[PATCH v2] PM: HIBERNATION: add resume_wait param to support MMC-like devices as resume file

[PATCH v2] PM: HIBERNATION: add resume_wait param to support MMC-like devices as resume file

[Barry Song]

From: Barry Song <baohua.song@csr.com>

Some devices like mmc are async detected very slow. For example,
drivers/mmc/host/sdhci.c launchs a 200ms delayed work to detect
mmc partitions then add disk.

we do have wait_for_device_probe and scsi_complete_async_scans
before calling swsusp_check, but it is not enough to wait mmc.

This patch adds resumewait kernel param just like rootwait so
that we have enough time to wait mmc ready. The differene is
here we wait for resume partition but rootwait waits for rootfs
partition.

This patch will make swsusp support many embedded products
without scsi devices but devices like mmc.

This was tested on one ARM platform with Frank Hofmann's patch.

Signed-off-by: Barry Song <Baohua.Song@csr.com>
Reviewed-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
---
 -v2:
 add document for the resumewait kernel parameter

 Documentation/kernel-parameters.txt |    4 ++++
 kernel/power/hibernate.c            |   16 ++++++++++++++++
 2 files changed, 20 insertions(+), 0 deletions(-)

diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
index 614d038..2aed0d2 100644
--- a/Documentation/kernel-parameters.txt
+++ b/Documentation/kernel-parameters.txt
@@ -2237,6 +2237,10 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
 			in <PAGE_SIZE> units (needed only for swap files).
 			See  Documentation/power/swsusp-and-swap-files.txt
 
+	resumewait	[HIBERNATION] Wait (indefinitely) for resume device to show up.
+			Useful for devices that are detected asynchronously
+			(e.g. USB and MMC devices).
+
 	hibernate=	[HIBERNATION]
 		noresume	Don't check if there's a hibernation image
 				present during boot.
diff --git a/kernel/power/hibernate.c b/kernel/power/hibernate.c
index 8f7b1db..9f4b4f3 100644
--- a/kernel/power/hibernate.c
+++ b/kernel/power/hibernate.c
@@ -14,6 +14,7 @@
 #include <linux/reboot.h>
 #include <linux/string.h>
 #include <linux/device.h>
+#include <linux/async.h>
 #include <linux/kmod.h>
 #include <linux/delay.h>
 #include <linux/fs.h>
@@ -31,6 +32,7 @@
 
 static int nocompress = 0;
 static int noresume = 0;
+static int resume_wait = 0;
 static char resume_file[256] = CONFIG_PM_STD_PARTITION;
 dev_t swsusp_resume_device;
 sector_t swsusp_resume_block;
@@ -732,6 +734,13 @@ static int software_resume(void)
 		 * to wait for this to finish.
 		 */
 		wait_for_device_probe();
+
+		if (resume_wait) {
+			while ((swsusp_resume_device = name_to_dev_t(resume_file)) == 0)
+				msleep(10);
+			async_synchronize_full();
+		}
+
 		/*
 		 * We can't depend on SCSI devices being available after loading
 		 * one of their modules until scsi_complete_async_scans() is
@@ -1060,7 +1069,14 @@ static int __init noresume_setup(char *str)
 	return 1;
 }
 
+static int __init resumewait_setup(char *str)
+{
+	resume_wait = 1;
+	return 1;
+}
+
 __setup("noresume", noresume_setup);
 __setup("resume_offset=", resume_offset_setup);
 __setup("resume=", resume_setup);
 __setup("hibernate=", hibernate_setup);
+__setup("resumewait", resumewait_setup);
-- 
1.7.0.4

Re: [PATCH v2] PM: HIBERNATION: add resume_wait param to support MMC-like devices as resume file

[Rafael J. Wysocki]

On Thursday, September 29, 2011, Barry Song wrote:
> From: Barry Song <baohua.song@csr.com>
> 
> Some devices like mmc are async detected very slow. For example,
> drivers/mmc/host/sdhci.c launchs a 200ms delayed work to detect
> mmc partitions then add disk.
> 
> we do have wait_for_device_probe and scsi_complete_async_scans
> before calling swsusp_check, but it is not enough to wait mmc.
> 
> This patch adds resumewait kernel param just like rootwait so
> that we have enough time to wait mmc ready. The differene is
> here we wait for resume partition but rootwait waits for rootfs
> partition.
> 
> This patch will make swsusp support many embedded products
> without scsi devices but devices like mmc.
> 
> This was tested on one ARM platform with Frank Hofmann's patch.
> 
> Signed-off-by: Barry Song <Baohua.Song@csr.com>
> Reviewed-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>

Sorry, I couldn't find the patch previously.

