| Message ID | 20210114152700.21916-1-muhammad.husaini.zulkifli@intel.com |
|---|---|
| Headers | show |
| Series | mmc: sdhci-of-arasan: Add UHS-1 support for Keem Bay SOC | expand |
On Thu, Jan 14, 2021 at 11:26:56PM +0800, Muhammad Husaini Zulkifli wrote: > Export inline function to encapsulate AON_CFG1 for controling the I/O Rail > supplied voltage levels which communicate with Trusted Firmware. Adding Sudeep for the SMCCC bits, not deleting any context for his benefit. > Signed-off-by: Muhammad Husaini Zulkifli <muhammad.husaini.zulkifli@intel.com> > Acked-by: Andy Shevchenko <andriy.shevchenko@intel.com> > --- > include/linux/firmware/intel/keembay.h | 82 ++++++++++++++++++++++++++ > 1 file changed, 82 insertions(+) > create mode 100644 include/linux/firmware/intel/keembay.h > > diff --git a/include/linux/firmware/intel/keembay.h b/include/linux/firmware/intel/keembay.h > new file mode 100644 > index 000000000000..f5a8dbfdb63b > --- /dev/null > +++ b/include/linux/firmware/intel/keembay.h > @@ -0,0 +1,82 @@ > +/* SPDX-License-Identifier: GPL-2.0 */ > +/* > + * Intel Keembay SOC Firmware API Layer > + * > + * Copyright (C) 2020, Intel Corporation > + * > + * Author: Muhammad Husaini Zulkifli <Muhammad.Husaini.Zulkifli@intel.com> > + */ > + > +#ifndef __FIRMWARE_KEEMBAY_SMC_H__ > +#define __FIRMWARE_KEEMBAY_SMC_H__ > + > +#include <linux/arm-smccc.h> > + > +/* > + * This file defines an API function that can be called by a device driver in order to > + * communicate with Trusted Firmware - A profile(TF-A) or Trusted Firmware - M profile (TF-M). > + */ > + > +#define KEEMBAY_SET_1V8_IO_RAIL 1 > +#define KEEMBAY_SET_3V3_IO_RAIL 0 > + > +#define KEEMBAY_IOV_1_8V_uV 1800000 > +#define KEEMBAY_IOV_3_3V_uV 3300000 > + > +#define KEEMBAY_SET_SD_VOLTAGE_ID 0xFF26 > +#define KEEMBAY_GET_SD_VOLTAGE_ID 0xFF2A > + > +#define ARM_SMCCC_SIP_KEEMBAY_SET_SD_VOLTAGE \ > + ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, \ > + ARM_SMCCC_SMC_32, \ > + ARM_SMCCC_OWNER_SIP, \ > + KEEMBAY_SET_SD_VOLTAGE_ID) > + > +#define ARM_SMCCC_SIP_KEEMBAY_GET_SD_VOLTAGE \ > + ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, \ > + ARM_SMCCC_SMC_32, \ > + ARM_SMCCC_OWNER_SIP, \ > + KEEMBAY_GET_SD_VOLTAGE_ID) > + > +#define KEEMBAY_REG_NUM_CONSUMERS 2 > + > +struct keembay_reg_supply { > + struct regulator *consumer; > +}; > + > +#if IS_ENABLED(CONFIG_HAVE_ARM_SMCCC_DISCOVERY) > +/* > + * Voltage applied on the IO Rail is controlled from the Always On Register using specific > + * bits in AON_CGF1 register. This is a secure register. Keem Bay SOC cannot exposed this > + * register address to the outside world. > + */ > +static inline int keembay_set_io_rail_supplied_voltage(int volt) > +{ > + struct arm_smccc_res res; > + > + arm_smccc_1_1_invoke(ARM_SMCCC_SIP_KEEMBAY_SET_SD_VOLTAGE, volt, &res); There is a SCMI voltage domain protocol intended for just this use case of controlling regulators managed by the firmware, why are you not using that for these systems? See drivers/firmware/arm_scmi/voltage.c. > + > + return res.a0; > +} > + > +static inline int keembay_get_io_rail_supplied_voltage(void) > +{ > + struct arm_smccc_res res; > + > + arm_smccc_1_1_invoke(ARM_SMCCC_SIP_KEEMBAY_GET_SD_VOLTAGE, &res); > + > + return res.a1; > +} > +#else > +static inline int keembay_set_io_rail_supplied_voltage(int volt) > +{ > + return -ENODEV; > +} > + > +static inline int keembay_get_io_rail_supplied_voltage(void) > +{ > + return -ENODEV; > +} > +#endif > + > +#endif /* __FIRMWARE_KEEMBAY_SMC_H__ */ > -- > 2.17.1 >
On Thu, Jan 14, 2021 at 11:26:59PM +0800, Muhammad Husaini Zulkifli wrote: > Keem Bay SD regulator driver module is added to encapsulate ARM > Secure Monitor Call Calling Convention (SMCCC) during set voltage > operations to control I/O Rail supplied voltage levels which communicate > with Trusted Firmware. Adding in Sudeep again for the SMCCC bits. I just checked and am I right in thinking this might already be shipping in production? > @@ -0,0 +1,112 @@ > +// SPDX-License-Identifier: GPL-2.0 > +/* > + * Intel Keem Bay SD Regulator > + * > + * Copyright (C) 2020, Intel Corporation > + * Author: Muhammad Husaini Zulkifli <Muhammad.Husaini.Zulkifli@intel.com> > + */ Please make the entire comment a C++ comment to improve legibility. > +static int keembay_regulator_set_voltage(struct regulator_dev *dev, > + int min_uV, int max_uV, > + unsigned *selector) > +{ > + int tmp_volt; > + > + if (min_uV == KEEMBAY_IOV_1_8V_uV && max_uV == KEEMBAY_IOV_1_8V_uV) > + tmp_volt = KEEMBAY_SET_1V8_IO_RAIL; > + else > + tmp_volt = KEEMBAY_SET_3V3_IO_RAIL; > + > + return keembay_set_io_rail_supplied_voltage(tmp_volt); > +} This has serious problems with input validation and is broken for most valid input. A set_voltage() function should set the voltage to the lowest valid voltage within the range specified in the arguments and return an error if it is not possible to set a voltage within the range specified by the arguments. This function will set 3.3V for any input range other than exactly 1.8V so for example if the caller validly sets a range of 1.7V-1.9V then 3.3V will be selected instead of 1.8V, or if the user sets 1.0-1.1V then it will set 3.3V instead of returning an error. > +static int keembay_regulator_get_voltage(struct regulator_dev *dev) > +{ > + int ret; > + > + ret = keembay_get_io_rail_supplied_voltage(); > + > + return ret ? KEEMBAY_IOV_1_8V_uV : KEEMBAY_IOV_3_3V_uV; > +} This ignores any errors or out of bounds values returned by the called function, and please write normal conditional statements rather than using the ternery operator to improve legibility. > +/* > + * Using subsys_initcall to ensure that Keem Bay regulator platform driver > + * is initialized before device driver try to utilize it. > + */ > +subsys_initcall(keembay_regulator_init); There is no need to do this, probe deferral will ensure that dependencies will be resolved.
