mbox series

[V2,0/3] Internal voltage control for qcom SDHC

Message ID 1590074615-10787-1-git-send-email-vbadigan@codeaurora.org
Headers show
Series Internal voltage control for qcom SDHC | expand

Message

Veerabhadrarao Badiganti May 21, 2020, 3:23 p.m. UTC
On qcom SD host controllers voltage switching be done after the HW
is ready for it. The HW informs its readiness through power irq.
The voltage switching should happen only then.

So added support to register voltage regulators from the msm driver
and use them.

This patchset was posted long back but not actively pursued
https://lore.kernel.org/linux-arm-msm/1539004739-32060-1-git-send-email-vbadigan@codeaurora.org/
So posting it as fresh patchset.  

Changes since V1:
	- Removed setting load for Vmmc regulator while turning it on/off.
	  Instead setting the active load once during probe.
	- Simplified handlng of supplies for BUS_ON/OFF cases in shci_msm_handle_pwr_irq().
	- Moved common code out of switch case in sdhci_msm_start_signal_voltage_switch().
	- Updated variable name to sdhci_core_to_disable_vqmmc.
	- Updated pr_err logs to dev_err logs.

Veerabhadrarao Badiganti (1):
  dt-bindings: mmc: Supply max load for mmc supplies
  mmc: sdhci-msm: Use internal voltage control

Vijay Viswanath (1):
  mmc: sdhci: Allow platform controlled voltage switching

 .../devicetree/bindings/mmc/mmc-controller.yaml    |  16 ++
 drivers/mmc/host/sdhci-msm.c                       | 207 ++++++++++++++++++++-
 drivers/mmc/host/sdhci.c                           |  32 ++--
 drivers/mmc/host/sdhci.h                           |   1 +
 4 files changed, 234 insertions(+), 22 deletions(-)

Comments

Veerabhadrarao Badiganti May 22, 2020, 1:27 p.m. UTC | #1
Hi Bjorn,

On 5/22/2020 12:37 AM, Bjorn Andersson wrote:
> On Thu 21 May 08:23 PDT 2020, Veerabhadrarao Badiganti wrote:
>
>> On qcom SD host controllers voltage switching be done after the HW
>> is ready for it. The HW informs its readiness through power irq.
>> The voltage switching should happen only then.
>>
>> Use the internal voltage switching and then control the voltage
>> switching using power irq.
>>
>> Set the regulator load as well so that regulator can be configured
>> in LPM mode when in is not being used.
>>
>> Co-developed-by: Asutosh Das <asutoshd@codeaurora.org>
>> Signed-off-by: Asutosh Das <asutoshd@codeaurora.org>
>> Co-developed-by: Vijay Viswanath <vviswana@codeaurora.org>
>> Signed-off-by: Vijay Viswanath <vviswana@codeaurora.org>
>> Co-developed-by: Veerabhadrarao Badiganti <vbadigan@codeaurora.org>
>> Signed-off-by: Veerabhadrarao Badiganti <vbadigan@codeaurora.org>
> Looks better, thanks.
>
>> ---
>>   drivers/mmc/host/sdhci-msm.c | 207 +++++++++++++++++++++++++++++++++++++++++--
>>   1 file changed, 198 insertions(+), 9 deletions(-)
>>
>> diff --git a/drivers/mmc/host/sdhci-msm.c b/drivers/mmc/host/sdhci-msm.c
> [..]
>>   static const struct sdhci_msm_offset *sdhci_priv_msm_offset(struct sdhci_host *host)
>> @@ -1298,6 +1302,71 @@ static void sdhci_msm_set_uhs_signaling(struct sdhci_host *host,
>>   		sdhci_msm_hs400(host, &mmc->ios);
>>   }
>>   
>> +static int sdhci_msm_set_vmmc(struct mmc_host *mmc)
>> +{
>> +	int ret;
>> +
>> +	if (IS_ERR(mmc->supply.vmmc))
>> +		return 0;
>> +
>> +	ret = mmc_regulator_set_ocr(mmc, mmc->supply.vmmc, mmc->ios.vdd);
>> +	if (ret)
>> +		dev_err(mmc_dev(mmc), "%s: vmmc set ocr with vdd=%d failed: %d\n",
>> +			mmc_hostname(mmc), mmc->ios.vdd, ret);
> Missed this one on v1, in the event that mmc_regulator_set_ocr() return
> a non-zero value it has already printed an error message. So please
> replace the tail with just:
>
> 	return mmc_regulator_set_ocr(...);
>
>> +
>> +	return ret;
>> +}
>> +
>> +static int sdhci_msm_set_vqmmc(struct sdhci_msm_host *msm_host,
>> +			      struct mmc_host *mmc, bool level)
>> +{
>> +	int load, ret;
>> +	struct mmc_ios ios;
>> +
>> +	if (IS_ERR(mmc->supply.vqmmc)			 ||
>> +	    (mmc->ios.power_mode == MMC_POWER_UNDEFINED) ||
>> +	    (msm_host->vqmmc_enabled == level))
>> +		return 0;
>> +
>> +	if (msm_host->vqmmc_load) {
>> +		load = level ? msm_host->vqmmc_load : 0;
>> +		ret = regulator_set_load(mmc->supply.vqmmc, load);
> Sorry for the late reply on v1, but please see my explanation regarding
> load and always-on regulators there.

