Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1483904 > unrolled thread

Re: [PATCH v3] mmc: sdhci-msm: Add pm_runtime and system PM support

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH v3] mmc: sdhci-msm: Add pm_runtime and system PM support Pramod Gurav <pramod.gurav@linaro.org> - 2016-09-15 10:00 +0200
    Re: [PATCH v3] mmc: sdhci-msm: Add pm_runtime and system PM support Ulf Hansson <ulf.hansson@linaro.org> - 2016-09-15 12:30 +0200
      Re: [PATCH v3] mmc: sdhci-msm: Add pm_runtime and system PM support Pramod Gurav <pramod.gurav@linaro.org> - 2016-09-15 16:00 +0200
        Re: [PATCH v3] mmc: sdhci-msm: Add pm_runtime and system PM support Ritesh Harjani <riteshh@codeaurora.org> - 2016-09-22 16:40 +0200
          Re: [PATCH v3] mmc: sdhci-msm: Add pm_runtime and system PM support Pramod Gurav <pramod.gurav@linaro.org> - 2016-09-23 08:30 +0200
          Re: [PATCH v3] mmc: sdhci-msm: Add pm_runtime and system PM support Ulf Hansson <ulf.hansson@linaro.org> - 2016-09-23 12:10 +0200

#1483904 — Re: [PATCH v3] mmc: sdhci-msm: Add pm_runtime and system PM support

On 9 September 2016 at 15:48, Georgi Djakov <georgi.djakov@linaro.org> wrote:
> On 09/08/2016 11:02 AM, Adrian Hunter wrote:
>>
>> On 01/09/16 17:23, Pramod Gurav wrote:
>>>
>>> Provides runtime PM callbacks to enable and disable clock resources
>>> when idle. Also support system PM callbacks to be called during system
>>> suspend and resume.
>>>
>>> Signed-off-by: Pramod Gurav <pramod.gurav@linaro.org>
>>
>>
>> Can we get some Tested/Reviewed/Acked-by from people using this driver?
>>
>
> Hi Pramod,
> Thanks for the patch. Unfortunately, my db410c board fails to
> boot when i apply it.
>

Thanks Georgi for testing the patch. Its my wrong I did not update my
kernel and continued fixing comments on old kernel.
After spending some time I came to know that below change is causing the issue:

Author: Dong Aisheng <aisheng.dong@nxp.com>
Date:   Tue Jul 12 15:46:17 2016 +0800

    mmc: sdhci: add standard hw auto retuning support

    If HW supports SDHCI_TUNING_MODE_3 which is auto retuning, we won't
    retune during runtime suspend and resume, instead we use Re-tuning
    Request signaled via SDHCI_INT_RETUNE interrupt to do retuning and
    hw auto retuning during data transfer to guarantee the signal sample
    window correction.

    This can avoid a mass of repeatedly retuning during small file system
    data access and improve the performance.

Specially these lines that was added to suspend path:

+       if (host->tuning_mode != SDHCI_TUNING_MODE_3)
+               mmc_retune_needed(host->mmc);

During sdhci setup in msm driver, the host returns the values to set
sdhci auto tuning as supported.
Hence host->tuning_mode is set to SDHCI_TUNING_MODE_3 during setup.
But some how the auto tuning is not happening.
Just to verify my case, I removed the 'if' part in above code and got
the FS mounted.

Is there anything else needed in msm sdhci driver so that the auto
tuning is taken care of?

> [    1.778433] mmc0: new HS200 MMC card at address 0001
> [    1.783115] mmcblk0: mmc0:0001 DS2008 7.28 GiB
> [    1.783337] mmcblk0boot0: mmc0:0001 DS2008 partition 1 4.00 MiB
> [    1.787025] mmcblk0boot1: mmc0:0001 DS2008 partition 2 4.00 MiB
> [    1.792893] mmcblk0rpmb: mmc0:0001 DS2008 partition 3 4.00 MiB
> [    1.802603]  mmcblk0: p1 p2 p3 p4 p5 p6 p7 p8 p9 p10
> [    2.693631] blk_update_request: I/O error, dev mmcblk0, sector 462880
> [    2.710381] blk_update_request: I/O error, dev mmcblk0, sector 462880
> [    2.710443] Buffer I/O error on dev mmcblk0p10, logical block 0, async
> page read
> [    2.724827] blk_update_request: I/O error, dev mmcblk0, sector 462881
> [    2.724853] Buffer I/O error on dev mmcblk0p10, logical block 1, async
> page read
> ...
>
> More I/O errors are following and it is unable to mount the rootfs from
> the eMMC. When i retried booting, got also the following:
>
> [    2.877149] mmcblk0: error -110 sending status command, retrying
> [    2.879408] mmcblk0: error -110 sending status command, retrying
> [    2.884436] mmcblk0: error -110 sending status command, aborting
> [    2.896826] mmc0: cache flush error -110
>
> BR,
> Georgi