I only have one question.  Namely, are there any systems that would
use resumewait and wouldn't use rootwait?

Rafael


> ---
>  -v2:
>  add document for the resumewait kernel parameter
> 
>  Documentation/kernel-parameters.txt |    4 ++++
>  kernel/power/hibernate.c            |   16 ++++++++++++++++
>  2 files changed, 20 insertions(+), 0 deletions(-)
> 
> diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
> index 614d038..2aed0d2 100644
> --- a/Documentation/kernel-parameters.txt
> +++ b/Documentation/kernel-parameters.txt
> @@ -2237,6 +2237,10 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
>  			in <PAGE_SIZE> units (needed only for swap files).
>  			See  Documentation/power/swsusp-and-swap-files.txt
>  
> +	resumewait	[HIBERNATION] Wait (indefinitely) for resume device to show up.
> +			Useful for devices that are detected asynchronously
> +			(e.g. USB and MMC devices).
> +
>  	hibernate=	[HIBERNATION]
>  		noresume	Don't check if there's a hibernation image
>  				present during boot.
> diff --git a/kernel/power/hibernate.c b/kernel/power/hibernate.c
> index 8f7b1db..9f4b4f3 100644
> --- a/kernel/power/hibernate.c
> +++ b/kernel/power/hibernate.c
> @@ -14,6 +14,7 @@
>  #include <linux/reboot.h>
>  #include <linux/string.h>
>  #include <linux/device.h>
> +#include <linux/async.h>
>  #include <linux/kmod.h>
>  #include <linux/delay.h>
>  #include <linux/fs.h>
> @@ -31,6 +32,7 @@
>  
>  static int nocompress = 0;
>  static int noresume = 0;
> +static int resume_wait = 0;
>  static char resume_file[256] = CONFIG_PM_STD_PARTITION;
>  dev_t swsusp_resume_device;
>  sector_t swsusp_resume_block;
> @@ -732,6 +734,13 @@ static int software_resume(void)
>  		 * to wait for this to finish.
>  		 */
>  		wait_for_device_probe();
> +
> +		if (resume_wait) {
> +			while ((swsusp_resume_device = name_to_dev_t(resume_file)) == 0)
> +				msleep(10);
> +			async_synchronize_full();
> +		}
> +
>  		/*
>  		 * We can't depend on SCSI devices being available after loading
>  		 * one of their modules until scsi_complete_async_scans() is
> @@ -1060,7 +1069,14 @@ static int __init noresume_setup(char *str)
>  	return 1;
>  }
>  
> +static int __init resumewait_setup(char *str)
> +{
> +	resume_wait = 1;
> +	return 1;
> +}
> +
>  __setup("noresume", noresume_setup);
>  __setup("resume_offset=", resume_offset_setup);
>  __setup("resume=", resume_setup);
>  __setup("hibernate=", hibernate_setup);
> +__setup("resumewait", resumewait_setup);
> 

Re: [PATCH v2] PM: HIBERNATION: add resume_wait param to support MMC-like devices as resume file

[Rafael J. Wysocki]

