mmc: use usleep_range() in mmc_delay()

Message ID 4EF18310.6080801@linaro.org
State New
Headers show

Commit Message

Dmitry Antipov Dec. 21, 2011, 6:56 a.m.
From f447d78db65c6675e69466e8ed08364ff065ac08 Mon Sep 17 00:00:00 2001
From: Dmitry Antipov <dmitry.antipov@linaro.org>
Date: Wed, 21 Dec 2011 10:51:03 +0400
Subject: [PATCH] mmc: use usleep_range() in mmc_delay()

---
  drivers/mmc/core/core.h |    8 ++------
  1 files changed, 2 insertions(+), 6 deletions(-)

Comments

Sujit Reddy Thumma Dec. 21, 2011, 11:25 a.m. | #1
On 12/21/2011 12:26 PM, Dmitry Antipov wrote:
>  From f447d78db65c6675e69466e8ed08364ff065ac08 Mon Sep 17 00:00:00 2001
> From: Dmitry Antipov <dmitry.antipov@linaro.org>
> Date: Wed, 21 Dec 2011 10:51:03 +0400
> Subject: [PATCH] mmc: use usleep_range() in mmc_delay()
>
> ---
> drivers/mmc/core/core.h | 8 ++------
> 1 files changed, 2 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/mmc/core/core.h b/drivers/mmc/core/core.h
> index 14664f1..a77851e 100644
> --- a/drivers/mmc/core/core.h
> +++ b/drivers/mmc/core/core.h
> @@ -47,12 +47,8 @@ void mmc_power_off(struct mmc_host *host);
>
> static inline void mmc_delay(unsigned int ms)
> {
> - if (ms < 1000 / HZ) {
> - cond_resched();
> - mdelay(ms);
> - } else {
> - msleep(ms);
> - }
> + unsigned long us = ms * USEC_PER_MSEC;
> + usleep_range(us, us + 1000);

I have posted similar patch some time back. 
http://comments.gmane.org/gmane.linux.ports.arm.msm/2119.

Would you like to comment on that?

Thanks,
Sujit

> }
>
> void mmc_rescan(struct work_struct *work);
Dmitry Antipov Dec. 21, 2011, 1:05 p.m. | #2
On 12/21/2011 03:25 PM, Sujit Reddy Thumma wrote:

> I have posted similar patch some time back.
> http://comments.gmane.org/gmane.linux.ports.arm.msm/2119.
>
> Would you like to comment on that?

- I believe we should forget about jiffies, HZ and other similar obsolete
   timekeeping stuff;

- I have no ideas where did you get 'most typical' 20 ms. MMC subsystem uses
   mmc_delay() with two compile-time fixed values 1 and 10 ms, with the only
   exception of card-dependent sleep/awake timeout. I was unable to find a table
   with typical values, but it's rounded up to >= 1 ms anyway.

Dmitry
Sujit Reddy Thumma Dec. 27, 2011, 4:40 a.m. | #3
Hi Antipov,

Sorry for the delayed response. Please find some comments below:

On 12/21/2011 6:35 PM, Dmitry Antipov wrote:
> On 12/21/2011 03:25 PM, Sujit Reddy Thumma wrote:
>
>> I have posted similar patch some time back.
>> http://comments.gmane.org/gmane.linux.ports.arm.msm/2119.
>>
>> Would you like to comment on that?
>
> - I believe we should forget about jiffies, HZ and other similar obsolete
> timekeeping stuff;
>
> - I have no ideas where did you get 'most typical' 20 ms. MMC subsystem
> uses
> mmc_delay() with two compile-time fixed values 1 and 10 ms, with the only
> exception of card-dependent sleep/awake timeout. I was unable to find a
> table
> with typical values, but it's rounded up to >= 1 ms anyway.

The main aim of my patch was to fix mmc_delay() to give accurate delay.
You might want to refer to Documentation/timers/timers-howto.txt 
(Section: SLEEPING FOR ~USECS OR SMALL MSECS) to know why usleep_range() 
must be used instead of msleep for delays less than 20ms (or more 
accurately two jiffies, since in HZ=100 systems this comes to 20ms).

Also, in the documentation it is suggested that for delays greater than 
10ms+ use msleep(). Although MMC subsystem doesn't use mmc_delay() for 
greater than 10ms today but I guess we should keep it for future purpose 
and just not have only usleep_range() as your patch did.

>
> Dmitry
Koskinen, Aaro (Nokia - FI/Espoo) Jan. 13, 2012, 1:22 p.m. | #4
Hi,

On Wed, 21 Dec 2011, Dmitry Antipov wrote:
> From f447d78db65c6675e69466e8ed08364ff065ac08 Mon Sep 17 00:00:00 2001
> From: Dmitry Antipov <dmitry.antipov@linaro.org>
> Date: Wed, 21 Dec 2011 10:51:03 +0400
> Subject: [PATCH] mmc: use usleep_range() in mmc_delay()
>
> ---

Shouldn't you add a proper patch description and a signed-off-by line?

> drivers/mmc/core/core.h |    8 ++------
> 1 files changed, 2 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/mmc/core/core.h b/drivers/mmc/core/core.h
> index 14664f1..a77851e 100644
> --- a/drivers/mmc/core/core.h
> +++ b/drivers/mmc/core/core.h
> @@ -47,12 +47,8 @@ void mmc_power_off(struct mmc_host *host);
>
> static inline void mmc_delay(unsigned int ms)
> {
> -	if (ms < 1000 / HZ) {
> -		cond_resched();
> -		mdelay(ms);
> -	} else {
> -		msleep(ms);
> -	}
> +	unsigned long us = ms * USEC_PER_MSEC;
> +	usleep_range(us, us + 1000);
> }

Anyway, I think the change is good. On systems with multiple MMC devices
the boot/probe can spend 100-200 ms alone just doing busylooping delays. I
think e.g. in mmc_rescan() the code uses frequently mmc_delay(10).

> void mmc_rescan(struct work_struct *work);
> -- 
> 1.7.7.4

A.

Patch

diff --git a/drivers/mmc/core/core.h b/drivers/mmc/core/core.h
index 14664f1..a77851e 100644
--- a/drivers/mmc/core/core.h
+++ b/drivers/mmc/core/core.h
@@ -47,12 +47,8 @@  void mmc_power_off(struct mmc_host *host);

  static inline void mmc_delay(unsigned int ms)
  {
-	if (ms < 1000 / HZ) {
-		cond_resched();
-		mdelay(ms);
-	} else {
-		msleep(ms);
-	}
+	unsigned long us = ms * USEC_PER_MSEC;
+	usleep_range(us, us + 1000);
  }

  void mmc_rescan(struct work_struct *work);