Re: [PATCH v3 0/3] Support PMIC operation region for CrystalCove and XPower

["Rafael J. Wysocki" <rjw@rjwysocki.net>]

On Wednesday, November 26, 2014 10:44:34 AM Aaron Lu wrote:
> On 11/26/2014 10:35 AM, Aaron Lu wrote:
> > On 11/26/2014 04:41 AM, Rafael J. Wysocki wrote:
> >> On Tuesday, November 25, 2014 07:52:31 PM Aaron Lu wrote:
> >>> On 11/25/2014 09:47 AM, Rafael J. Wysocki wrote:
> >>>> On Friday, November 21, 2014 03:11:48 PM Aaron Lu wrote:
> >>>>> v3:
> >>>>> Only some function/variable name changes, no functiona changes:
> >>>>> - Replace the dptf/DPTF word originate from the BIOS ACPI table with more
> >>>>>   meaningful word thermal/THERMAL in all places;
> >>>>> - Eliminate the soc part in various structure and function names to make
> >>>>>   them shorter:
> >>>>>   intel_soc_pmic_opregion -> intel_pmic_opregion
> >>>>>   intel_soc_pmic_pmop_handler -> intel_pmic_pmop_handler
> >>>>>   intel_soc_pmic_install_opregion_handler -> intel_pmic_install_opregion_handler
> >>>>>   etc.
> >>>>>
> >>>>>
> >>>>> v2:
> >>>>> Place PMIC operation files under drivers/acpi/pmic instead of
> >>>>> drivers/acpi/pmic_opregion as suggested by Rafael;
> >>>>> Rename PMIC operation files to make them shorter as suggested by Rafael.
> >>>>>
> >>>>> v1:
> >>>>> On Intel Baytrail-T and Baytrail-T-CR platforms, there are two customized
> >>>>> ACPI operation regions defined for the Power Management Integrated Circuit
> >>>>> device, one is for power resource handling and one is for thermal: sensor
> >>>>> temperature reporting, trip point setting, etc. There are different PMIC
> >>>>> chips used on those platforms and though each has the same two operation
> >>>>> regions and functionality, their implementation is different so every PMIC
> >>>>> will need a driver. But since their functionality is similar, some common
> >>>>> code is abstracted into the intel_soc_pmic_opregion.c.
> >>>>>
> >>>>> The last version is posted here:
> >>>>> https://lkml.org/lkml/2014/9/8/801
> >>>>>
> >>>>> Changes since then:
> >>>>> 1 Move to drivers/acpi as discussed on the above thread;
> >>>>> 2 Added support for XPower AXP288 PMIC operation region support;
> >>>>> 3 Since operation region handler can not be removed(at the moment at least),
> >>>>>   use bool for the two operation region driver configs instead of tristate;
> >>>>>   Another reason to do this is that, with Mika's MFD ACPI support patch, all
> >>>>>   those MFD cell devices created will have the same modalias as their parent's
> >>>>>   so it doesn't make much sense to compile these drivers into modules.
> >>>>>
> >>>>> Patch 1 applies on top of Rafael's pm-next branch, and then patch 2 and
> >>>>> patch 3 needs merge of Lee's mfd/ib-mfd-iio-3.19 branch where the PMIC
> >>>>> driver XPower AXP288 and iio driver axp288_adc is located.
> >>>>>
> >>>>>
> >>>>> Aaron Lu (3):
> >>>>>   ACPI / PMIC: support PMIC operation region for CrystalCove
> >>>>>   ACPI / PMIC: support PMIC operation region for XPower AXP288
> >>>>>   ACPI / PMIC: AXP288: support virtual GPIO in ACPI table
> >>>>
> >>>> OK
> >>>>
> >>>> I've pulled the Lee's 'mfd/ib-mfd-iio-3.19' branch and applied your updated
> >>>> three on top of that.  Please check the bleeding-edge branch of linux-pm.git
> >>>> for the result.
> >>>
> >>> Thanks, and a fix patch is here:
> >>>
> >>> From: Aaron Lu <aaron.lu@intel.com>
> >>> Date: Tue, 25 Nov 2014 14:35:38 +0800
> >>> Subject: [PATCH] ACPI / PMIC: Make it possible to build PMIC driver as a module
> >>>
> >>> This can solve a problem that when axp288_adc driver is built as a
> >>> module and the PMIC driver is builtin, following error would ocur:
> >>
> >> I would prefer that to be sloved by requiring axp288_adc to be built in
> >> if the PMIC stuff is selected.  Otherwise we may need to deal with some
> >> nasty module load ordering dependencies.
> > 
> > Good point, and I saw you have solved this problem in the bleeding-edge
> > branch, thanks!

