El mié, 07-10-2020 a las 22:38 +0200, Hans de Goede escribió:
> Hi,
>
> On 10/7/20 9:58 PM, Limonciello, Mario wrote:
> > > -----Original Message-----
> > > From: Gerardo Esteban Malazdrewicz <gerardo@xxxxxxxxxxxxxxxxxxx>
> > > Sent: Wednesday, October 7, 2020 14:55
> > > To: Limonciello, Mario; Pali Rohár; Hans de Goede
> > > Subject: Re: [ PATCH v2 1/1] dell smbios driver : Consider
> > > Alienware a valid
> > > OEM String
> > >
> > >
> > > [EXTERNAL EMAIL]
> > >
> > > El mié, 07-10-2020 a las 15:53 +0000, Limonciello, Mario
> > > escribió:
> > > > > Hans, there are more drivers which checks for Dell DMI
> > > > > strings.
> > > > > Probably
> > > > > it would be needed to update Alienware on more places, not
> > > > > only in
> > > > > dell-smbios-base.c driver.
> > > >
> > > > I would prefer that each of those be checked on a case by case
> > > > basis
> > > > and only
> > > > added if actually necessary. Gerardo if you can please check
> > > > any
> > > > other drivers
> > > > that should need this string added to their allow list.
> > >
> > > I didn't find other instances of that string in this subsystem,
> > > but see
> > > below.
> > >
> > > There is one in pci, another in hotplug.
> > >
> > > However, this is an extract from kernel logs:
> > >
> > > [ 138.093686] dell-smbios A80593CE-A997-11DA-B012-B622A1EF5492:
> > > WMI
> > > SMBIOS userspace interface not supported(0), try upgrading to a
> > > newer
> > > BIOS
> >
> > Considering that messaging - does the non-WMI interface actually
> > work?
> > dell-smbios has two backends available.
>
> Yes that is a very good question.
>
> Gerardo, I guess you started looking into this because of the:
>
> pr_err("Unable to run on non-Dell system\n");
>
> In dell-smbios-base.c triggering on your system?
Correct, plus the top of dmidecode output, declaring SMBIOS present.
# dmidecode 3.2
Getting SMBIOS data from sysfs.
SMBIOS 3.2.0 present.
Table at 0x848B6000.
>
> As Pali mentioned in another mail, you probably should
> be looking at the dell-laptop code, which also has a
> has a DMI string check and which uses the dell-smbios code,
> another consumer of the dell-smbios code is the dell-wmi
> driver.
>
> If neither of those drivers add additional functionality
> (e.g. extra hotkey events, being able the control the kbd
> backlight), then the right fix might be to silence the
> error you see being thrown by dell-smbios-base.c, rather
> then allowing it to load.
>
dell-laptop doesn't even load as is ('No such device')
However, adding an entry in dell_device_table, based on dmidecode
output (Vendor: Alienware, Type: 10), allows it to load.
But I don't know how to test for any of this additional functionality.
What should I look for?
If any is found, would do a v3.
dell|alienware lsmod output, if that helps:
$ sudo lsmod | egrep "(dell|alienware)"
dell_laptop 28672 0
dell_rbtn 20480 0
dell_rbu 20480 0
dell_wmi 20480 0
dell_smo8800 20480 0
dell_smbios 32768 2 dell_wmi,dell_laptop
dell_smm_hwmon 24576 0
ledtrig_audio 16384 3
snd_hda_codec_generic,snd_sof,dell_laptop
dcdbas 20480 1 dell_smbios
alienware_wmi 20480 0
dell_wmi_descriptor 20480 2 dell_wmi,dell_smbios
rfkill 28672 11
bluetooth,dell_laptop,dell_rbtn,cfg80211
sparse_keymap 16384 2 intel_hid,dell_wmi
wmi 36864 7
intel_wmi_thunderbolt,alienware_wmi,dell_wmi,wmi_bmof,dell_smbios,dell_
wmi_descriptor,mxm_wmi
video 53248 3 dell_wmi,dell_laptop,i915
dell|alienware lsmod output, if that helps:
> For now I'll drop your patch from my review-hans branch,
> as we first need to clear this up.
>
> > The SMI based backend you can check by using dcdbas.
> >
> > I had presumed from your patch that it actually worked.
> >
> > > [ 1275.987716] dell_smm_hwmon: not running on a supported Dell
> > > system.
> > > [ 1275.987734] dell_smm_hwmon: vendor=Alienware, model=Alienware
> > > Area-
> > > 51m R2, version=1.3.0
> > >
> > >
> > > dell_smm_hwmon ignore_dmi=1
> > >
> > > /sys/class/hwmon/hwmonX/pwm{1,3} access correctly the left and
> > > right
> > > fan, respectively
>
> Ok, so that looks good and fixing the DMI check there probably
> makes sense. Note that this code is independent from the
> dell-smbios code from drivers/platform/x86 so you can
> do another patch to fix the DMI check there independent of the
> dell-smbios discussion.
>
> Regards,
>
> Hans
Regards, Gerardo
>