<Merging your comment from V1 here>

 >> You should still call regulator_enable()/regulator_disable() on your
 >> consumer regulator in this driver. When you do this the regulator core
 >> will conclude that the regulator_dev (i.e. the part that represents the
 >> hardware) is marked always_on and will not enable/disable the regulator.

 >> But it will still invoke _regulator_handle_consumer_enable() and
 >> _regulator_handle_consumer_disable(), which will aggregate the "load" of
 >> all client regulators and update the regulator's load.

 >> So this will apply the load as you expect regardless of it being
 >> supplied by a regulator marked as always_on.

Since I'm not turning off this regulator for eMMC, I wanted to keep it 
in LPM mode
to save some power.
When the regulator configured in auto mode (RPMH_REGULATOR_MODE_AUTO) it
switches to LPM/HPM mode based on the active load.
So i have to minimize my driver load requirement so that I can let this 
regulator
in LPM mode.
So i need to set load every-time I disable/enable the regulator.

>> +		if (ret) {
>> +			dev_err(mmc_dev(mmc), "%s: vqmmc set load failed: %d\n",
>> +				mmc_hostname(mmc), ret);
>> +			goto out;
>> +		}
>> +	}
>> +
>> +	if (level) {
>> +		/* Set the IO voltage regulator to default voltage level */
>> +		if (msm_host->caps_0 & CORE_3_0V_SUPPORT)
>> +			ios.signal_voltage = MMC_SIGNAL_VOLTAGE_330;
>> +		else if (msm_host->caps_0 & CORE_1_8V_SUPPORT)
>> +			ios.signal_voltage = MMC_SIGNAL_VOLTAGE_180;
>> +
>> +		if (msm_host->caps_0 & CORE_VOLT_SUPPORT) {
>> +			ret = mmc_regulator_set_vqmmc(mmc, &ios);
>> +			if (ret < 0) {
>> +				dev_err(mmc_dev(mmc), "%s: vqmmc set volgate failed: %d\n",
>> +					mmc_hostname(mmc), ret);
>> +				goto out;
>> +			}
>> +		}
>> +		ret = regulator_enable(mmc->supply.vqmmc);
>> +	} else {
>> +		ret = regulator_disable(mmc->supply.vqmmc);
>> +	}
>> +
>> +	if (ret)
>> +		dev_err(mmc_dev(mmc), "%s: vqmm %sable failed: %d\n",
>> +			mmc_hostname(mmc), level ? "en":"dis", ret);
>> +	else
>> +		msm_host->vqmmc_enabled = level;
>> +out:
>> +	return ret;
>> +}
> [..]
>> +static int sdhci_msm_start_signal_voltage_switch(struct mmc_host *mmc,
>> +				      struct mmc_ios *ios)
>> +{
>> +	struct sdhci_host *host = mmc_priv(mmc);
>> +	u16 ctrl, status;
>> +
>> +	/*
>> +	 * Signal Voltage Switching is only applicable for Host Controllers
>> +	 * v3.00 and above.
>> +	 */
>> +	if (host->version < SDHCI_SPEC_300)
>> +		return 0;
>> +
>> +	ctrl = sdhci_readw(host, SDHCI_HOST_CONTROL2);
>> +
>> +	switch (ios->signal_voltage) {
>> +	case MMC_SIGNAL_VOLTAGE_330:
>> +		if (!(host->flags & SDHCI_SIGNALING_330))
>> +			return -EINVAL;
>> +
>> +		/* Set 1.8V Signal Enable in the Host Control2 register to 0 */
>> +		ctrl &= ~SDHCI_CTRL_VDD_180;
>> +		break;
>> +	case MMC_SIGNAL_VOLTAGE_180:
>> +		if (!(host->flags & SDHCI_SIGNALING_180))
>> +			return -EINVAL;
>> +
>> +		/*
>> +		 * Enable 1.8V Signal Enable in the Host Control2
>> +		 * register
>> +		 */
>> +		ctrl |= SDHCI_CTRL_VDD_180;
>> +		break;
>> +	case MMC_SIGNAL_VOLTAGE_120:
>> +		if (!(host->flags & SDHCI_SIGNALING_120))
>> +			return -EINVAL;
>> +		return 0;
>> +	default:
>> +		/* No signal voltage switch required */
>> +		return 0;
>> +	}
>> +
>> +	sdhci_writew(host, ctrl, SDHCI_HOST_CONTROL2);
>> +
>> +	/* Wait for 5ms */
>> +	usleep_range(5000, 5500);
>> +
>> +	/* regulator output should be stable within 5 ms */
>> +	status = !!(ctrl & SDHCI_CTRL_VDD_180);
>> +	ctrl = sdhci_readw(host, SDHCI_HOST_CONTROL2);
>> +	if (!!(ctrl &  SDHCI_CTRL_VDD_180) == status)
> You should be able to drop the !! both here and when assigning status.
>
> Overall this looks neater, thanks for reworking it.
>
> Regards,
> Bjorn