On Thu, Jan 14, 2021 at 04:48:11PM +0000, Mark Brown wrote: > On Thu, Jan 14, 2021 at 11:26:56PM +0800, Muhammad Husaini Zulkifli wrote: > > Export inline function to encapsulate AON_CFG1 for controling the I/O Rail > > supplied voltage levels which communicate with Trusted Firmware. > > Adding Sudeep for the SMCCC bits, not deleting any context for his > benefit. > Thanks Mark for cc-ing me and joining the dots. I completely forgot about that fact that this platform was using SCMI using SMC as transport. Sorry for that and it is my fault. I did review the SCMI/SMC support for this platform sometime in June/July last year and forgot the fact it is same platform when voltage/regulator support patches for SD/MMC was posted sometime later last year. I concentrated on SMCCC conventions and other details. [...] > > +#define ARM_SMCCC_SIP_KEEMBAY_SET_SD_VOLTAGE \ > > + ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, \ > > + ARM_SMCCC_SMC_32, \ > > + ARM_SMCCC_OWNER_SIP, \ > > + KEEMBAY_SET_SD_VOLTAGE_ID) > > + > > +#define ARM_SMCCC_SIP_KEEMBAY_GET_SD_VOLTAGE \ > > + ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, \ > > + ARM_SMCCC_SMC_32, \ > > + ARM_SMCCC_OWNER_SIP, \ > > + KEEMBAY_GET_SD_VOLTAGE_ID) > > + > > +#define KEEMBAY_REG_NUM_CONSUMERS 2 > > + > > +struct keembay_reg_supply { > > + struct regulator *consumer; > > +}; > > + > > +#if IS_ENABLED(CONFIG_HAVE_ARM_SMCCC_DISCOVERY) > > +/* > > + * Voltage applied on the IO Rail is controlled from the Always On Register using specific > > + * bits in AON_CGF1 register. This is a secure register. Keem Bay SOC cannot exposed this > > + * register address to the outside world. > > + */ > > +static inline int keembay_set_io_rail_supplied_voltage(int volt) > > +{ > > + struct arm_smccc_res res; > > + > > + arm_smccc_1_1_invoke(ARM_SMCCC_SIP_KEEMBAY_SET_SD_VOLTAGE, volt, &res); > > There is a SCMI voltage domain protocol intended for just this use case > of controlling regulators managed by the firmware, why are you not using > that for these systems? See drivers/firmware/arm_scmi/voltage.c. > Indeed. Please switch to using the new voltage protocol added for this without any extra code. You just need to wire up DT for this. Just for curiosity, where is SCMI platform firmware implemented ? On Cortex-A, secure side or external processor. Does this platform run TF-A ? -- Regards, Sudeep
On Thu, Jan 14, 2021 at 05:14:34PM +0000, Mark Brown wrote: > On Thu, Jan 14, 2021 at 11:26:59PM +0800, Muhammad Husaini Zulkifli wrote: > > > Keem Bay SD regulator driver module is added to encapsulate ARM > > Secure Monitor Call Calling Convention (SMCCC) during set voltage > > operations to control I/O Rail supplied voltage levels which communicate > > with Trusted Firmware. > > Adding in Sudeep again for the SMCCC bits. I just checked and am I > right in thinking this might already be shipping in production? > OK you seem to have asked the most important question here directly. I have asked for the details of platform firmware implementation in the other email basically for the same reason(to check how feasible is it to move to new SCMI voltage protocol). Some work in the firmware, but if the implement is on the A-profile itself in TF-A or any other equivalent, it should not be too difficult. -- Regards, Sudeep
Hi Sudeep and Mark, Thanks for the review. I replied inline. >-----Original Message----- >From: Sudeep Holla <sudeep.holla@arm.com> >Sent: Saturday, January 16, 2021 2:58 AM >To: Mark Brown <broonie@kernel.org> >Cc: Zulkifli, Muhammad Husaini <muhammad.husaini.zulkifli@intel.com>; >ulf.hansson@linaro.org; lgirdwood@gmail.com; robh+dt@kernel.org; >devicetree@vger.kernel.org; Hunter, Adrian <adrian.hunter@intel.com>; >michal.simek@xilinx.com; linux-mmc@vger.kernel.org; linux- >kernel@vger.kernel.org; Shevchenko, Andriy ><andriy.shevchenko@intel.com>; A, Rashmi <rashmi.a@intel.com>; Vaidya, >Mahesh R <mahesh.r.vaidya@intel.com>; Sudeep Holla ><sudeep.holla@arm.com> >Subject: Re: [PATCH v1 5/9] firmware: keembay: Add support for Trusted >Firmware Service call > >On Thu, Jan 14, 2021 at 04:48:11PM +0000, Mark Brown wrote: >> On Thu, Jan 14, 2021 at 11:26:56PM +0800, Muhammad Husaini Zulkifli >wrote: >> > Export inline function to encapsulate AON_CFG1 for controling the >> > I/O Rail supplied voltage levels which communicate with Trusted Firmware. >> >> Adding Sudeep for the SMCCC bits, not deleting any context for his >> benefit. >> > >Thanks Mark for cc-ing me and joining the dots. I completely forgot about that >fact that this platform was using SCMI using SMC as transport. Sorry for that and >it is my fault. I did review the SCMI/SMC support for this platform sometime in >June/July last year and forgot the fact it is same platform when >voltage/regulator support patches for SD/MMC was posted sometime later last >year. I concentrated on SMCCC conventions and other details. Yes Sudeep. I redesigned the way I handle the smccc call. Previously it was handled directly in mmc driver. After few discussion, we conclude to create an abstraction using regulator framework to encapsulate this smccc call during set voltage operation. Using standard abstraction will make things easier for the maintainer. > >[...] > >> > +#define ARM_SMCCC_SIP_KEEMBAY_SET_SD_VOLTAGE \ >> > + ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, \ >> > + ARM_SMCCC_SMC_32, \ >> > + ARM_SMCCC_OWNER_SIP, \ >> > + KEEMBAY_SET_SD_VOLTAGE_ID) >> > + >> > +#define ARM_SMCCC_SIP_KEEMBAY_GET_SD_VOLTAGE \ >> > + ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, \ >> > + ARM_SMCCC_SMC_32, \ >> > + ARM_SMCCC_OWNER_SIP, \ >> > + KEEMBAY_GET_SD_VOLTAGE_ID) >> > + >> > +#define KEEMBAY_REG_NUM_CONSUMERS 2 >> > + >> > +struct keembay_reg_supply { >> > + struct regulator *consumer; >> > +}; >> > + >> > +#if IS_ENABLED(CONFIG_HAVE_ARM_SMCCC_DISCOVERY) >> > +/* >> > + * Voltage applied on the IO Rail is controlled from the Always On >> > +Register using specific >> > + * bits in AON_CGF1 register. This is a secure register. Keem Bay >> > +SOC cannot exposed this >> > + * register address to the outside world. >> > + */ >> > +static inline int keembay_set_io_rail_supplied_voltage(int volt) { >> > + struct arm_smccc_res res; >> > + >> > + > arm_smccc_1_1_invoke(ARM_SMCCC_SIP_KEEMBAY_SET_SD_VOLTA >GE, volt, >> > +&res); >> >> There is a SCMI voltage domain protocol intended for just this use >> case of controlling regulators managed by the firmware, why are you >> not using that for these systems? See drivers/firmware/arm_scmi/voltage.c. From mmc maintainer's perspective, I should use the common modelling either using regulator framework or pinctrl to perform voltage operation. Not just directly invoke smccc call in the mmc driver. That is why I came up with this regulator driver to perform voltage operation. >> > >Indeed. Please switch to using the new voltage protocol added for this without >any extra code. You just need to wire up DT for this. May I know even if I wire up the DT, how should I call this from the mmc driver For set/get voltage operation? Any example? > >Just for curiosity, where is SCMI platform firmware implemented ? On Cortex- >A, secure side or external processor. Does this platform run TF-A ? The KMB SCMI framework is implemented in secure runtime firmware (TF-A BL31). Hopefully I am answering your question. > >-- >Regards, >Sudeep
On Mon, Jan 18, 2021 at 10:28:33AM +0000, Zulkifli, Muhammad Husaini wrote: > >> There is a SCMI voltage domain protocol intended for just this use > >> case of controlling regulators managed by the firmware, why are you > >> not using that for these systems? See drivers/firmware/arm_scmi/voltage.c. > From mmc maintainer's perspective, I should use the common modelling either using > regulator framework or pinctrl to perform voltage operation. Not just directly invoke > smccc call in the mmc driver. That is why I came up with this regulator driver to perform > voltage operation. The above is a standard way of controlling regulators via SMCCC which already has a regulator driver, you're duplicating this functionality. > >Indeed. Please switch to using the new voltage protocol added for this without > >any extra code. You just need to wire up DT for this. > May I know even if I wire up the DT, how should I call this from the mmc driver > For set/get voltage operation? Any example? There's one in the binding document for the driver.