[toc] | [next] | [standalone]

#1483957

On 15 September 2016 at 09:59, Pramod Gurav <pramod.gurav@linaro.org> wrote:
> On 9 September 2016 at 15:48, Georgi Djakov <georgi.djakov@linaro.org> wrote:
>> On 09/08/2016 11:02 AM, Adrian Hunter wrote:
>>>
>>> On 01/09/16 17:23, Pramod Gurav wrote:
>>>>
>>>> Provides runtime PM callbacks to enable and disable clock resources
>>>> when idle. Also support system PM callbacks to be called during system
>>>> suspend and resume.
>>>>
>>>> Signed-off-by: Pramod Gurav <pramod.gurav@linaro.org>
>>>
>>>
>>> Can we get some Tested/Reviewed/Acked-by from people using this driver?
>>>
>>
>> Hi Pramod,
>> Thanks for the patch. Unfortunately, my db410c board fails to
>> boot when i apply it.
>>
>
> Thanks Georgi for testing the patch. Its my wrong I did not update my
> kernel and continued fixing comments on old kernel.
> After spending some time I came to know that below change is causing the issue:
>
> Author: Dong Aisheng <aisheng.dong@nxp.com>
> Date:   Tue Jul 12 15:46:17 2016 +0800
>
>     mmc: sdhci: add standard hw auto retuning support
>
>     If HW supports SDHCI_TUNING_MODE_3 which is auto retuning, we won't
>     retune during runtime suspend and resume, instead we use Re-tuning
>     Request signaled via SDHCI_INT_RETUNE interrupt to do retuning and
>     hw auto retuning during data transfer to guarantee the signal sample
>     window correction.
>
>     This can avoid a mass of repeatedly retuning during small file system
>     data access and improve the performance.
>
> Specially these lines that was added to suspend path:
>
> +       if (host->tuning_mode != SDHCI_TUNING_MODE_3)
> +               mmc_retune_needed(host->mmc);
>
> During sdhci setup in msm driver, the host returns the values to set
> sdhci auto tuning as supported.
> Hence host->tuning_mode is set to SDHCI_TUNING_MODE_3 during setup.
> But some how the auto tuning is not happening.
> Just to verify my case, I removed the 'if' part in above code and got
> the FS mounted.
>
> Is there anything else needed in msm sdhci driver so that the auto
> tuning is taken care of?

I am not familiar with any other than sdhci-esdhc-imx which supports
the SDHCI_TUNING_MODE_3. I may be wrong though.

In the sdhci-esdhc-imx case, enabling of auto tuning seems to be done
in esdhc_post_tuning(), where a vendor specific register
(ESDHC_MIX_CTRL) is being written to. Perhaps something similar in
your case?

[...]

Kind regards
Uffe

[toc] | [prev] | [next] | [standalone]

