public inbox for linux-mmc@vger.kernel.org
 help / color / mirror / Atom feed
From: Adrian Hunter <adrian.hunter@intel.com>
To: Ulf Hansson <ulf.hansson@linaro.org>
Cc: Ulf Hansson <ulf.hansson@stericsson.com>,
	linux-mmc@vger.kernel.org, Chris Ball <cjb@laptop.org>,
	Maya Erez <merez@codeaurora.org>,
	Subhash Jadavani <subhashj@codeaurora.org>,
	Arnd Bergmann <arnd@arndb.de>, Kevin Liu <kliu5@marvell.com>,
	Daniel Drake <dsd@laptop.org>, Ohad Ben-Cohen <ohad@wizery.com>
Subject: Re: [PATCH V3 4/4] mmc: core: Support aggressive power management for (e)MMC/SD
Date: Thu, 02 May 2013 15:24:38 +0300	[thread overview]
Message-ID: <51825B06.7040709@intel.com> (raw)
In-Reply-To: <CAPDyKFp3NGC13qOT=y=MxiANO9ecU3y3vMFTYHfiM-56AT_6-g@mail.gmail.com>

On 02/05/13 14:35, Ulf Hansson wrote:
> On 2 May 2013 12:38, Adrian Hunter <adrian.hunter@intel.com> wrote:
>> On 16/04/13 13:00, Ulf Hansson wrote:
>>> Aggressive power management is suitable when saving power is
>>> essential. At request inactivity timeout, aka pm runtime
>>> autosuspend timeout, the card will be suspended.
>>>
>>> Once a new request arrives, the card will be re-initalized and
>>> thus the first request will suffer from a latency. This latency
>>> is card-specific, experiments has shown in general that SD-cards
>>> has quite poor initialization time, around 300ms-1100ms. eMMC is
>>> not surprisingly far better but still a couple of hundreds of ms
>>> has been observed.
>>>
>>> Except for the request latency, it is important to know that
>>> suspending the card will also prevent the card from executing
>>> internal house-keeping operations in idle mode. This could mean
>>> degradation in performance.
>>>
>>> To use this feature make sure the request inactivity timeout is
>>> chosen carefully. This has not been done as a part of this patch.
>>>
>>> Enable this feature by using host cap MMC_CAP_AGGRESSIVE_PM and
>>> by setting CONFIG_MMC_UNSAFE_RESUME.
>>>
>>> Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
>>> Cc: Maya Erez <merez@codeaurora.org>
>>> Cc: Subhash Jadavani <subhashj@codeaurora.org>
>>> Cc: Arnd Bergmann <arnd@arndb.de>
>>> Cc: Kevin Liu <kliu5@marvell.com>
>>> Cc: Adrian Hunter <adrian.hunter@intel.com>
>>> Cc: Daniel Drake <dsd@laptop.org>
>>> Cc: Ohad Ben-Cohen <ohad@wizery.com>
>>> ---
>>>  drivers/mmc/core/mmc.c   |   43 +++++++++++++++++++++++++++++++++++++++++++
>>>  drivers/mmc/core/sd.c    |   42 ++++++++++++++++++++++++++++++++++++++++++
>>>  include/linux/mmc/host.h |    2 +-
>>>  3 files changed, 86 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/mmc/core/mmc.c b/drivers/mmc/core/mmc.c
>>> index bf19058..8dfbc84 100644
>>> --- a/drivers/mmc/core/mmc.c
>>> +++ b/drivers/mmc/core/mmc.c
>>> @@ -1454,6 +1454,47 @@ static int mmc_resume(struct mmc_host *host)
>>>       return err;
>>>  }
>>>
>>> +
>>> +/*
>>> + * Callback for runtime_suspend.
>>> + */
>>> +static int mmc_runtime_suspend(struct mmc_host *host)
>>> +{
>>> +     int err;
>>> +
>>> +     if (!(host->caps & MMC_CAP_AGGRESSIVE_PM))
>>> +             return 0;
>>> +
>>
>> mmc_power_off() needs to be within mmc_claim_host() / mmc_release_host().
