Skip to contentSkip to conversation

i2c: tegra: Set ACPI node as primary fwnode

PATCH Applied Akhil R 4 participants 8 messages Nov 17, 2022 – Dec 1, 2022

Subsystems (3)
lore
Share & tools
lore

Akhil R

author
#0 lore raw

PATCH 1/1 i2c: tegra: Set ACPI node as primary fwnode

Set ACPI node as the primary fwnode of I2C adapter to allow enumeration of child devices from the ACPI table

---
 drivers/i2c/busses/i2c-tegra.c | 1 +
 1 file changed, 1 insertion(+)
drivers/i2c/busses/i2c-tegra.c +1 -0
index 954022c04cc4..69c9ae161bbe 100644
@@ -1826,6 +1826,7 @@ static int tegra_i2c_probe(struct platform_device *pdev)
 	i2c_dev->adapter.class = I2C_CLASS_DEPRECATED;
 	i2c_dev->adapter.algo = &tegra_i2c_algo;
 	i2c_dev->adapter.nr = pdev->id;
+	ACPI_COMPANION_SET(&i2c_dev->adapter.dev, ACPI_COMPANION(&pdev->dev));
 
 	if (i2c_dev->hw->supports_bus_clear)
 		i2c_dev->adapter.bus_recovery_info = &tegra_i2c_recovery_info;
Signature · 1 line
2.17.1
maintainer
#2 lore raw
In reply to Akhil R
Quoted 14 lines · Akhil R

Set ACPI node as the primary fwnode of I2C adapter to allow enumeration of child devices from the ACPI table

---
  drivers/i2c/busses/i2c-tegra.c | 1 +
  1 file changed, 1 insertion(+)
drivers/i2c/busses/i2c-tegra.c +0 -0
index 954022c04cc4..69c9ae161bbe 100644

drivers/i2c/busses/i2c-tegra.c static int tegra_i2c_probe(struct platform_device *pdev) · line 1832

Quoted 7 lines
drivers/i2c/busses/i2c-tegra.c +1 -0
@@ -1826,6 +1826,7 @@ static int tegra_i2c_probe(struct platform_device *pdev)
  	i2c_dev->adapter.class = I2C_CLASS_DEPRECATED;
  	i2c_dev->adapter.algo = &tegra_i2c_algo;
  	i2c_dev->adapter.nr = pdev->id;
+	ACPI_COMPANION_SET(&i2c_dev->adapter.dev, ACPI_COMPANION(&pdev->dev));
 
  	if (i2c_dev->hw->supports_bus_clear)
  		i2c_dev->adapter.bus_recovery_info = &tegra_i2c_recovery_info;

Do we always want to set as the primary fwnode even when booting with device-tree? I some other drivers do, but I also see some others ...

if (has_acpi_companion(dev))

        ACPI_COMPANION_SET(&i2c_dev->adapter.dev,
                           ACPI_COMPANION(&pdev->dev));

It would be nice to know why it is OK to always do this even for device-tree because it is not clear to me.

Jon

Signature · 1 line
nvpublic
#3 lore raw
Quoted 27 lines · Jon Hunter
Quoted 14 lines · Akhil R

Set ACPI node as the primary fwnode of I2C adapter to allow enumeration of child devices from the ACPI table

---
  drivers/i2c/busses/i2c-tegra.c | 1 +
  1 file changed, 1 insertion(+)
drivers/i2c/busses/i2c-tegra.c +0 -0
index 954022c04cc4..69c9ae161bbe 100644

drivers/i2c/busses/i2c-tegra.c static int tegra_i2c_probe(struct platform_device *pdev) · line 1831

Quoted 6 lines
drivers/i2c/busses/i2c-tegra.c +1 -0
@@ -1826,6 +1826,7 @@ static int tegra_i2c_probe(struct platform_device *pdev)
  	i2c_dev->adapter.class = I2C_CLASS_DEPRECATED;
  	i2c_dev->adapter.algo = &tegra_i2c_algo;
  	i2c_dev->adapter.nr = pdev->id;
+	ACPI_COMPANION_SET(&i2c_dev->adapter.dev, ACPI_COMPANION(&pdev->dev));
  	if (i2c_dev->hw->supports_bus_clear)
  		i2c_dev->adapter.bus_recovery_info = &tegra_i2c_recovery_info;

Do we always want to set as the primary fwnode even when booting with device-tree? I some other drivers do, but I also see some others ...