Thanks

Veera
Veerabhadrarao Badiganti May 28, 2020, 7:13 a.m. UTC | #2
On 5/22/2020 10:34 PM, Bjorn Andersson wrote:
> On Fri 22 May 06:27 PDT 2020, Veerabhadrarao Badiganti wrote:
>
>> Hi Bjorn,
>>
>> On 5/22/2020 12:37 AM, Bjorn Andersson wrote:
>>> On Thu 21 May 08:23 PDT 2020, Veerabhadrarao Badiganti wrote:
>>>
>>>> On qcom SD host controllers voltage switching be done after the HW
>>>> is ready for it. The HW informs its readiness through power irq.
>>>> The voltage switching should happen only then.
>>>>
>>>> Use the internal voltage switching and then control the voltage
>>>> switching using power irq.
>>>>
>>>> Set the regulator load as well so that regulator can be configured
>>>> in LPM mode when in is not being used.
>>>>
>>>> Co-developed-by: Asutosh Das <asutoshd@codeaurora.org>
>>>> Signed-off-by: Asutosh Das <asutoshd@codeaurora.org>
>>>> Co-developed-by: Vijay Viswanath <vviswana@codeaurora.org>
>>>> Signed-off-by: Vijay Viswanath <vviswana@codeaurora.org>
>>>> Co-developed-by: Veerabhadrarao Badiganti <vbadigan@codeaurora.org>
>>>> Signed-off-by: Veerabhadrarao Badiganti <vbadigan@codeaurora.org>
>>> Looks better, thanks.
>>>
>>>> ---
>>>>    drivers/mmc/host/sdhci-msm.c | 207 +++++++++++++++++++++++++++++++++++++++++--
>>>>    1 file changed, 198 insertions(+), 9 deletions(-)
>>>>
>>>> diff --git a/drivers/mmc/host/sdhci-msm.c b/drivers/mmc/host/sdhci-msm.c
>>> [..]
>>>>    static const struct sdhci_msm_offset *sdhci_priv_msm_offset(struct sdhci_host *host)
>>>> @@ -1298,6 +1302,71 @@ static void sdhci_msm_set_uhs_signaling(struct sdhci_host *host,
>>>>    		sdhci_msm_hs400(host, &mmc->ios);
>>>>    }
>>>> +static int sdhci_msm_set_vmmc(struct mmc_host *mmc)
>>>> +{
>>>> +	int ret;
>>>> +
>>>> +	if (IS_ERR(mmc->supply.vmmc))
>>>> +		return 0;
>>>> +
>>>> +	ret = mmc_regulator_set_ocr(mmc, mmc->supply.vmmc, mmc->ios.vdd);
>>>> +	if (ret)
>>>> +		dev_err(mmc_dev(mmc), "%s: vmmc set ocr with vdd=%d failed: %d\n",
>>>> +			mmc_hostname(mmc), mmc->ios.vdd, ret);
>>> Missed this one on v1, in the event that mmc_regulator_set_ocr() return
>>> a non-zero value it has already printed an error message. So please
>>> replace the tail with just:
>>>
>>> 	return mmc_regulator_set_ocr(...);
>>>
>>>> +
>>>> +	return ret;
>>>> +}
>>>> +
>>>> +static int sdhci_msm_set_vqmmc(struct sdhci_msm_host *msm_host,
>>>> +			      struct mmc_host *mmc, bool level)
>>>> +{
>>>> +	int load, ret;
>>>> +	struct mmc_ios ios;
>>>> +
>>>> +	if (IS_ERR(mmc->supply.vqmmc)			 ||
>>>> +	    (mmc->ios.power_mode == MMC_POWER_UNDEFINED) ||
>>>> +	    (msm_host->vqmmc_enabled == level))
>>>> +		return 0;
>>>> +
>>>> +	if (msm_host->vqmmc_load) {
>>>> +		load = level ? msm_host->vqmmc_load : 0;
>>>> +		ret = regulator_set_load(mmc->supply.vqmmc, load);
>>> Sorry for the late reply on v1, but please see my explanation regarding
>>> load and always-on regulators there.
>> <Merging your comment from V1 here>
>>
>>>> You should still call regulator_enable()/regulator_disable() on your
>>>> consumer regulator in this driver. When you do this the regulator core
>>>> will conclude that the regulator_dev (i.e. the part that represents the
>>>> hardware) is marked always_on and will not enable/disable the regulator.
>>>> But it will still invoke _regulator_handle_consumer_enable() and
>>>> _regulator_handle_consumer_disable(), which will aggregate the "load" of
>>>> all client regulators and update the regulator's load.
>>>> So this will apply the load as you expect regardless of it being
>>>> supplied by a regulator marked as always_on.
>> Since I'm not turning off this regulator for eMMC, I wanted to keep it in
>> LPM mode
>> to save some power.
>> When the regulator configured in auto mode (RPMH_REGULATOR_MODE_AUTO) it
>> switches to LPM/HPM mode based on the active load.
>> So i have to minimize my driver load requirement so that I can let this
>> regulator
>> in LPM mode.
>> So i need to set load every-time I disable/enable the regulator.
>>
> You call regulator_enable(vqmmc) and regulator_disable() below, so you
> are telling the regulator framework that your struct regulator should be
> "on" or "off".
>
> This will cause the sum of all struct regulator's for the underlying
> struct regulator_dev to be recalculated. So after calling
> regulator_disable() below your effective addition to the load
> calculation is 0, regardless of which load you have specified.
>
> Independent of this the property regulator-always-on (always_on in
> struct regulator_dev) will determine if the enable/disable request will
> actually be sent to the RPMh.
>
>
> So, if you where to not call regulator_disable() for eMMC your argument
> is correct, but as far as I can see you are and you're relying on the
> regulator core to keep it always-on - and then the load logic is in
> effect still.
Thanks for the details Bjorn.
My requirement is, for eMMC i shouldn't be turning this regulator off. 
But has to configure in LPM mode.
For SD-card, i have to turn-off this regulator.
So I'm planning to update the logic as below, let me know if you have 
any other suggestions.