On Mon, Jan 18, 2021 at 10:28:33AM +0000, Zulkifli, Muhammad Husaini wrote: > Hi Sudeep and Mark, > > Thanks for the review. I replied inline. > > >-----Original Message----- > >From: Sudeep Holla <sudeep.holla@arm.com> > >Sent: Saturday, January 16, 2021 2:58 AM > >To: Mark Brown <broonie@kernel.org> > >Cc: Zulkifli, Muhammad Husaini <muhammad.husaini.zulkifli@intel.com>; > >ulf.hansson@linaro.org; lgirdwood@gmail.com; robh+dt@kernel.org; > >devicetree@vger.kernel.org; Hunter, Adrian <adrian.hunter@intel.com>; > >michal.simek@xilinx.com; linux-mmc@vger.kernel.org; linux- > >kernel@vger.kernel.org; Shevchenko, Andriy > ><andriy.shevchenko@intel.com>; A, Rashmi <rashmi.a@intel.com>; Vaidya, > >Mahesh R <mahesh.r.vaidya@intel.com>; Sudeep Holla > ><sudeep.holla@arm.com> > >Subject: Re: [PATCH v1 5/9] firmware: keembay: Add support for Trusted > >Firmware Service call > > > >On Thu, Jan 14, 2021 at 04:48:11PM +0000, Mark Brown wrote: > >> On Thu, Jan 14, 2021 at 11:26:56PM +0800, Muhammad Husaini Zulkifli > >wrote: > >> > Export inline function to encapsulate AON_CFG1 for controling the > >> > I/O Rail supplied voltage levels which communicate with Trusted Firmware. > >> > >> Adding Sudeep for the SMCCC bits, not deleting any context for his > >> benefit. > >> > > > >Thanks Mark for cc-ing me and joining the dots. I completely forgot about that > >fact that this platform was using SCMI using SMC as transport. Sorry for that and > >it is my fault. I did review the SCMI/SMC support for this platform sometime in > >June/July last year and forgot the fact it is same platform when > >voltage/regulator support patches for SD/MMC was posted sometime later last > >year. I concentrated on SMCCC conventions and other details. > > Yes Sudeep. I redesigned the way I handle the smccc call. Previously it was handled directly in mmc driver. > After few discussion, we conclude to create an abstraction using regulator framework to encapsulate this smccc call > during set voltage operation. Using standard abstraction will make things easier for the maintainer. > > > > >[...] > > > >> > +#define ARM_SMCCC_SIP_KEEMBAY_SET_SD_VOLTAGE \ > >> > + ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, \ > >> > + ARM_SMCCC_SMC_32, \ > >> > + ARM_SMCCC_OWNER_SIP, \ > >> > + KEEMBAY_SET_SD_VOLTAGE_ID) > >> > + > >> > +#define ARM_SMCCC_SIP_KEEMBAY_GET_SD_VOLTAGE \ > >> > + ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, \ > >> > + ARM_SMCCC_SMC_32, \ > >> > + ARM_SMCCC_OWNER_SIP, \ > >> > + KEEMBAY_GET_SD_VOLTAGE_ID) > >> > + > >> > +#define KEEMBAY_REG_NUM_CONSUMERS 2 > >> > + > >> > +struct keembay_reg_supply { > >> > + struct regulator *consumer; > >> > +}; > >> > + > >> > +#if IS_ENABLED(CONFIG_HAVE_ARM_SMCCC_DISCOVERY) > >> > +/* > >> > + * Voltage applied on the IO Rail is controlled from the Always On > >> > +Register using specific > >> > + * bits in AON_CGF1 register. This is a secure register. Keem Bay > >> > +SOC cannot exposed this > >> > + * register address to the outside world. > >> > + */ > >> > +static inline int keembay_set_io_rail_supplied_voltage(int volt) { > >> > + struct arm_smccc_res res; > >> > + > >> > + > > arm_smccc_1_1_invoke(ARM_SMCCC_SIP_KEEMBAY_SET_SD_VOLTA > >GE, volt, > >> > +&res); > >> > >> There is a SCMI voltage domain protocol intended for just this use > >> case of controlling regulators managed by the firmware, why are you > >> not using that for these systems? See drivers/firmware/arm_scmi/voltage.c. > > From mmc maintainer's perspective, I should use the common modelling either > using regulator framework or pinctrl to perform voltage operation. Not just > directly invoke smccc call in the mmc driver. That is why I came up with > this regulator driver to perform voltage operation. > That's correct. Since the platform uses SCMI and SCMI spec[1] supports Voltage protocol and there is upstream driver[2] to support it, I see no point in duplicating the support with another custom/non-standard solution. > >> > > > >Indeed. Please switch to using the new voltage protocol added for this without > >any extra code. You just need to wire up DT for this. > > May I know even if I wire up the DT, how should I call this from the mmc driver > For set/get voltage operation? Any example? > Mark has already pointed you to the binding document[3] > > > >Just for curiosity, where is SCMI platform firmware implemented ? On Cortex- > >A, secure side or external processor. Does this platform run TF-A ? > > The KMB SCMI framework is implemented in secure runtime firmware (TF-A BL31). > Hopefully I am answering your question. > Yes, it should be easy to extend the implementation and add support for voltage protocol. If you still face any issues, please ask any queries on the list cc-ing me and Cristian Marussi(cc-ed) -- Regards, Sudeep [1] https://developer.arm.com/documentation/den0056/latest [2] drivers/firmware/arm_scmi/voltage.c + drivers/regulator/scmi-regulator.c [3] Documentation/devicetree/bindings/arm/arm,scmi.txt
