Re: [PATCH] SPI LPC information kernel module

[Arnd Bergmann]

On Tue, Jun 30, 2020 at 3:57 PM Richard Hughes <hughsient@xxxxxxxxx> wrote:
>
> On Tue, 30 Jun 2020 at 09:56, Greg Kroah-Hartman
> <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> > Again, which makes it seem like securityfs is not the thing for this, as
> > it describes the hardware, not a security model which is what securityfs
> > has been for in the past, right?
>
> It describes the hardware platform. From a fwupd perspective I don't
> mind if the BC attributes are read from securityfs, sysfs or even read
> from an offset in a file from /proc... I guess sysfs makes sense if
> securityfs is defined for things like the LSM or lockdown status,
> although I also thought sysfs was for devices *in* the platform, not
> the platform itself. I guess exposing the platform registers in sysfs
> is no more weird than exposing things like the mei device and rfkill.

Why does fwupd care about the platform then? If these are
register values that relate to the flash device and that device is
what the firmware update gets written to, shouldn't it just use
an interface from that device?

       Arnd