| Message ID | CAJcYhpbxWicEAOheSw9mdHibxQ2RPVQfZ4ncDvtQ5YEwtY=xWg@mail.gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 06/05/2017 09:25 AM, Paulo Sergio wrote: > Hi Pierre,how are you? > >>> +static int is_byt_cr(struct device *dev, struct sst_acpi_mach *mach, bool >>> *bytcr) >>> { >>> int status = 0; >>> + /* Lenovo Yoga2 exception - acpi_ipc_irq_index is the 1st index, >>> + but iosf_mbi_read (bios_status) returns 0 for bits 26:27 */ >>> + if (!strcmp(mach->drv_name, "bytcr_wm5102")) { >>> + *bytcr = true; >>> + return status; >>> + } >> >> well no, this isn't the right way to do this. quirks need to be based on >> real hardware information (e.g. DMI) and not a driver name. > Thanks for the review. Could I use dmi_match function inside is_byt_cr? Yes, however it's the first time I see a quirk based on DMI_CHASSIS_VERSION. we use PRODUCT or BOARD usually. > > git diff HEAD sound/soc/intel/atom/sst/sst_acpi.c > diff --git a/sound/soc/intel/atom/sst/sst_acpi.c > b/sound/soc/intel/atom/sst/sst_acpi.c > index dd250b8..a99d5a5 100644 > --- a/sound/soc/intel/atom/sst/sst_acpi.c > +++ b/sound/soc/intel/atom/sst/sst_acpi.c > @@ -244,6 +244,13 @@ static int is_byt_cr(struct device *dev, bool *bytcr) > { > int status = 0; > > + /* Lenovo Yoga2 exception - acpi_ipc_irq_index is the 1st index, > + but iosf_mbi_read (bios_status) returns 0 for bits 26:27 */ > + if (dmi_match(DMI_SYS_VENDOR, "LENOVO") && > dmi_match(DMI_CHASSIS_VERSION, "1051F")) { > + *bytcr = true; > + return status; > + } > + > if (IS_ENABLED(CONFIG_IOSF_MBI)) { > ... > > >> The rest looks more of less ok, it's not clear to me if all the quirks are >> needed and make sense. > Inside machine driver code, I can confirm these as necessary: > > - SSP0 usage > - 25Mhz Clock > > About the other quirks: To be honest, as I started using bycr_rt5640 > as base, I avoided touching other parts of the code. > > I'll understand if this is not adequated to be pushed upstream. > Besides, the codec side changes (arizona/wm5102 - not sent with this > patch) > are not ready to be pushed due some hardcoding & an issue with > regulator part (arizona-code will only detected codec chip if > arizona-ldo1 is marked as built-in; as a module it is not working) > > Regards > Paulo Sérgio Travaglia (Pstglia)
>>> well no, this isn't the right way to do this. quirks need to be based on >>> real hardware information (e.g. DMI) and not a driver name. >> >> Thanks for the review. Could I use dmi_match function inside is_byt_cr? > > > Yes, however it's the first time I see a quirk based on DMI_CHASSIS_VERSION. > we use PRODUCT or BOARD usually. Not a problem. I can change it. I have these info available from dmi: bios_date: 08/01/2014 bios_vendor: LENOVO bios_version: 01WT17WW board_asset_tag: NO Asset Tag board_name: INVALID board_serial: HA07YER9 board_vendor: LENOVO board_version: SDK9A6N2HVWIN chassis_asset_tag: NO Asset Tag chassis_serial: HA07YER9 chassis_type: 11 chassis_vendor: LENOVO chassis_version: 1051F modalias: dmi:bvnLENOVO:bvr01WT17WW:bd08/01/2014:svnLENOVO:pn60073:pvr1051F:rvnLENOVO:rnINVALID:rvrSDK9A6N2HVWIN:cvnLENOVO:ct11:cvr1051F: product_name: 60073 product_serial: HA07YER9 product_uuid: 39FFE49E-1918-884A-9499-CC7705A289C9 product_version: 1051F sys_vendor: LENOVO uevent: MODALIAS=dmi:bvnLENOVO:bvr01WT17WW:bd08/01/2014:svnLENOVO:pn60073:pvr1051F:rvnLENOVO:rnINVALID:rvrSDK9A6N2HVWIN:cvnLENOVO:ct11:cvr1051F:
diff --git a/sound/soc/intel/atom/sst/sst_acpi.c b/sound/soc/intel/atom/sst/sst_acpi.c index dd250b8..a99d5a5 100644 --- a/sound/soc/intel/atom/sst/sst_acpi.c +++ b/sound/soc/intel/atom/sst/sst_acpi.c @@ -244,6 +244,13 @@ static int is_byt_cr(struct device *dev, bool *bytcr) { int status = 0; + /* Lenovo Yoga2 exception - acpi_ipc_irq_index is the 1st index, + but iosf_mbi_read (bios_status) returns 0 for bits 26:27 */ + if (dmi_match(DMI_SYS_VENDOR, "LENOVO") && dmi_match(DMI_CHASSIS_VERSION, "1051F")) { + *bytcr = true; + return status; + } + if (IS_ENABLED(CONFIG_IOSF_MBI)) { ...