Hi sorry I'm a bit late on this. On Mon, Jan 18, 2021 at 10:28:33AM +0000, Zulkifli, Muhammad Husaini wrote: > Hi Sudeep and Mark, > > Thanks for the review. I replied inline. > > >-----Original Message----- > >From: Sudeep Holla <sudeep.holla@arm.com> > >Sent: Saturday, January 16, 2021 2:58 AM > >To: Mark Brown <broonie@kernel.org> > >Cc: Zulkifli, Muhammad Husaini <muhammad.husaini.zulkifli@intel.com>; > >ulf.hansson@linaro.org; lgirdwood@gmail.com; robh+dt@kernel.org; > >devicetree@vger.kernel.org; Hunter, Adrian <adrian.hunter@intel.com>; > >michal.simek@xilinx.com; linux-mmc@vger.kernel.org; linux- > >kernel@vger.kernel.org; Shevchenko, Andriy > ><andriy.shevchenko@intel.com>; A, Rashmi <rashmi.a@intel.com>; Vaidya, > >Mahesh R <mahesh.r.vaidya@intel.com>; Sudeep Holla > ><sudeep.holla@arm.com> > >Subject: Re: [PATCH v1 5/9] firmware: keembay: Add support for Trusted > >Firmware Service call > > > >On Thu, Jan 14, 2021 at 04:48:11PM +0000, Mark Brown wrote: > >> On Thu, Jan 14, 2021 at 11:26:56PM +0800, Muhammad Husaini Zulkifli > >wrote: > >> > Export inline function to encapsulate AON_CFG1 for controling the > >> > I/O Rail supplied voltage levels which communicate with Trusted Firmware. > >> > >> Adding Sudeep for the SMCCC bits, not deleting any context for his > >> benefit. > >> > > > >Thanks Mark for cc-ing me and joining the dots. I completely forgot about that > >fact that this platform was using SCMI using SMC as transport. Sorry for that and > >it is my fault. I did review the SCMI/SMC support for this platform sometime in > >June/July last year and forgot the fact it is same platform when > >voltage/regulator support patches for SD/MMC was posted sometime later last > >year. I concentrated on SMCCC conventions and other details. > > Yes Sudeep. I redesigned the way I handle the smccc call. Previously it was handled directly in mmc driver. > After few discussion, we conclude to create an abstraction using regulator framework to encapsulate this smccc call > during set voltage operation. Using standard abstraction will make things easier for the maintainer. > > > > >[...] > > > >> > +#define ARM_SMCCC_SIP_KEEMBAY_SET_SD_VOLTAGE \ > >> > + ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, \ > >> > + ARM_SMCCC_SMC_32, \ > >> > + ARM_SMCCC_OWNER_SIP, \ > >> > + KEEMBAY_SET_SD_VOLTAGE_ID) > >> > + > >> > +#define ARM_SMCCC_SIP_KEEMBAY_GET_SD_VOLTAGE \ > >> > + ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, \ > >> > + ARM_SMCCC_SMC_32, \ > >> > + ARM_SMCCC_OWNER_SIP, \ > >> > + KEEMBAY_GET_SD_VOLTAGE_ID) > >> > + > >> > +#define KEEMBAY_REG_NUM_CONSUMERS 2 > >> > + > >> > +struct keembay_reg_supply { > >> > + struct regulator *consumer; > >> > +}; > >> > + > >> > +#if IS_ENABLED(CONFIG_HAVE_ARM_SMCCC_DISCOVERY) > >> > +/* > >> > + * Voltage applied on the IO Rail is controlled from the Always On > >> > +Register using specific > >> > + * bits in AON_CGF1 register. This is a secure register. Keem Bay > >> > +SOC cannot exposed this > >> > + * register address to the outside world. > >> > + */ > >> > +static inline int keembay_set_io_rail_supplied_voltage(int volt) { > >> > + struct arm_smccc_res res; > >> > + > >> > + > > arm_smccc_1_1_invoke(ARM_SMCCC_SIP_KEEMBAY_SET_SD_VOLTA > >GE, volt, > >> > +&res); > >> > >> There is a SCMI voltage domain protocol intended for just this use > >> case of controlling regulators managed by the firmware, why are you > >> not using that for these systems? See drivers/firmware/arm_scmi/voltage.c. > > From mmc maintainer's perspective, I should use the common modelling either using > regulator framework or pinctrl to perform voltage operation. Not just directly invoke > smccc call in the mmc driver. That is why I came up with this regulator driver to perform > voltage operation. > There is an SCMI regulator driver built on top of SCMIv3.0 Voltage Domain Protocol, so as long as your SCMI platform firmware support the new VD protocol you should be able to wire up a generic SCMI regulator in the DT (as described in the arm,scmi.txt bindings) and then just use the usual regulator framework ops with such SCMI regulator without the need to add a custom regulator using custom SMCCC calls). Thanks Cristian > >> > > > >Indeed. Please switch to using the new voltage protocol added for this without > >any extra code. You just need to wire up DT for this. > > May I know even if I wire up the DT, how should I call this from the mmc driver > For set/get voltage operation? Any example? > > > > >Just for curiosity, where is SCMI platform firmware implemented ? On Cortex- > >A, secure side or external processor. Does this platform run TF-A ? > > The KMB SCMI framework is implemented in secure runtime firmware (TF-A BL31). > Hopefully I am answering your question. > > > > >-- > >Regards, > >Sudeep
