i2c/designware: enable i2c controller to suspend/resume asynchronously

PATCH In review Fu, Zhonghui 3 participants 5 messages Dec 24, 2015 – Jan 15, 2016

Subsystems (3)

I2C subsystem host drivers Intel LPSS (Low Power SubSystem) drivers Synopsys DesignWare I2C driver

lore

Share & tools

lore

Fu, Zhonghui

author

2015-12-24 14:30 UTC #0 lore raw

PATCH 1/1 i2c/designware: enable i2c controller to suspend/resume asynchronously

Now, PM core supports asynchronous suspend/resume mode for devices during system suspend/resume, and the power state transition of one device may be completed in separate kernel thread. PM core ensures all power state transition dependency between devices. This patch enables designware i2c controllers to suspend/resume asynchronously. This will take advantage of multicore and improve system suspend/resume speed. After enabling all i2c devices, i2c adapters and i2c controllers on ASUS T100TA tablet, the system suspend-to-idle time is reduced to about 510ms from 750ms, and the system resume time is reduced to about 790ms from 900ms.

Signed-off-by Zhonghui Fu <[email protected]>

---
 drivers/i2c/busses/i2c-designware-platdrv.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

drivers/i2c/busses/i2c-designware-platdrv.c +1 -0

index 6b00061..395130b 100644
@@ -230,6 +230,7 @@ static int dw_i2c_plat_probe(struct platform_device *pdev)
 	}
 
 	adap = &dev->adapter;
+	device_enable_async_suspend(&pdev->dev);
 	adap->owner = THIS_MODULE;
 	adap->class = I2C_CLASS_DEPRECATED;
 	ACPI_COMPANION_SET(&adap->dev, ACPI_COMPANION(&pdev->dev));

-- 1.7.1

Jarkko Nikula

2016-01-05 08:53 UTC #1 lore raw

Quoted 11 lines · Zhonghui

Now, PM core supports asynchronous suspend/resume mode for devices during system suspend/resume, and the power state transition of one device may be completed in separate kernel thread. PM core ensures all power state transition dependency between devices. This patch enables designware i2c controllers to suspend/resume asynchronously. This will take advantage of multicore and improve system suspend/resume speed. After enabling all i2c devices, i2c adapters and i2c controllers on ASUS T100TA tablet, the system suspend-to-idle time is reduced to about 510ms from 750ms, and the system resume time is reduced to about 790ms from 900ms.

Nice reduction :-)

drivers/i2c/busses/i2c-designware-platdrv.c static int dw_i2c_plat_probe(struct platform_device *pdev) · line 236

Quoted 12 lines

drivers/i2c/busses/i2c-designware-platdrv.c +1 -0

index 6b00061..395130b 100644
@@ -230,6 +230,7 @@ static int dw_i2c_plat_probe(struct platform_device *pdev)
  	}
 
  	adap = &dev->adapter;
+	device_enable_async_suspend(&pdev->dev);
  	adap->owner = THIS_MODULE;
  	adap->class = I2C_CLASS_DEPRECATED;
  	ACPI_COMPANION_SET(&adap->dev, ACPI_COMPANION(&pdev->dev));

Does device_enable_async_suspend() need to be called before enabling runtime PM? I suppose not since there appears to have also related sysfs node for toggling it runtime.

I'm thinking if you could move the device_enable_async_suspend() call into drivers/i2c/busses/i2c-designware-core.c: i2c_dw_probe() and then also PCI enumerated adapter could take advantage of it.

Signature · 1 line

Jarkko

Andy Shevchenko

reviewer

2016-01-05 12:14 UTC #2 lore raw

Quoted 42 lines · Jarkko Nikula

Hi
Quoted 7 lines · Zhonghui
Now, PM core supports asynchronous suspend/resume mode for devices during system suspend/resume, and the power state transition of one device may be completed in separate kernel thread. PM core ensures all power state transition dependency between devices. This patch enables designware i2c controllers to suspend/resume asynchronously.
This will take advantage of multicore and improve system
Quoted 9 lines
suspend/resume
speed. After enabling all i2c devices, i2c adapters and i2c controllers
on ASUS T100TA tablet, the system suspend-to-idle time is reduced to
about 510ms from 750ms, and the system resume time is reduced to about
790ms from 900ms.
Nice reduction :-)
Quoted 13 lines
drivers/i2c/busses/i2c-designware-platdrv.c +1 -0
index 6b00061..395130b 100644
@@ -230,6 +230,7 @@ static int dw_i2c_plat_probe(struct
 platform_device *pdev)
   	}
 
   	adap = &dev->adapter;
+	device_enable_async_suspend(&pdev->dev);
   	adap->owner = THIS_MODULE;
   	adap->class = I2C_CLASS_DEPRECATED;
Quoted 2 lines
ACPI_COMPANION_SET(&adap->dev, ACPI_COMPANION(&pdev-
Quoted 1 line
dev));
Does device_enable_async_suspend() need to be called before enabling runtime PM? I suppose not since there appears to have also related sysfs
node for toggling it runtime.

Quoted 4 lines

I'm thinking if you could move the device_enable_async_suspend() call into drivers/i2c/busses/i2c-designware-core.c: i2c_dw_probe() and then
also PCI enumerated adapter could take advantage of it.

I concern about Intel BayTrail-T / Braswell / CherryTrail cases, since we have non-trivial PM for LPSS there. Zhonghui, have you a chance to stress test this on platforms based on mentioned SoCs?

Signature · 2 lines

Andy Shevchenko <[email protected]>
Intel Finland Oy

Fu, Zhonghui

author

2016-01-15 06:42 UTC #4 lore raw

In reply to Jarkko Nikula

Quoted 31 lines · Jarkko Nikula

Hi
Quoted 11 lines · Zhonghui
Now, PM core supports asynchronous suspend/resume mode for devices during system suspend/resume, and the power state transition of one device may be completed in separate kernel thread. PM core ensures all power state transition dependency between devices. This patch enables designware i2c controllers to suspend/resume asynchronously. This will take advantage of multicore and improve system suspend/resume speed. After enabling all i2c devices, i2c adapters and i2c controllers on ASUS T100TA tablet, the system suspend-to-idle time is reduced to about 510ms from 750ms, and the system resume time is reduced to about 790ms from 900ms.
Nice reduction :-)
drivers/i2c/busses/i2c-designware-platdrv.c static int dw_i2c_plat_probe(struct platform_device *pdev) · line 236
Quoted 12 lines
drivers/i2c/busses/i2c-designware-platdrv.c +1 -0
index 6b00061..395130b 100644
@@ -230,6 +230,7 @@ static int dw_i2c_plat_probe(struct platform_device *pdev)
      }
 
      adap = &dev->adapter;
+    device_enable_async_suspend(&pdev->dev);
      adap->owner = THIS_MODULE;
      adap->class = I2C_CLASS_DEPRECATED;
      ACPI_COMPANION_SET(&adap->dev, ACPI_COMPANION(&pdev->dev));
Does device_enable_async_suspend() need to be called before enabling runtime PM? I suppose not since there appears to have also related sysfs node for toggling it runtime.

Quoted 1 line

I'm thinking if you could move the device_enable_async_suspend() call into drivers/i2c/busses/i2c-designware-core.c: i2c_dw_probe() and then also PCI enumerated adapter could take advantage of it.

Because of long leave, so sorry for late reply.

Your proposal is right. I have submitted a new patch according to your comments - "[PATCH v2] i2c/designware: enable i2c controller to suspend/resume asynchronously".

Thanks,
Zhonghui

Fu, Zhonghui

author

2016-01-15 07:01 UTC #3 lore raw

In reply to Andy Shevchenko

Quoted 42 lines · Andy Shevchenko

Quoted 41 lines · Jarkko Nikula
Hi
Quoted 7 lines · Zhonghui
Now, PM core supports asynchronous suspend/resume mode for devices during system suspend/resume, and the power state transition of one device may be completed in separate kernel thread. PM core ensures all power state transition dependency between devices. This patch enables designware i2c controllers to suspend/resume asynchronously.
This will take advantage of multicore and improve system
Quoted 9 lines
suspend/resume
speed. After enabling all i2c devices, i2c adapters and i2c controllers
on ASUS T100TA tablet, the system suspend-to-idle time is reduced to
about 510ms from 750ms, and the system resume time is reduced to about
790ms from 900ms.
Nice reduction :-)
Quoted 13 lines
drivers/i2c/busses/i2c-designware-platdrv.c +1 -0
index 6b00061..395130b 100644
@@ -230,6 +230,7 @@ static int dw_i2c_plat_probe(struct
 platform_device *pdev)
  	}
 
  	adap = &dev->adapter;
+	device_enable_async_suspend(&pdev->dev);
  	adap->owner = THIS_MODULE;
  	adap->class = I2C_CLASS_DEPRECATED;
Quoted 2 lines
ACPI_COMPANION_SET(&adap->dev, ACPI_COMPANION(&pdev-
Quoted 1 line
dev));
Does device_enable_async_suspend() need to be called before enabling runtime PM? I suppose not since there appears to have also related sysfs
node for toggling it runtime.

Quoted 7 lines

Quoted 4 lines
I'm thinking if you could move the device_enable_async_suspend() call into drivers/i2c/busses/i2c-designware-core.c: i2c_dw_probe() and then
also PCI enumerated adapter could take advantage of it.
I concern about Intel BayTrail-T / Braswell / CherryTrail cases, since we have non-trivial PM for LPSS there. Zhonghui, have you a chance to stress test this on platforms based on mentioned SoCs?

Because of long leave, so sorry for late reply.

I understand what you said, if enable all LPSS devices suspend/resume asynchronously, the system can't resume sometimes on ASUS T100TA(BayTrail-T SoC). But, I have verified that the system can resume normally every time if enable only i2c controller async mode and let other LPSS devices in sync mode on ASUS T100TA. I have no Braswell and cherryTrail platforms, so no verification test on them.

I have submitted a new version of this patch to move the device_enable_async_suspend() call into i2c_dw_proble() function - "[PATCH v2] i2c/designware: enable i2c controller to suspend/resume asynchronously".

Thanks,
Zhonghui

Quoted 2 lines