Hi, On 10/18/21 12:31, Andy Shevchenko wrote: > On Mon, Oct 18, 2021 at 12:16 PM Hans de Goede <hdegoede@xxxxxxxxxx> wrote: >> On 10/17/21 18:15, Tsuchiya Yuto wrote: >>> On Microsoft Surface 3 (uses Intel's Atom Cherry Trail SoC), executing > > ... > >> As Andy said we could use a DMI quirk for this, but chances are that the Microsoft Surface >> DSDT is not the only one with the wrong HRV value. So instead it might be better to >> just test for the SoC type as the attached patch does. >> >> Tsuchiya, can you give the attached patch a try. >> >> Andy, what do you think, should we go with the attached patch or would you prefer using >> a DMI quirk ? > > TBH I have no strong opinion. Only one remark on your patch, I am not > a fan of removing COMPILE_TEST but at the same time I'm not a fan of > ifdeffery. All on all I think having COMPILE_TEST is preferable even > if we have ifdeffery. Btw, IIRC similar code (i.e. BYT vs CHT by CPU > ID) is being used elsewhere. Perhaps we might have some common > (library) under arc/x86, PDx86 or so (headers?)? We already have helpers for this defined in: sound/soc/intel/common/soc-intel-quirks.h We could move those to some header under include, maybe: include/linux/platform_data/x86/atom.h And add #ifdef-ery there so that things will also build on non x86 ? Then we could do a 2 patch series adding the include/linux/platform_data/x86/atom.h file + the drivers/mfd/intel_soc_pmic_core.c change and Lee can merge both through the MFD tree. And then we can do further clean-ups of e.g. sound/soc on top (we can ask Lee to provide an immutable branch). How does that sound ? Regards, Hans
