> On 10/7/20 11:26 PM, Limonciello, Mario wrote: > >> 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? > > > > If that's the case, I would ask why this driver even auto-loaded? > > The module load table is very prescriptive. > > https://github.com/torvalds/linux/blob/master/drivers/platform/x86/dell- > smbios-wmi.c#L277 > > https://github.com/torvalds/linux/blob/master/drivers/platform/x86/dell- > smbios-smm.c#L56 > > Was it because it was compiled into the kernel? > > I believe that the laptops ACPI tables do define the WMI GUID > that dell-smbios-wmi.c has in its module-id-table otherwise > the "WMI SMBIOS userspace interface not supported(0)" error > would not trigger. > That's a good point, you're right thanks. I should note - that is a warning, it's valid and it only turns off userspace access. If the BIOS doesn't advertise that hotfix many calls can fail. Some of the more simple calls used by the kernel modules such as dell-laptop would not be prohibited.