Re: [PATCH 5/6] soc: aspeed: Add eSPI driver

[Joel Stanley]

On Thu, 7 Jan 2021 at 02:39, ChiaWei Wang <chiawei_wang@xxxxxxxxxxxxxx> wrote:
>
> Hi Rob,
>
> > -----Original Message-----
> > From: Rob Herring <robh@xxxxxxxxxx>
> > Sent: Wednesday, January 6, 2021 11:32 PM
> > To: ChiaWei Wang <chiawei_wang@xxxxxxxxxxxxxx>
> > Subject: Re: [PATCH 5/6] soc: aspeed: Add eSPI driver
> >
> > On Wed, Jan 06, 2021 at 01:59:38PM +0800, Chia-Wei, Wang wrote:
> > > The Aspeed eSPI controller is slave device to communicate with the
> > > master through the Enhanced Serial Peripheral Interface (eSPI).
> > > All of the four eSPI channels, namely peripheral, virtual wire,
> > > out-of-band, and flash are supported.
> > >
> > > Signed-off-by: Chia-Wei, Wang <chiawei_wang@xxxxxxxxxxxxxx>
> > > ---
> > >  drivers/soc/aspeed/Kconfig                  |  49 ++
> > >  drivers/soc/aspeed/Makefile                 |   5 +
> > >  drivers/soc/aspeed/aspeed-espi-ctrl.c       | 197 ++++++
> > >  drivers/soc/aspeed/aspeed-espi-flash.c      | 490 ++++++++++++++
> > >  drivers/soc/aspeed/aspeed-espi-oob.c        | 706
> > ++++++++++++++++++++
> > >  drivers/soc/aspeed/aspeed-espi-peripheral.c | 613 +++++++++++++++++
> > >  drivers/soc/aspeed/aspeed-espi-vw.c         | 211 ++++++
> > >  include/uapi/linux/aspeed-espi.h            | 160 +++++
> > >  8 files changed, 2431 insertions(+)
> > >  create mode 100644 drivers/soc/aspeed/aspeed-espi-ctrl.c
> > >  create mode 100644 drivers/soc/aspeed/aspeed-espi-flash.c
> > >  create mode 100644 drivers/soc/aspeed/aspeed-espi-oob.c
> > >  create mode 100644 drivers/soc/aspeed/aspeed-espi-peripheral.c
> > >  create mode 100644 drivers/soc/aspeed/aspeed-espi-vw.c
> >
> > drivers/spi/ is the correct location for a SPI controller.
> >
> > >  create mode 100644 include/uapi/linux/aspeed-espi.h
> >
> > This userspace interface is not going to be accepted upstream.
> >
> > I'd suggest you look at similar SPI flash capable SPI controller drivers upstream
> > and model your driver after them. This looks like it needs major reworking.
> >
> eSPI resues the timing and electrical specification of SPI but runs completely different protocol.
> Only the flash channel is related to SPI and the other 3 channels are for EC/BMC/SIO.
> Therefore, an eSPI driver might not fit into the SPI model.

I agree, the naming is confusing but eSPI doesn't belong in drivers/spi.

As it is a bus that is common to more than just the Aspeed BMC, we may
want to implement it as a new bus type that has devices hanging off
it, similar to FSI.

Cheers,

Joel