#1484196

On 15 September 2016 at 15:49, Ulf Hansson <ulf.hansson@linaro.org> wrote:
> On 15 September 2016 at 09:59, Pramod Gurav <pramod.gurav@linaro.org> wrote:
>> On 9 September 2016 at 15:48, Georgi Djakov <georgi.djakov@linaro.org> wrote:
>>> On 09/08/2016 11:02 AM, Adrian Hunter wrote:
>>>>
>>>> On 01/09/16 17:23, Pramod Gurav wrote:
>>>>>
>>>>> Provides runtime PM callbacks to enable and disable clock resources
>>>>> when idle. Also support system PM callbacks to be called during system
>>>>> suspend and resume.
>>>>>
>>>>> Signed-off-by: Pramod Gurav <pramod.gurav@linaro.org>
>>>>
>>>>
>>>> Can we get some Tested/Reviewed/Acked-by from people using this driver?
>>>>
>>>
>>> Hi Pramod,
>>> Thanks for the patch. Unfortunately, my db410c board fails to
>>> boot when i apply it.
>>>
>>
>> Thanks Georgi for testing the patch. Its my wrong I did not update my
>> kernel and continued fixing comments on old kernel.
>> After spending some time I came to know that below change is causing the issue:
>>
>> Author: Dong Aisheng <aisheng.dong@nxp.com>
>> Date:   Tue Jul 12 15:46:17 2016 +0800
>>
>>     mmc: sdhci: add standard hw auto retuning support
>>
>>     If HW supports SDHCI_TUNING_MODE_3 which is auto retuning, we won't
>>     retune during runtime suspend and resume, instead we use Re-tuning
>>     Request signaled via SDHCI_INT_RETUNE interrupt to do retuning and
>>     hw auto retuning during data transfer to guarantee the signal sample
>>     window correction.
>>
>>     This can avoid a mass of repeatedly retuning during small file system
>>     data access and improve the performance.
>>
>> Specially these lines that was added to suspend path:
>>
>> +       if (host->tuning_mode != SDHCI_TUNING_MODE_3)
>> +               mmc_retune_needed(host->mmc);
>>
>> During sdhci setup in msm driver, the host returns the values to set
>> sdhci auto tuning as supported.
>> Hence host->tuning_mode is set to SDHCI_TUNING_MODE_3 during setup.
>> But some how the auto tuning is not happening.
>> Just to verify my case, I removed the 'if' part in above code and got
>> the FS mounted.
>>
>> Is there anything else needed in msm sdhci driver so that the auto
>> tuning is taken care of?
>
> I am not familiar with any other than sdhci-esdhc-imx which supports
> the SDHCI_TUNING_MODE_3. I may be wrong though.
>
> In the sdhci-esdhc-imx case, enabling of auto tuning seems to be done
> in esdhc_post_tuning(), where a vendor specific register
> (ESDHC_MIX_CTRL) is being written to. Perhaps something similar in
> your case?
>
Thanks Ulf for the comments. Will check this and see if there is
something of this sort we have to do to achieve auto tuning.
Adding Ritesh who has been posting some SDHCI MSM patches recently in
case he knows about this.

Regards,
Pramod

[toc] | [prev] | [next] | [standalone]

#1488942

Hi Pramod,

On 9/15/2016 7:28 PM, Pramod Gurav wrote:
> On 15 September 2016 at 15:49, Ulf Hansson <ulf.hansson@linaro.org> wrote:
>> On 15 September 2016 at 09:59, Pramod Gurav <pramod.gurav@linaro.org> wrote:
>>> On 9 September 2016 at 15:48, Georgi Djakov <georgi.djakov@linaro.org> wrote:
>>>> On 09/08/2016 11:02 AM, Adrian Hunter wrote:
>>>>>
>>>>> On 01/09/16 17:23, Pramod Gurav wrote:
>>>>>>
>>>>>> Provides runtime PM callbacks to enable and disable clock resources
>>>>>> when idle. Also support system PM callbacks to be called during system
>>>>>> suspend and resume.
>>>>>>
>>>>>> Signed-off-by: Pramod Gurav <pramod.gurav@linaro.org>
>>>>>
>>>>>
>>>>> Can we get some Tested/Reviewed/Acked-by from people using this driver?
>>>>>
>>>>
>>>> Hi Pramod,
>>>> Thanks for the patch. Unfortunately, my db410c board fails to
>>>> boot when i apply it.
>>>>
>>>
>>> Thanks Georgi for testing the patch. Its my wrong I did not update my
>>> kernel and continued fixing comments on old kernel.
>>> After spending some time I came to know that below change is causing the issue:
>>>
>>> Author: Dong Aisheng <aisheng.dong@nxp.com>
>>> Date:   Tue Jul 12 15:46:17 2016 +0800
>>>
>>>     mmc: sdhci: add standard hw auto retuning support
>>>
>>>     If HW supports SDHCI_TUNING_MODE_3 which is auto retuning, we won't
>>>     retune during runtime suspend and resume, instead we use Re-tuning
>>>     Request signaled via SDHCI_INT_RETUNE interrupt to do retuning and
>>>     hw auto retuning during data transfer to guarantee the signal sample
>>>     window correction.
>>>
>>>     This can avoid a mass of repeatedly retuning during small file system
>>>     data access and improve the performance.
>>>
>>> Specially these lines that was added to suspend path:
>>>
>>> +       if (host->tuning_mode != SDHCI_TUNING_MODE_3)
>>> +               mmc_retune_needed(host->mmc);
>>>
>>> During sdhci setup in msm driver, the host returns the values to set
>>> sdhci auto tuning as supported.
>>> Hence host->tuning_mode is set to SDHCI_TUNING_MODE_3 during setup.
>>> But some how the auto tuning is not happening.
>>> Just to verify my case, I removed the 'if' part in above code and got
>>> the FS mounted.
>>>
>>> Is there anything else needed in msm sdhci driver so that the auto
>>> tuning is taken care of?
>>
>> I am not familiar with any other than sdhci-esdhc-imx which supports
>> the SDHCI_TUNING_MODE_3. I may be wrong though.
>>
>> In the sdhci-esdhc-imx case, enabling of auto tuning seems to be done
>> in esdhc_post_tuning(), where a vendor specific register
>> (ESDHC_MIX_CTRL) is being written to. Perhaps something similar in
>> your case?
>>
> Thanks Ulf for the comments. Will check this and see if there is
> something of this sort we have to do to achieve auto tuning.
> Adding Ritesh who has been posting some SDHCI MSM patches recently in
> case he knows about this.