>>
>> Claiming is nested, so you can out put mmc_claim_host() here:
>>
>>         mmc_claim_host(host);
> 
> OK
> 
>>
>>
>>> +     err = mmc_suspend(host);
>>> +     if (err) {
>>> +             pr_err("%s: error %d doing aggessive suspend\n",
>>> +                     mmc_hostname(host), err);
>>> +             return err;
>>
>>                 goto out;
>>
>>> +     }
>>> +
>>> +     mmc_power_off(host);
>>
>> out:
>>         mmc_release_host(host);
>>
> 
> OK
> 
>>> +     return err;
>>> +}
>>> +
>>> +/*
>>> + * Callback for runtime_resume.
>>> + */
>>> +static int mmc_runtime_resume(struct mmc_host *host)
>>> +{
>>> +     int err;
>>> +
>>> +     if (!(host->caps & MMC_CAP_AGGRESSIVE_PM))
>>> +             return 0;
>>> +
>>> +     mmc_power_up(host);
>>
>> As above
>>
> 
> OK
> 
>>> +     err = mmc_resume(host);
>>> +     if (err)
>>> +             pr_err("%s: error %d doing aggessive resume\n",
>>> +                     mmc_hostname(host), err);
>>> +
>>> +     return err;
>>
>> The power is on - leaving the device in a RPM_SUSPENDED state does not seem
>> useful so better to return zero here.
> 
> OK
> 
>>
>>> +}
>>> +
>>>  static int mmc_power_restore(struct mmc_host *host)
>>>  {
>>>       int ret;
>>> @@ -1514,6 +1555,8 @@ static const struct mmc_bus_ops mmc_ops_unsafe = {
>>>       .detect = mmc_detect,
>>>       .suspend = mmc_suspend,
>>>       .resume = mmc_resume,
>>> +     .runtime_suspend = mmc_runtime_suspend,
>>> +     .runtime_resume = mmc_runtime_resume,
>>>       .power_restore = mmc_power_restore,
>>>       .alive = mmc_alive,
>>>  };
>>> diff --git a/drivers/mmc/core/sd.c b/drivers/mmc/core/sd.c
>>> index 30387d6..e0458f9 100644
>>> --- a/drivers/mmc/core/sd.c
>>> +++ b/drivers/mmc/core/sd.c
>>> @@ -1095,6 +1095,46 @@ static int mmc_sd_resume(struct mmc_host *host)
>>>       return err;
>>>  }
>>>
>>> +/*
>>> + * Callback for runtime_suspend.
>>> + */
>>> +static int mmc_sd_runtime_suspend(struct mmc_host *host)
>>> +{
>>> +     int err;
>>> +
>>> +     if (!(host->caps & MMC_CAP_AGGRESSIVE_PM))
>>> +             return 0;
>>> +
>>> +     err = mmc_sd_suspend(host);
>>> +     if (err) {
>>> +             pr_err("%s: error %d doing aggessive suspend\n",
>>> +                     mmc_hostname(host), err);
>>> +             return err;
>>> +     }
>>> +
>>> +     mmc_power_off(host);
>>
>> As above
> 
> OK
> 
>>
>>> +     return err;
>>> +}
>>> +
>>> +/*
>>> + * Callback for runtime_resume.
>>> + */
>>> +static int mmc_sd_runtime_resume(struct mmc_host *host)
>>> +{
>>> +     int err;
>>> +
>>> +     if (!(host->caps & MMC_CAP_AGGRESSIVE_PM))
>>> +             return 0;
>>> +
>>> +     mmc_power_up(host);
>>
>> As above
> 
> OK
> 
>>
>>> +     err = mmc_sd_resume(host);
>>> +     if (err)
>>> +             pr_err("%s: error %d doing aggessive resume\n",
>>> +                     mmc_hostname(host), err);
>>> +
>>> +     return err;
>>
>> As above
> 
> OK
> 
>>
>>         return 0
>>
>>> +}
>>> +
>>>  static int mmc_sd_power_restore(struct mmc_host *host)
>>>  {
>>>       int ret;
>>> @@ -1119,6 +1159,8 @@ static const struct mmc_bus_ops mmc_sd_ops = {
>>>  static const struct mmc_bus_ops mmc_sd_ops_unsafe = {
>>>       .remove = mmc_sd_remove,
>>>       .detect = mmc_sd_detect,
>>> +     .runtime_suspend = mmc_sd_runtime_suspend,
>>> +     .runtime_resume = mmc_sd_runtime_resume,
>>>       .suspend = mmc_sd_suspend,
>>>       .resume = mmc_sd_resume,
>>>       .power_restore = mmc_sd_power_restore,
>>> diff --git a/include/linux/mmc/host.h b/include/linux/mmc/host.h
>>> index 17d7148..cec6684 100644
>>> --- a/include/linux/mmc/host.h
>>> +++ b/include/linux/mmc/host.h
>>> @@ -239,7 +239,7 @@ struct mmc_host {
>>>  #define MMC_CAP_SPI          (1 << 4)        /* Talks only SPI protocols */
>>>  #define MMC_CAP_NEEDS_POLL   (1 << 5)        /* Needs polling for card-detection */
>>>  #define MMC_CAP_8_BIT_DATA   (1 << 6)        /* Can the host do 8 bit transfers */
>>> -
>>> +#define MMC_CAP_AGGRESSIVE_PM        (1 << 7)        /* Suspend (e)MMC/SD at idle  */
>>
>> Using a "cap" is not ideal here - it should really be under the control of
>> user space.