+static int sdhci_msm_set_vqmmc(struct sdhci_msm_host *msm_host,
+                             struct mmc_host *mmc, bool level)
+{
+       int ret;
+       bool always_on;
+
+       if (IS_ERR(mmc->supply.vqmmc)           ||
+           (mmc->ios.power_mode == MMC_POWER_UNDEFINED))
+               return 0;
+       /*
+        * For eMMC don't turn off Vqmmc, Instead just configure it in LPM
+        * and HPM modes by setting the right amount of load.
+        */
+       always_on = mmc->card && mmc_card_mmc(mmc->card);
+
+       if (always_on)
+               ret = msm_config_vqmmc_mode(msm_host, mmc, level);
+       else
+               ret = msm_toggle_vqmmc(msm_host, mmc, level);
+
+       return ret;
+}
> Regards,
> Bjorn
>
>>>> +		if (ret) {
>>>> +			dev_err(mmc_dev(mmc), "%s: vqmmc set load failed: %d\n",
>>>> +				mmc_hostname(mmc), ret);
>>>> +			goto out;
>>>> +		}
>>>> +	}
>>>> +
>>>> +	if (level) {
>>>> +		/* Set the IO voltage regulator to default voltage level */
>>>> +		if (msm_host->caps_0 & CORE_3_0V_SUPPORT)
>>>> +			ios.signal_voltage = MMC_SIGNAL_VOLTAGE_330;
>>>> +		else if (msm_host->caps_0 & CORE_1_8V_SUPPORT)
>>>> +			ios.signal_voltage = MMC_SIGNAL_VOLTAGE_180;
>>>> +
>>>> +		if (msm_host->caps_0 & CORE_VOLT_SUPPORT) {
>>>> +			ret = mmc_regulator_set_vqmmc(mmc, &ios);
>>>> +			if (ret < 0) {
>>>> +				dev_err(mmc_dev(mmc), "%s: vqmmc set volgate failed: %d\n",
>>>> +					mmc_hostname(mmc), ret);
>>>> +				goto out;
>>>> +			}
>>>> +		}
>>>> +		ret = regulator_enable(mmc->supply.vqmmc);
>>>> +	} else {
>>>> +		ret = regulator_disable(mmc->supply.vqmmc);
>>>> +	}
>>>> +
>>>> +	if (ret)
>>>> +		dev_err(mmc_dev(mmc), "%s: vqmm %sable failed: %d\n",
>>>> +			mmc_hostname(mmc), level ? "en":"dis", ret);
>>>> +	else
>>>> +		msm_host->vqmmc_enabled = level;
>>>> +out:
>>>> +	return ret;
>>>> +}