Internally, we don't use this Auto re-tuning and rely on explicit 
re-tune by host driver.

Question though -
1. why do we need to call sdhci_runtime_resume/suspend from 
sdhci_msm_runtime_suspend/resume?
 From what I see is, sdhci_runtime_susend/resume will do reset and 
re-program of host->pwr and host->clk because of which a retune will be 
required for the next command after runtime resume.

We can *only* disable and enable the clocks in 
sdhci_msm_runtime_suspend/resume?
Thoughts? With this, I suppose you would not see any issue.


Though for this issue, since internally also auto retuning is never 
used, we can have this mode disabled. I can once again check with HW 
team to get more details about this mode for MSM controller.

>
> Regards,
> Pramod
>

[toc] | [prev] | [next] | [standalone]

#1489786

Hi Ritesh,

Thanks for the inputs.

On 22 September 2016 at 20:02, Ritesh Harjani <riteshh@codeaurora.org> wrote:
> Hi Pramod,

<snip>

>> Thanks Ulf for the comments. Will check this and see if there is
>> something of this sort we have to do to achieve auto tuning.
>> Adding Ritesh who has been posting some SDHCI MSM patches recently in
>> case he knows about this.
>
>
> Internally, we don't use this Auto re-tuning and rely on explicit re-tune by
> host driver.
>
> Question though -
> 1. why do we need to call sdhci_runtime_resume/suspend from
> sdhci_msm_runtime_suspend/resume?
> From what I see is, sdhci_runtime_susend/resume will do reset and re-program
> of host->pwr and host->clk because of which a retune will be required for
> the next command after runtime resume.

Honestly I took reference from existing SDHCI HC driver which
implement the runtime PM and each one uses this function.

>
> We can *only* disable and enable the clocks in
> sdhci_msm_runtime_suspend/resume?
> Thoughts? With this, I suppose you would not see any issue.
This should work as I did not have this funtion call in my V1 but
later included when I referred the other sdhci drivers.

>
>
> Though for this issue, since internally also auto retuning is never used, we
> can have this mode disabled. I can once again check with HW team to get more
> details about this mode for MSM controller.
This seems a read only register. And I could not find any other
reference of this mode in any of the docs.
>
>>
>> Regards,
>> Pramod
>>
>

[toc] | [prev] | [next] | [standalone]

#1489918

[...]

>>>> Is there anything else needed in msm sdhci driver so that the auto
>>>> tuning is taken care of?
>>>
>>>
>>> I am not familiar with any other than sdhci-esdhc-imx which supports
>>> the SDHCI_TUNING_MODE_3. I may be wrong though.
>>>
>>> In the sdhci-esdhc-imx case, enabling of auto tuning seems to be done
>>> in esdhc_post_tuning(), where a vendor specific register
>>> (ESDHC_MIX_CTRL) is being written to. Perhaps something similar in
>>> your case?
>>>
>> Thanks Ulf for the comments. Will check this and see if there is
>> something of this sort we have to do to achieve auto tuning.
>> Adding Ritesh who has been posting some SDHCI MSM patches recently in
>> case he knows about this.
>
>
> Internally, we don't use this Auto re-tuning and rely on explicit re-tune by
> host driver.
>
> Question though -
> 1. why do we need to call sdhci_runtime_resume/suspend from
> sdhci_msm_runtime_suspend/resume?
> From what I see is, sdhci_runtime_susend/resume will do reset and re-program
> of host->pwr and host->clk because of which a retune will be required for
> the next command after runtime resume.
>
> We can *only* disable and enable the clocks in
> sdhci_msm_runtime_suspend/resume?
> Thoughts? With this, I suppose you would not see any issue.

I see.

I assumes that means saving/restoring register context will
automatically handled by some other outer logic, when doing clock
gating/ungating?

In other words, if the controller has valid tuning values, those will
be re-used and restored when clock ungating happens?

>
>
> Though for this issue, since internally also auto retuning is never used, we
> can have this mode disabled. I can once again check with HW team to get more
> details about this mode for MSM controller.
>
>>
>> Regards,
>> Pramod
>>
>

Kind regards
Uffe

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web