On Mon, 11 May 2020 at 11:45, Mika Westerberg
<mika.westerberg@linux.intel.com> wrote:
> I think you may want to look at drivers/mfd/lpc_ich.c and see if some of
> this can be placed there. It is the "LPC" driver that binds to the
> LPC/eSPI PCI device so it already has at least some of these PCI IDs.

Ahh, that's useful, thanks. I've attached something based on lpc_ich
that works, although there are a few things needing discussion:

* Some of the reported eSPI IDs are missing, and I can't easily find
references to them in the official Intel specifications. Do you need
me to hunt them down, or can I add DIDs without referencing a document
number? I can certainly split out the new DIDs and the securityfs
stuff when this patchset looks half-acceptable.

* Do you want the CONFIG_SECURITY functionality put in a different
file, perhaps with a different Kconfig entry? e.g. LPC_SCH_SECURITYFS
-- if so, how do you want the hooks implemented? Declare a dummy
lpc_ich_init_securityfs() if there is no support for securityfs like
iommu does? Put the securityfs dentries static in this new file rather
than in the lpc_ich_priv struct? Any advice welcome.

* My hardware seems to not set RCBA and so res->start never gets
defined.I don't think it's a problem from my naive point of view, but
the -ENODEV is presumably there for a reason.

If you want the patch inline, that's fine too, please just ask and I
can resend. Thanks for your help so far, and sorry for all the silly
mistakes -- I'm new to all this kernel stuff.

Richard.