On Wednesday, October 05, 2011, Barry Song wrote:
> 在 2011-10-5 上午6:30，"Rafael J. Wysocki" <rjw@sisk.pl>写道：
> >
> > On Thursday, September 29, 2011, Barry Song wrote:
> > > From: Barry Song <baohua.song@csr.com>
> > >
> > > Some devices like mmc are async detected very slow. For example,
> > > drivers/mmc/host/sdhci.c launchs a 200ms delayed work to detect
> > > mmc partitions then add disk.
> > >
> > > we do have wait_for_device_probe and scsi_complete_async_scans
> > > before calling swsusp_check, but it is not enough to wait mmc.
> > >
> > > This patch adds resumewait kernel param just like rootwait so
> > > that we have enough time to wait mmc ready. The differene is
> > > here we wait for resume partition but rootwait waits for rootfs
> > > partition.
> > >
> > > This patch will make swsusp support many embedded products
> > > without scsi devices but devices like mmc.
> > >
> > > This was tested on one ARM platform with Frank Hofmann's patch.
> > >
> > > Signed-off-by: Barry Song <Baohua.Song@csr.com>
> > > Reviewed-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
> >
> > Sorry, I couldn't find the patch previously.
> >
> > I only have one question.  Namely, are there any systems that would
> > use resumewait and wouldn't use rootwait?
> 
> I think yes. Some devices use nand
> as root and SD as resume swap.
> Still some devices use both SD as root and swap. But rootfs is mounted by
> ramfs without rootwait.

OK, fair enough.  I applied the patch to linux-pm/linux-next and will push
it for 3.2.

Thanks,
Rafael


> > > ---
> > >  -v2:
> > >  add document for the resumewait kernel parameter
> > >
> > >  Documentation/kernel-parameters.txt |    4 ++++
> > >  kernel/power/hibernate.c            |   16 ++++++++++++++++
> > >  2 files changed, 20 insertions(+), 0 deletions(-)
> > >
> > > diff --git a/Documentation/kernel-parameters.txt
> b/Documentation/kernel-parameters.txt
> > > index 614d038..2aed0d2 100644
> > > --- a/Documentation/kernel-parameters.txt
> > > +++ b/Documentation/kernel-parameters.txt
> > > @@ -2237,6 +2237,10 @@ bytes respectively. Such letter suffixes can also
> be entirely omitted.
> > >                       in <PAGE_SIZE> units (needed only for swap files).
> > >                       See  Documentation/power/swsusp-and-swap-files.txt
> > >
> > > +     resumewait      [HIBERNATION] Wait (indefinitely) for resume
> device to show up.
> > > +                     Useful for devices that are detected
> asynchronously
> > > +                     (e.g. USB and MMC devices).
> > > +
> > >       hibernate=      [HIBERNATION]
> > >               noresume        Don't check if there's a hibernation image
> > >                               present during boot.
> > > diff --git a/kernel/power/hibernate.c b/kernel/power/hibernate.c
> > > index 8f7b1db..9f4b4f3 100644
> > > --- a/kernel/power/hibernate.c
> > > +++ b/kernel/power/hibernate.c
> > > @@ -14,6 +14,7 @@
> > >  #include <linux/reboot.h>
> > >  #include <linux/string.h>
> > >  #include <linux/device.h>
> > > +#include <linux/async.h>
> > >  #include <linux/kmod.h>
> > >  #include <linux/delay.h>
> > >  #include <linux/fs.h>
> > > @@ -31,6 +32,7 @@
> > >
> > >  static int nocompress = 0;
> > >  static int noresume = 0;
> > > +static int resume_wait = 0;
> > >  static char resume_file[256] = CONFIG_PM_STD_PARTITION;
> > >  dev_t swsusp_resume_device;
> > >  sector_t swsusp_resume_block;
> > > @@ -732,6 +734,13 @@ static int software_resume(void)
> > >                * to wait for this to finish.
> > >                */
> > >               wait_for_device_probe();
> > > +
> > > +             if (resume_wait) {
> > > +                     while ((swsusp_resume_device =
> name_to_dev_t(resume_file)) == 0)
> > > +                             msleep(10);
> > > +                     async_synchronize_full();
> > > +             }
> > > +
> > >               /*
> > >                * We can't depend on SCSI devices being available after
> loading
> > >                * one of their modules until scsi_complete_async_scans()
> is
> > > @@ -1060,7 +1069,14 @@ static int __init noresume_setup(char *str)
> > >       return 1;
> > >  }
> > >
> > > +static int __init resumewait_setup(char *str)
> > > +{
> > > +     resume_wait = 1;
> > > +     return 1;
> > > +}
> > > +
> > >  __setup("noresume", noresume_setup);
> > >  __setup("resume_offset=", resume_offset_setup);
> > >  __setup("resume=", resume_setup);
> > >  __setup("hibernate=", hibernate_setup);
> > > +__setup("resumewait", resumewait_setup);
> > >
> >
>