Quoted 6 lines

if (has_acpi_companion(dev))

        ACPI_COMPANION_SET(&i2c_dev->adapter.dev,
                           ACPI_COMPANION(&pdev->dev));

It would be nice to know why it is OK to always do this even for device-tree because it is not clear to me.

ACPI_COMPANION() returns NULL if there is no ACPI companion, which will cause ACPI_COMPANION_SET() to set the primary fwnode to NULL. If I read the code for set_primary_fwnode() correctly, that's essentially a no-op for DT devices.

I guess that the extra check might save a few cycles by not having to run through all the various conditionals, but it seems a rather minor saving.

Either way is fine with me, though.

Thierry

maintainer
#4 lore raw
Quoted 32 lines · Thierry Reding
Quoted 31 lines · Jon Hunter
Quoted 14 lines · Akhil R

Set ACPI node as the primary fwnode of I2C adapter to allow enumeration of child devices from the ACPI table

---
   drivers/i2c/busses/i2c-tegra.c | 1 +
   1 file changed, 1 insertion(+)
drivers/i2c/busses/i2c-tegra.c +0 -0
index 954022c04cc4..69c9ae161bbe 100644

drivers/i2c/busses/i2c-tegra.c static int tegra_i2c_probe(struct platform_device *pdev) · line 1831

Quoted 6 lines
drivers/i2c/busses/i2c-tegra.c +1 -0
@@ -1826,6 +1826,7 @@ static int tegra_i2c_probe(struct platform_device *pdev)
   	i2c_dev->adapter.class = I2C_CLASS_DEPRECATED;
   	i2c_dev->adapter.algo = &tegra_i2c_algo;
   	i2c_dev->adapter.nr = pdev->id;
+	ACPI_COMPANION_SET(&i2c_dev->adapter.dev, ACPI_COMPANION(&pdev->dev));
   	if (i2c_dev->hw->supports_bus_clear)
   		i2c_dev->adapter.bus_recovery_info = &tegra_i2c_recovery_info;

Do we always want to set as the primary fwnode even when booting with device-tree? I some other drivers do, but I also see some others ...

  if (has_acpi_companion(dev))
         ACPI_COMPANION_SET(&i2c_dev->adapter.dev,
                            ACPI_COMPANION(&pdev->dev));
Quoted 7 lines
Quoted 2 lines

It would be nice to know why it is OK to always do this even for device-tree because it is not clear to me.

ACPI_COMPANION() returns NULL if there is no ACPI companion, which will cause ACPI_COMPANION_SET() to set the primary fwnode to NULL. If I read the code for set_primary_fwnode() correctly, that's essentially a no-op for DT devices.

Yes it does, but doesn't it is not clear to me if it is a good idea to pass NULL to set_primary_fwnode(). It does seem to handle this but my biggest gripe is the lack of explanation in the commit message why this is OK.

Jon

Signature · 1 line
nvpublic
author
#5 lore raw
Quoted 44 lines · Thierry Reding
Quoted 12 lines · Jon Hunter
Quoted 11 lines · Akhil R
Quoted 9 lines

Set ACPI node as the primary fwnode of I2C adapter to allow enumeration of child devices from the ACPI table

---
   drivers/i2c/busses/i2c-tegra.c | 1 +
   1 file changed, 1 insertion(+)
Quoted 5 lines
Quoted 5 lines

drivers/i2c/busses/i2c-tegra.c static int tegra_i2c_probe(struct platform_device

Quoted 5 lines
drivers/i2c/busses/i2c-tegra.c +0 -0
index 954022c04cc4..69c9ae161bbe 100644
@@ -1826,6 +1826,7 @@ static int tegra_i2c_probe(struct platform_device

*pdev)

Quoted 4 lines
Quoted 4 lines

drivers/i2c/busses/i2c-tegra.c static int tegra_i2c_probe(struct platform_device *pdev)

Quoted 4 lines
drivers/i2c/busses/i2c-tegra.c +1 -0
   	i2c_dev->adapter.class = I2C_CLASS_DEPRECATED;
   	i2c_dev->adapter.algo = &tegra_i2c_algo;
   	i2c_dev->adapter.nr = pdev->id;
+	ACPI_COMPANION_SET(&i2c_dev->adapter.dev,

ACPI_COMPANION(&pdev->dev));

Quoted 2 lines
Quoted 2 lines

drivers/i2c/busses/i2c-tegra.c static int tegra_i2c_probe(struct platform_device · line 1831

Quoted 2 lines
drivers/i2c/busses/i2c-tegra.c +0 -0
   	if (i2c_dev->hw->supports_bus_clear)
   		i2c_dev->adapter.bus_recovery_info =

&tegra_i2c_recovery_info;

Quoted 9 lines
Quoted 9 lines

Do we always want to set as the primary fwnode even when booting with device-tree? I some other drivers do, but I also see some others ...

  if (has_acpi_companion(dev))
         ACPI_COMPANION_SET(&i2c_dev->adapter.dev,
                            ACPI_COMPANION(&pdev->dev));
Quoted 7 lines
Quoted 2 lines

It would be nice to know why it is OK to always do this even for device-tree because it is not clear to me.

ACPI_COMPANION() returns NULL if there is no ACPI companion, which will cause ACPI_COMPANION_SET() to set the primary fwnode to NULL. If I read the code for set_primary_fwnode() correctly, that's essentially a no-op for DT devices.

Quoted 4 lines

Yes it does, but doesn't it is not clear to me if it is a good idea to pass NULL to set_primary_fwnode(). It does seem to handle this but my biggest gripe is the lack of explanation in the commit message why this is OK.

I saw ACPI_COMPANION_SET() as an empty function if CONFIG_ACPI is not set. Yes, I agree that I should have mentioned this in the commit message. Shall I send a v2 with the details added in the commit description?

Regards,
Akhil

maintainer
#6 lore raw
Quoted 44 lines · Akhil R
Quoted 44 lines · Thierry Reding
Quoted 12 lines · Jon Hunter
Quoted 11 lines · Akhil R
Quoted 9 lines

Set ACPI node as the primary fwnode of I2C adapter to allow enumeration of child devices from the ACPI table

---
    drivers/i2c/busses/i2c-tegra.c | 1 +
    1 file changed, 1 insertion(+)
Quoted 5 lines
Quoted 5 lines

drivers/i2c/busses/i2c-tegra.c static int tegra_i2c_probe(struct platform_device

Quoted 5 lines
drivers/i2c/busses/i2c-tegra.c +0 -0
index 954022c04cc4..69c9ae161bbe 100644
@@ -1826,6 +1826,7 @@ static int tegra_i2c_probe(struct platform_device

*pdev)

Quoted 4 lines
Quoted 4 lines

drivers/i2c/busses/i2c-tegra.c static int tegra_i2c_probe(struct platform_device *pdev)

Quoted 4 lines
drivers/i2c/busses/i2c-tegra.c +1 -0
    	i2c_dev->adapter.class = I2C_CLASS_DEPRECATED;
    	i2c_dev->adapter.algo = &tegra_i2c_algo;
    	i2c_dev->adapter.nr = pdev->id;
+	ACPI_COMPANION_SET(&i2c_dev->adapter.dev,

ACPI_COMPANION(&pdev->dev));

Quoted 2 lines
Quoted 2 lines

drivers/i2c/busses/i2c-tegra.c static int tegra_i2c_probe(struct platform_device · line 1831

Quoted 2 lines
drivers/i2c/busses/i2c-tegra.c +0 -0
    	if (i2c_dev->hw->supports_bus_clear)
    		i2c_dev->adapter.bus_recovery_info =

&tegra_i2c_recovery_info;

Quoted 16 lines
Quoted 11 lines

Do we always want to set as the primary fwnode even when booting with device-tree? I some other drivers do, but I also see some others ...

   if (has_acpi_companion(dev))
          ACPI_COMPANION_SET(&i2c_dev->adapter.dev,
                             ACPI_COMPANION(&pdev->dev));

It would be nice to know why it is OK to always do this even for device-tree because it is not clear to me.

ACPI_COMPANION() returns NULL if there is no ACPI companion, which will cause ACPI_COMPANION_SET() to set the primary fwnode to NULL. If I read the code for set_primary_fwnode() correctly, that's essentially a no-op for DT devices.

Quoted 5 lines
Quoted 4 lines

Yes it does, but doesn't it is not clear to me if it is a good idea to pass NULL to set_primary_fwnode(). It does seem to handle this but my biggest gripe is the lack of explanation in the commit message why this is OK.

I saw ACPI_COMPANION_SET() as an empty function if CONFIG_ACPI is not set.

That's not the issue. By default CONFIG_ACPI is enabled for arm64 but for Tegra we typically boot with device-tree. So I was more concerned about the case where ACPI_COMPANION_SET() is not an empty function.

Quoted 2 lines

Yes, I agree that I should have mentioned this in the commit message. Shall I send a v2 with the details added in the commit description?

No need, especially as Thierry has already applied. I am not familiar with this function and primary/secondary fwnodes so wanted to understand there is no issue for device tree.

Jon

Signature · 1 line
nvpublic