    1 file changed, 3 insertions(+)
Quoted 16 lines

drivers/power/domain/imx8m-power-domain.c static int imx8m_power_domain_on(struct power_domain *power_domain) · line 337

Quoted 13 lines
drivers/power/domain/imx8m-power-domain.c +3 -0
index 145f6ec0cd..d8e9ce3291 100644
@@ -330,6 +330,9 @@ static int imx8m_power_domain_on(struct power_domain *power_domain)
    	u32 pgc;
    	int ret;
+	if (pdata->has_pd)
+		power_domain_on(&pdata->pd);
+
    	if (pdata->clk.count) {
    		ret = clk_enable_bulk(&pdata->clk);
    		if (ret) {

One problem with this patch is that it does not turn the power domain back OFF in imx8m_power_domain_off().

That's not true. That function already turns the power domain off, it just never enables them in the first place. The enabling was lost when the code was rewritten to not use smcc but the disabling is still there:

static int imx8m_power_domain_off(struct power_domain *power_domain)

Quoted 8 lines
{
	(...)
	if (pdata->has_pd)
		power_domain_off(&pdata->pd);

	return 0;
	(...)
}

Doh, sorry, I did miss that one. Indeed, the imx power domain driver is not symmetrical now.

Quoted 18 lines

What might be missing though is error handling to turn the power domain back off when enabling clocks/power up fails. In case this needs to be fixed elsewhere, then maybe we need to *remove* the call to power_domain_ off().

Quoted 13 lines

However, the driver should not have to care about that. It is the uclass job to turn on any prerequisite power domains, which I believe happens in:

drivers/power/domain/power-domain-uclass.c
dev_power_domain_ctrl()

However, I suspect the code fails to recurse through more than one level of power domains and therefore doesn't enable the power domains all the way up the power domain tree. I also think this would be the fix -- recurse in dev_power_domain_ctrl() if there are upstream domains to enable, enable them in that recursive call, and then enable the current power domain using power_domain_on() .

Quoted 5 lines
Quoted 2 lines

Can you take a closer look at the uclass and the way it enables (or fail to) the upstream domains instead ?

I guess one issue is that I was calling power_domain_on() directly instead of dev_power_domain_on() in my PCIe driver.

That sounds like a good idea.

Quoted 3 lines

And maybe I shouldn't even need to
have the PCIe driver call that one function anyway, maybe uclass is supposed to turn on my PCIe controllers power domain automatically.

device_probe() does call dev_power_domain_on(), so that sounds like a good idea too.

Quoted 4 lines

As you mentioned, there's no "parent handling" in dev_power_domain_ctrl(). Something like this could work, but I'm not sure that's better. Also I can't judge if there are any other side-effects of this. Still feels a bit like a layer violation

How so ?

Quoted 18 lines

, but maybe it's the right thing to do.

	for (i = 0; i < count; i++) {
		ret = power_domain_get_by_index(dev, &pd, i);
		if (ret)
			return ret;
		if (on) {
			ret = dev_power_domain_on(pd.dev);
			if (!ret)
				ret = power_domain_on(&pd);
		} else {
			ret = power_domain_off(&pd);
			if (!ret)
				ret = dev_power_domain_off(pd.dev);
		}
	}

I'll give this a go after some sleep.

Quoted 8 lines
Quoted 5 lines

I think this makes sense -- the domain should enable all its upstream domains first, before enabling itself.

It might also make sense to remove the aforementioned power_domain_off() from imx8m-power-domain.c once the dev_power_domain_ctrl() is fixed.

Just tested, this diff works for me. Should I send it as a patch to the list?

Yes please !

Quoted 3 lines

One thing that worries me is that there's no refcounting to make sure that a parent that is used twice doesn't get disabled. That might be something to follow-up on.

I agree. Could you try and add that too ? Separate patch is fine.

drivers/power/domain/power-domain-uclass.c static int dev_power_domain_ctrl(struct udevice *dev, bool on) · line 142

Quoted 13 lines
drivers/power/domain/power-domain-uclass.c +2 -3
index f6286c70c1..3b3caf6164 100644
@@ -138,10 +138,15 @@ static int dev_power_domain_ctrl(struct udevice *dev, bool on)
  		ret = power_domain_get_by_index(dev, &pd, i);
  		if (ret)
  			return ret;
-		if (on)
-			ret = power_domain_on(&pd);
-		else
+		if (on) {
+			ret = dev_power_domain_on(pd.dev);

You might want to add a fail path here:

if (ret)
goto err_power_on;

...
err_poewr_on:
... un-do the power on ...

Quoted 10 lines
drivers/power/domain/power-domain-uclass.c +6 -0
+			if (!ret)
+				ret = power_domain_on(&pd);
+		} else {
  			ret = power_domain_off(&pd);
+			if (!ret)
+				ret = dev_power_domain_off(pd.dev);
+		}
  	}
 
  	/*

[...]