> 
> Several other caps could be debated whether these should exist here as
> well, but let's leave that to a separate discussion.
> Until there are a solution for how to add new mmc features per host,
> which should be runtime configurable, I believe this is the only way
> we can do it.

Yes, I guess it can be improved later.

> 
> Do note, that runtime pm can be disabled from user space, which
> indirectly will prevent this feature from being used.
> 
>>
>>>  #define MMC_CAP_NONREMOVABLE (1 << 8)        /* Nonremovable e.g. eMMC */
>>>  #define MMC_CAP_WAIT_WHILE_BUSY      (1 << 9)        /* Waits while card is busy */
>>>  #define MMC_CAP_ERASE                (1 << 10)       /* Allow erase/trim commands */
>>>
>>
> 
> 


      reply	other threads:[~2013-05-02 12:19 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-04-16 10:00 [PATCH V3 0/4] mmc: Use runtime pm for blkdevice Ulf Hansson
2013-04-16 10:00 ` [PATCH V3 1/4] mmc: core: Stop bkops for eMMC only from mmc suspend Ulf Hansson
2013-04-18  7:17   ` Jaehoon Chung
2013-04-16 10:00 ` [PATCH V3 2/4] mmc: core: Add bus_ops for runtime pm callbacks Ulf Hansson
2013-04-26 13:11   ` Adrian Hunter
2013-04-29  7:54     ` Adrian Hunter
2013-04-29 13:42       ` Ulf Hansson
2013-04-16 10:00 ` [PATCH V3 3/4] mmc: block: Enable runtime pm for mmc blkdevice Ulf Hansson
2013-05-02  8:58   ` Adrian Hunter
2013-05-02  9:52     ` Ulf Hansson
2013-05-02  9:57       ` Asutosh Das
2013-05-02 11:09         ` Ulf Hansson
2013-05-02 12:22           ` Adrian Hunter
     [not found]   ` <CAMj5BkiOmh8sz-=b0z1VF9owGPX0KpbZeNfPzETemCb=C2odGQ@mail.gmail.com>
2013-05-24  8:27     ` Ulf Hansson
     [not found]       ` <CAMj5Bki+1=DSzQWYyEC1L=Pa6LpSFQKF3YvoUkkuq62wHuMWow@mail.gmail.com>
2013-05-27  7:51         ` Ulf Hansson
2013-05-27  7:52           ` Ulf Hansson
2013-05-28  6:49             ` zhangfei gao
2013-04-16 10:00 ` [PATCH V3 4/4] mmc: core: Support aggressive power management for (e)MMC/SD Ulf Hansson
2013-05-02 10:38   ` Adrian Hunter
2013-05-02 11:35     ` Ulf Hansson
2013-05-02 12:24       ` Adrian Hunter [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=51825B06.7040709@intel.com \
    --to=adrian.hunter@intel.com \
    --cc=arnd@arndb.de \
    --cc=cjb@laptop.org \
    --cc=dsd@laptop.org \
    --cc=kliu5@marvell.com \
    --cc=linux-mmc@vger.kernel.org \
    --cc=merez@codeaurora.org \
    --cc=ohad@wizery.com \
    --cc=subhashj@codeaurora.org \
    --cc=ulf.hansson@linaro.org \
    --cc=ulf.hansson@stericsson.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox