mmc: enable mmc host device to suspend/resume asynchronously

PATCH In review Fu, Zhonghui 3 participants 6 messages Aug 3, 2015 – Sep 21, 2015 MMC, SD and SDIO

lore

Share & tools

lore

Fu, Zhonghui

author

2015-08-17 03:36 UTC #1 lore raw

Hi,

Any comments are welcome.

Thanks,
Zhonghui

Quoted 22 lines · Zhonghui

Enable mmc host device to suspend/resume asynchronously. This can improve system suspend/resume speed.

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

---
 drivers/mmc/core/host.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

drivers/mmc/core/host.c +1 -1

index 99a9c90..85f2bbb 100644
@@ -577,6 +577,7 @@ struct mmc_host *mmc_alloc_host(int extra, struct device *dev)
 	host->class_dev.parent = dev;
 	host->class_dev.class = &mmc_host_class;
 	device_initialize(&host->class_dev);
+	device_enable_async_suspend(&host->class_dev);
 
 	if (mmc_gpio_alloc(host)) {
 		put_device(&host->class_dev);
-- 1.7.1

Adrian Hunter

2015-08-17 06:55 UTC #2 lore raw

Quoted 4 lines · Zhonghui

Hi,
Any comments are welcome.

Same comments as here:

	http://marc.info/?l=linux-kernel&m=143979428424353&w=2

Quoted 31 lines

Thanks,
Zhonghui
Quoted 22 lines · Zhonghui
Enable mmc host device to suspend/resume asynchronously. This can improve system suspend/resume speed.
Signed-off-by Zhonghui Fu <[email protected]>
---
 drivers/mmc/core/host.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
drivers/mmc/core/host.c +1 -1
index 99a9c90..85f2bbb 100644
@@ -577,6 +577,7 @@ struct mmc_host *mmc_alloc_host(int extra, struct device *dev)
 	host->class_dev.parent = dev;
 	host->class_dev.class = &mmc_host_class;
 	device_initialize(&host->class_dev);
+	device_enable_async_suspend(&host->class_dev);
 
 	if (mmc_gpio_alloc(host)) {
 		put_device(&host->class_dev);
-- 1.7.1

Fu, Zhonghui

author

2015-08-24 15:11 UTC #3 lore raw

Quoted 7 lines · Adrian Hunter

Quoted 3 lines · Zhonghui
Hi,
Any comments are welcome.
Same comments as here:
	http://marc.info/?l=linux-kernel&m=143979428424353&w=2

Now, PM core support asynchronous device suspend/resume mode. If one device has been set to support asynchronous PM mode, it's suspend/resume operation can be performed in a separate kernel thread and take advantage of multicore feature to improve overall system suspend/resume speed. The worst case is that all device suspend/resume threads will be scheduled to the same CPU, it hardly occur.

PM core ensure all the suspend/resume dependency related to one device. Actually, async suspend/resume mode is one feature of PM core, every device subsystem may use it or not use it. Once one device subsystem choose to use this feature, its safety is up to PM core as long as device subsystem has initialized fully this device.

Thanks,
Zhonghui

Quoted 35 lines

Quoted 24 lines
Thanks,
Zhonghui
Quoted 18 lines · Zhonghui
Enable mmc host device to suspend/resume asynchronously. This can improve system suspend/resume speed.
Signed-off-by Zhonghui Fu <[email protected]>
---
 drivers/mmc/core/host.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
drivers/mmc/core/host.c +1 -0
index 99a9c90..85f2bbb 100644
@@ -577,6 +577,7 @@ struct mmc_host *mmc_alloc_host(int extra, struct device *dev)
 	host->class_dev.parent = dev;
 	host->class_dev.class = &mmc_host_class;
 	device_initialize(&host->class_dev);
+	device_enable_async_suspend(&host->class_dev);
 
Quoted 6 lines
drivers/mmc/core/host.c struct mmc_host *mmc_alloc_host(int extra, struct device *dev)
Quoted 4 lines
drivers/mmc/core/host.c +0 -1
 	if (mmc_gpio_alloc(host)) {
 		put_device(&host->class_dev);
-- 1.7.1
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Ulf Hansson

2015-08-27 13:05 UTC #4 lore raw

In reply to Fu, Zhonghui

Quoted 2 lines · Zhonghui

Enable mmc host device to suspend/resume asynchronously. This can improve system suspend/resume speed.

Can or will?

It would be nice to see some statistics of this to justify the change. Can you share that?

Moreover we already have the runtime PM support, which enables the reinitialization sequence of the mmc/sd card to be postponed from the system PM resume path. Instead that's done when the next pm_runtime_get_sync() for the card's device gets called. You my tried that feature by enabling MMC_CAP_RUNTIME_RESUME for the mmc hosts.

Kind regards
Uffe

Quoted 20 lines

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

---
 drivers/mmc/core/host.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

drivers/mmc/core/host.c +1 -1

index 99a9c90..85f2bbb 100644
@@ -577,6 +577,7 @@ struct mmc_host *mmc_alloc_host(int extra, struct device *dev)
        host->class_dev.parent = dev;
        host->class_dev.class = &mmc_host_class;
        device_initialize(&host->class_dev);
+       device_enable_async_suspend(&host->class_dev);
 
        if (mmc_gpio_alloc(host)) {
                put_device(&host->class_dev);
-- 1.7.1

Fu, Zhonghui

author

2015-09-21 06:27 UTC #5 lore raw

Quoted 5 lines · Ulf Hansson

Quoted 2 lines · Zhonghui
Enable mmc host device to suspend/resume asynchronously. This can improve system suspend/resume speed.
Can or will?

Quoted 8 lines

It would be nice to see some statistics of this to justify the change. Can you share that?
Moreover we already have the runtime PM support, which enables the reinitialization sequence of the mmc/sd card to be postponed from the system PM resume path. Instead that's done when the next pm_runtime_get_sync() for the card's device gets called. You my tried that feature by enabling MMC_CAP_RUNTIME_RESUME for the mmc hosts.

This "will" improve system suspend/resume speed. I have resent this patch with updated commit message - "[PATCH v2] mmc: enable mmc host device to suspend/resume asynchronously".

The original suspend time is 1645ms and resume time is 940ms on ASUS T100TA machine. After enabling "wiphy device", "SDIO device", "mmc host" and "sdhci-acpi device" to suspend/resume asynchronously, the suspend time is 1096ms and resume time is 908ms. The test environment is listed as follows:

OS: Ubuntu 14.04
Kernel: mainline v4.1
Machine: ASUS T100TA(Baytrail-T platform)
Tool: analyze_suspend
“analyze_suspend.py –f –m freeze” to suspend system Press power button to resume system

System PM is independent from runtime PM. 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 timing dependency between devices. This patch enables mmc host device to suspend/resume asynchronously. This will take advantage of multicore and improve system suspend/resume speed.

Thanks,
Zhonghui

Quoted 27 lines

Kind regards
Uffe
Quoted 11 lines
Signed-off-by Zhonghui Fu <[email protected]>
---
 drivers/mmc/core/host.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
drivers/mmc/core/host.c +0 -0
index 99a9c90..85f2bbb 100644
drivers/mmc/core/host.c struct mmc_host *mmc_alloc_host(int extra, struct device *dev)
Quoted 8 lines
drivers/mmc/core/host.c +1 -1
@@ -577,6 +577,7 @@ struct mmc_host *mmc_alloc_host(int extra, struct device *dev)
        host->class_dev.parent = dev;
        host->class_dev.class = &mmc_host_class;
        device_initialize(&host->class_dev);
+       device_enable_async_suspend(&host->class_dev);
 
        if (mmc_gpio_alloc(host)) {
                put_device(&host->class_dev);
-- 1.7.1
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/