Hi Cristian, Sudeep and Mark, >-----Original Message----- >From: Cristian Marussi <cristian.marussi@arm.com> >Sent: Monday, January 18, 2021 8:02 PM >To: Zulkifli, Muhammad Husaini <muhammad.husaini.zulkifli@intel.com> >Cc: Sudeep Holla <sudeep.holla@arm.com>; Mark Brown ><broonie@kernel.org>; ulf.hansson@linaro.org; lgirdwood@gmail.com; >robh+dt@kernel.org; devicetree@vger.kernel.org; Hunter, Adrian ><adrian.hunter@intel.com>; michal.simek@xilinx.com; linux- >mmc@vger.kernel.org; linux-kernel@vger.kernel.org; Shevchenko, Andriy ><andriy.shevchenko@intel.com>; A, Rashmi <rashmi.a@intel.com>; Vaidya, >Mahesh R <mahesh.r.vaidya@intel.com> >Subject: Re: [PATCH v1 5/9] firmware: keembay: Add support for Trusted >Firmware Service call > >Hi > >sorry I'm a bit late on this. > >On Mon, Jan 18, 2021 at 10:28:33AM +0000, Zulkifli, Muhammad Husaini >wrote: >> Hi Sudeep and Mark, >> >> Thanks for the review. I replied inline. >> >> >-----Original Message----- >> >From: Sudeep Holla <sudeep.holla@arm.com> >> >Sent: Saturday, January 16, 2021 2:58 AM >> >To: Mark Brown <broonie@kernel.org> >> >Cc: Zulkifli, Muhammad Husaini <muhammad.husaini.zulkifli@intel.com>; >> >ulf.hansson@linaro.org; lgirdwood@gmail.com; robh+dt@kernel.org; >> >devicetree@vger.kernel.org; Hunter, Adrian <adrian.hunter@intel.com>; >> >michal.simek@xilinx.com; linux-mmc@vger.kernel.org; linux- >> >kernel@vger.kernel.org; Shevchenko, Andriy >> ><andriy.shevchenko@intel.com>; A, Rashmi <rashmi.a@intel.com>; >> >Vaidya, Mahesh R <mahesh.r.vaidya@intel.com>; Sudeep Holla >> ><sudeep.holla@arm.com> >> >Subject: Re: [PATCH v1 5/9] firmware: keembay: Add support for >> >Trusted Firmware Service call >> > >> >On Thu, Jan 14, 2021 at 04:48:11PM +0000, Mark Brown wrote: >> >> On Thu, Jan 14, 2021 at 11:26:56PM +0800, Muhammad Husaini Zulkifli >> >wrote: >> >> > Export inline function to encapsulate AON_CFG1 for controling the >> >> > I/O Rail supplied voltage levels which communicate with Trusted >Firmware. >> >> >> >> Adding Sudeep for the SMCCC bits, not deleting any context for his >> >> benefit. >> >> >> > >> >Thanks Mark for cc-ing me and joining the dots. I completely forgot >> >about that fact that this platform was using SCMI using SMC as >> >transport. Sorry for that and it is my fault. I did review the >> >SCMI/SMC support for this platform sometime in June/July last year >> >and forgot the fact it is same platform when voltage/regulator >> >support patches for SD/MMC was posted sometime later last year. I >concentrated on SMCCC conventions and other details. >> >> Yes Sudeep. I redesigned the way I handle the smccc call. Previously it was >handled directly in mmc driver. >> After few discussion, we conclude to create an abstraction using >> regulator framework to encapsulate this smccc call during set voltage >operation. Using standard abstraction will make things easier for the >maintainer. >> >> > >> >[...] >> > >> >> > +#define ARM_SMCCC_SIP_KEEMBAY_SET_SD_VOLTAGE > \ >> >> > + ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, > \ >> >> > + ARM_SMCCC_SMC_32, \ >> >> > + ARM_SMCCC_OWNER_SIP, \ >> >> > + KEEMBAY_SET_SD_VOLTAGE_ID) >> >> > + >> >> > +#define ARM_SMCCC_SIP_KEEMBAY_GET_SD_VOLTAGE > \ >> >> > + ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, > \ >> >> > + ARM_SMCCC_SMC_32, \ >> >> > + ARM_SMCCC_OWNER_SIP, \ >> >> > + KEEMBAY_GET_SD_VOLTAGE_ID) >> >> > + >> >> > +#define KEEMBAY_REG_NUM_CONSUMERS 2 >> >> > + >> >> > +struct keembay_reg_supply { >> >> > + struct regulator *consumer; >> >> > +}; >> >> > + >> >> > +#if IS_ENABLED(CONFIG_HAVE_ARM_SMCCC_DISCOVERY) >> >> > +/* >> >> > + * Voltage applied on the IO Rail is controlled from the Always >> >> > +On Register using specific >> >> > + * bits in AON_CGF1 register. This is a secure register. Keem >> >> > +Bay SOC cannot exposed this >> >> > + * register address to the outside world. >> >> > + */ >> >> > +static inline int keembay_set_io_rail_supplied_voltage(int volt) { >> >> > + struct arm_smccc_res res; >> >> > + >> >> > + >> > arm_smccc_1_1_invoke(ARM_SMCCC_SIP_KEEMBAY_SET_SD_VOLTA >> >GE, volt, >> >> > +&res); >> >> >> >> There is a SCMI voltage domain protocol intended for just this use >> >> case of controlling regulators managed by the firmware, why are you >> >> not using that for these systems? See >drivers/firmware/arm_scmi/voltage.c. >> >> From mmc maintainer's perspective, I should use the common modelling >> either using regulator framework or pinctrl to perform voltage >> operation. Not just directly invoke smccc call in the mmc driver. That >> is why I came up with this regulator driver to perform voltage operation. >> > >There is an SCMI regulator driver built on top of SCMIv3.0 Voltage Domain >Protocol, so as long as your SCMI platform firmware support the new VD >protocol you should be able to wire up a generic SCMI regulator in the DT >(as described in the arm,scmi.txt bindings) and then just use the usual >regulator framework ops with such SCMI regulator without the need to add >a custom regulator using custom SMCCC calls). I try to hook up the DT last night. Seems like the SCMI Protocol 17 is not implemented at ATF side. Double check with ATF Team, currently we don't have SCMI voltage domain control in ARM Trusted Firmware yet as of now, that is why even if I map the function to scmi, my call will be fail. [ 2.648989] arm-scmi firmware:scmi: SCMI Notifications - Core Enabled. [ 