No problem, I can fix up things like that occasionally.

> >>
> >>>  drivers/built-in.o: In function `intel_xpower_pmic_get_raw_temp':
> >>>  intel_pmic_xpower.c:(.text+0xdfaa7): undefined reference to `iio_channel_get'
> >>>  intel_pmic_xpower.c:(.text+0xdfb24): undefined reference to `iio_read_channel_raw'
> >>>  intel_pmic_xpower.c:(.text+0xdfb4e): undefined reference to `iio_channel_release'
> >>>
> >>> Also, with the fix commit: 52870786ff5d ("ACPI: Use ACPI companion to
> >>> match only the first physical device"), the MFD cell device will have
> >>> its own platform modalias instead of its parent's ACPI modalias, this
> >>> made it possible for the module to be autoloaded.
> >>>
> >>> Reported-by: kbuild test robot <fengguang.wu@intel.com>
> >>> Signed-off-by: Aaron Lu <aaron.lu@intel.com>
> >>> ---
> >>>  drivers/acpi/Kconfig                  |  6 +++---
> >>>  drivers/acpi/pmic/intel_pmic_crc.c    | 12 +++++++++++-
> >>>  drivers/acpi/pmic/intel_pmic_xpower.c | 12 +++++++++++-
> >>>  3 files changed, 25 insertions(+), 5 deletions(-)
> >>>
> >>> diff --git a/drivers/acpi/Kconfig b/drivers/acpi/Kconfig
> >>> index 227f0692cbff..f9459ba4ce59 100644
> >>> --- a/drivers/acpi/Kconfig
> >>> +++ b/drivers/acpi/Kconfig
> >>> @@ -394,7 +394,7 @@ config ACPI_EXTLOG
> >>>  	  tracepoint which carries that information to userspace.
> >>>  
> >>>  menuconfig PMIC_OPREGION
> >>> -	bool "PMIC (Power Management Integrated Circuit) operation region support"
> >>> +	tristate "PMIC (Power Management Integrated Circuit) operation region support"
> >>>  	help
> >>>  	  Select this option to enable support for ACPI operation
> >>>  	  region of the PMIC chip. The operation region can be used
> >>> @@ -403,13 +403,13 @@ menuconfig PMIC_OPREGION
> >>>  
> >>>  if PMIC_OPREGION
> >>>  config CRC_PMIC_OPREGION
> >>> -	bool "ACPI operation region support for CrystalCove PMIC"
> >>> +	tristate "ACPI operation region support for CrystalCove PMIC"
> >>>  	depends on INTEL_SOC_PMIC
> >>>  	help
> >>>  	  This config adds ACPI operation region support for CrystalCove PMIC.
> >>>  
> >>>  config XPOWER_PMIC_OPREGION
> >>> -	bool "ACPI operation region support for XPower AXP288 PMIC"
> >>> +	tristate "ACPI operation region support for XPower AXP288 PMIC"
> >>>  	depends on AXP288_ADC
> >>>  	help
> >>>  	  This config adds ACPI operation region support for XPower AXP288 PMIC.
> >>> diff --git a/drivers/acpi/pmic/intel_pmic_crc.c b/drivers/acpi/pmic/intel_pmic_crc.c
> >>> index 8955e5b41195..8a193381b5ee 100644
> >>> --- a/drivers/acpi/pmic/intel_pmic_crc.c
> >>> +++ b/drivers/acpi/pmic/intel_pmic_crc.c
> >>> @@ -194,13 +194,23 @@ static int intel_crc_pmic_opregion_probe(struct platform_device *pdev)
> >>>  			&intel_crc_pmic_opregion_data);
> >>>  }
> >>>  
> >>> +#define DRV_NAME "crystal_cove_region"
> >>
> >> This name is just horrible, BTW.
> > 
> > Heh. I tried to follow the pattern in CrystalCove's MFD driver, where
> > all its cell devices are named crystal_cove_XXX, and since the platform
> > bus' match is done by comparing the device's name and the name field of
> > the driver's id table entry, they have to be the same.

So I called this one "crystal_cove_pmic" and the other one "axp288_pmic_acpi".

> The problem is,
> > the name field of the platform_device_id structure has a size limit of
> > 20, I can't even put the 'op' before the 'region' there. Perhaps time to
> > enlarge that size limit?
> 
> Or we can simply use 'acpi' instead of 'region' or 'opregion' here, to
> mean that this cell device is for ACPI purpose. How does this sound?

Better, but see above.

-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.