2.656157] arm-scmi firmware:scmi: SCMI Protocol v1.0 'INTEL:KMB' Firmware version 0x1 [ 2.664513] arm-scmi firmware:scmi: SCMI protocol 23 not implemented [ 2.675898] arm-scmi firmware:scmi: SCMI protocol 17 not implemented Any possibilities that for UHS patch we go with my current regulator driver implementation? > >Thanks > >Cristian > >> >> >> > >> >Indeed. Please switch to using the new voltage protocol added for this >without >> >any extra code. You just need to wire up DT for this. >> >> May I know even if I wire up the DT, how should I call this from the mmc >driver >> For set/get voltage operation? Any example? >> >> > >> >Just for curiosity, where is SCMI platform firmware implemented ? On >Cortex- >> >A, secure side or external processor. Does this platform run TF-A ? >> >> The KMB SCMI framework is implemented in secure runtime firmware (TF-A >BL31). >> Hopefully I am answering your question. >> >> > >> >-- >> >Regards, >> >Sudeep
On Thu, 14 Jan 2021 at 16:28, Muhammad Husaini Zulkifli <muhammad.husaini.zulkifli@intel.com> wrote: > > Hi, > > This patch series adds Ultra High Speed(UHS-1) Bus Speed Mode Support for Keem Bay SoC SD Card. > Summary of each patches as per below: > > Patch 1: Use of_device_get_match_data() helper to get the match-data. > Patch 2: Convert to use np pointer instead of using pdev->dev.of_node. > Patch 3: Add struct device *dev in probe func(), so that dev pointer can be widely use in probe to make code more readable. > Patch 4: Change from dev_err to dev_err_probe() to avoid spamming logs when probe is deferred. > Patch 5: Export function to be use by device driver to configure i/o voltage rail output which communicate with Trusted Firmware. > Patch 6: Update phy and regulator supply for Keem Bay SoC. > Patch 7: Add DT Binding for Keem Bay SoC SD Regulator. > Patch 8: Add SD Regulator driver to support Keem Bay SoC. This is to model using standard regulator abstraction during voltage operation > as for Keem Bay SoC, i/o voltage rail need to be configure by setting specific bit in the AON_CFG1 Register. > AON_CFG1 Register is a secure register. Direct access to AON_CFG1 register will cause firewall violation in secure system. > Patch 9: Add Ultra High Speed (UHS-1) Support for Keem Bay SOC. For Keem Bay hardware, two regulators are been used to change the I/O bus line voltage which are "vqmmc-supply" and "sdvrail-supply". > > All of these patches was tested with Keem Bay evaluation module board. > > Kindly help to review this patch set. > > Muhammad Husaini Zulkifli (9): > mmc: sdhci-of-arasan: use of_device_get_match_data() > mmc: sdhci-of-arasan: Convert to use np instead of pdev->dev.of_node > mmc: sdhci-of-arasan: Add structure device pointer in probe function > mmc: sdhci-of-arasan: Use dev_err_probe() to avoid spamming logs > firmware: keembay: Add support for Trusted Firmware Service call > dt-bindings: mmc: Update phy and regulator supply for Keem Bay SOC > dt-bindings: regulator: keembay: Add DT binding documentation > regulator: keembay: Add regulator for Keem Bay SoC > mmc: sdhci-of-arasan: Add UHS-1 support for Keem Bay SOC > > .../devicetree/bindings/mmc/arasan,sdhci.yaml | 7 +- > .../bindings/regulator/keembay-regulator.yaml | 36 ++ > drivers/mmc/host/sdhci-of-arasan.c | 313 ++++++++++++++++-- > drivers/regulator/Kconfig | 10 + > drivers/regulator/Makefile | 1 + > drivers/regulator/keembay-sd-regulator.c | 112 +++++++ > include/linux/firmware/intel/keembay.h | 82 +++++ > 7 files changed, 532 insertions(+), 29 deletions(-) > create mode 100644 Documentation/devicetree/bindings/regulator/keembay-regulator.yaml > create mode 100644 drivers/regulator/keembay-sd-regulator.c > create mode 100644 include/linux/firmware/intel/keembay.h > > -- > 2.17.1 > Applied patch 1 to patch 4. I assume you will be respinning the rest? Thanks and kind regards Uffe
On Tue, Jan 19, 2021 at 02:38:32AM +0000, Zulkifli, Muhammad Husaini wrote: [...] > > I try to hook up the DT last night. Seems like the SCMI Protocol 17 is not > implemented at ATF side. I had guessed that. > Double check with ATF Team, currently we don't have SCMI voltage domain > control in ARM Trusted Firmware yet as of now, that is why even if I map the > function to scmi, my call will be fail. Correct, but if you already have this custom SMCCC for voltage already implemented in TF-A, I don't see it is a big deal to support voltage protocol there. > > [ 2.648989] arm-scmi firmware:scmi: SCMI Notifications - Core Enabled. > [ 2.656157] arm-scmi firmware:scmi: SCMI Protocol v1.0 'INTEL:KMB' Firmware version 0x1 > [ 2.664513] arm-scmi firmware:scmi: SCMI protocol 23 not implemented > [ 2.675898] arm-scmi firmware:scmi: SCMI protocol 17 not implemented > > Any possibilities that for UHS patch we go with my current regulator driver > implementation? Sorry absolutely no. If this platform was not using SCMI, I wouldn't have pushed back hard on this custom SMCCC. Please update TF-A to add this support. There is no point in having custom interface just for this when everything else is already using SCMI on this platform.
Hi Ulf, >-----Original Message----- >From: Ulf Hansson <ulf.hansson@linaro.org> >Sent: Tuesday, January 19, 2021 9:43 PM >To: Zulkifli, Muhammad Husaini <muhammad.husaini.zulkifli@intel.com> >Cc: Mark Brown <broonie@kernel.org>; Liam Girdwood ><lgirdwood@gmail.com>; Rob Herring <robh+dt@kernel.org>; DTML ><devicetree@vger.kernel.org>; Hunter, Adrian <adrian.hunter@intel.com>; >Michal Simek <michal.simek@xilinx.com>; linux-mmc@vger.kernel.org; Linux >Kernel Mailing List <linux-kernel@vger.kernel.org>; Shevchenko, Andriy ><andriy.shevchenko@intel.com>; A, Rashmi <rashmi.a@intel.com>; Vaidya, >Mahesh R <mahesh.r.vaidya@intel.com> >Subject: Re: [PATCH v1 0/9] mmc: sdhci-of-arasan: Add UHS-1 support for >Keem Bay SOC > >On Thu, 14 Jan 2021 at 16:28, Muhammad Husaini Zulkifli ><muhammad.husaini.zulkifli@intel.com> wrote: >> >> Hi, >> >> This patch series adds Ultra High Speed(UHS-1) Bus Speed Mode Support >for Keem Bay SoC SD Card. >> Summary of each patches as per below: >> >> Patch 1: Use of_device_get_match_data() helper to get the match-data. >> Patch 2: Convert to use np pointer instead of using pdev->dev.of_node. >> Patch 3: Add struct device *dev in probe func(), so that dev pointer can be >widely use in probe to make code more readable. >> Patch 4: Change from dev_err to dev_err_probe() to avoid spamming logs >when probe is deferred. >> Patch 5: Export function to be use by device driver to configure i/o voltage >rail output which communicate with Trusted Firmware. >> Patch 6: Update phy and regulator supply for Keem Bay SoC. >> Patch 7: Add DT Binding for Keem Bay SoC SD Regulator. >> Patch 8: Add SD Regulator driver to support Keem Bay SoC. This is to >> model using standard regulator abstraction during voltage operation as for >Keem Bay SoC, i/o voltage rail need to be configure by setting specific bit in >the AON_CFG1 Register. >> AON_CFG1 Register is a secure register. Direct access to AON_CFG1 register >will cause firewall violation in secure system. >> Patch 9: Add Ultra High Speed (UHS-1) Support for Keem Bay SOC. For >Keem Bay hardware, two regulators are been used to change the I/O bus line >voltage which are "vqmmc-supply" and "sdvrail-supply". >> >> All of these patches was tested with Keem Bay evaluation module board. >> >> Kindly help to review this patch set. >> >> Muhammad Husaini Zulkifli (9): >> mmc: sdhci-of-arasan: use of_device_get_match_data() >> mmc: sdhci-of-arasan: Convert to use np instead of pdev->dev.of_node >> mmc: sdhci-of-arasan: Add structure device pointer in probe function >> mmc: sdhci-of-arasan: Use dev_err_probe() to avoid spamming logs >> firmware: keembay: Add support for Trusted Firmware Service call >> dt-bindings: mmc: Update phy and regulator supply for Keem Bay SOC >> dt-bindings: regulator: keembay: Add DT binding documentation >> regulator: keembay: Add regulator for Keem Bay SoC >> mmc: sdhci-of-arasan: Add UHS-1 support for Keem Bay SOC >> >> .../devicetree/bindings/mmc/arasan,sdhci.yaml | 7 +- >> .../bindings/regulator/keembay-regulator.yaml | 36 ++ >> drivers/mmc/host/sdhci-of-arasan.c | 313 ++++++++++++++++-- >> drivers/regulator/Kconfig | 10 + >> drivers/regulator/Makefile | 1 + >> drivers/regulator/keembay-sd-regulator.c | 112 +++++++ >> include/linux/firmware/intel/keembay.h | 82 +++++ >> 7 files changed, 532 insertions(+), 29 deletions(-) create mode >> 100644 >> Documentation/devicetree/bindings/regulator/keembay-regulator.yaml >> create mode 100644 drivers/regulator/keembay-sd-regulator.c >> create mode 100644 include/linux/firmware/intel/keembay.h >> >> -- >> 2.17.1 >> > >Applied patch 1 to patch 4. I assume you will be respinning the rest? Thanks!! Yeah. Seems like I need to request to ATF Team to implement SCMI voltage domain control first in ARM trusted firmware. > >Thanks and kind regards >Uffe
On Thu, Jan 14, 2021 at 11:27:00PM +0800, Muhammad Husaini Zulkifli wrote: > Keem Bay SOC can support dual voltage operations for GPIO SD pins to > either 1.8V or 3.3V for bus IO line power. In order to operate the GPIOs > line for Clk, Cmd and Data on Keem Bay hardware, it is important to > configure the supplied voltage applied to their I/O Rail and the output > of the I²C expander pin. Final Voltage applied on the GPIOs line are > dependent by both supplied voltage rail and expander pin output as it is > been set after passing through the voltage sense resistor. > > Keem Bay hardware is somewhat unique in the way of how IO bus line > voltage are been controlled. > > Expander pin output is controlled by gpio-regulator. Voltage rail output > is controlled by Keem Bay SD regulator. Keem Bay SD regulator encapsulated > the Secure Monitor Calling Convention (SMCCC) to communicate with Trusted > Firmware during set voltage operation. > > Signed-off-by: Muhammad Husaini Zulkifli <muhammad.husaini.zulkifli@intel.com> > Acked-by: Adrian Hunter <adrian.hunter@intel.com> > Acked-by: Andy Shevchenko <andriy.shevchenko@intel.com> > --- > drivers/mmc/host/sdhci-of-arasan.c | 263 +++++++++++++++++++++++++++++ > 1 file changed, 263 insertions(+) > + u32 otap_delay, sel_clk_buffer; > + > + phys = of_parse_phandle(dev->of_node, "phys", 0); Normally, you'd use the phy API here. Though not required from a DT perspective. > + if (!phys) { > + dev_err(dev, "Can't get phys for sd0\n"); > + return -ENODEV; > + } > + > + of_property_read_u32(phys, "intel,keembay-emmc-phy-otap-dly", &otap_delay); > + of_property_read_u32(phys, "intel,keembay-emmc-phy-sel-clkbuf", &sel_clk_buffer); Not doucmented? For property names, I'd leave out the SoC name. Might want to use it in the next chip. Rob