Re: [PATCH v2 2/3] spi: Add HiSilicon v3xx SPI NOR flash controller driver

[John Garry]

On 31/01/2020 11:39, Andy Shevchenko wrote:
> On Fri, Jan 31, 2020 at 12:08 PM John Garry <john.garry@xxxxxxxxxx> wrote:
> > On 13/01/2020 14:34, Andy Shevchenko wrote:
> > > On Mon, Jan 13, 2020 at 02:27:54PM +0000, Mark Brown wrote:
> > > > On Mon, Jan 13, 2020 at 04:17:32PM +0200, Andy Shevchenko wrote:
> > > > > On Mon, Jan 13, 2020 at 4:07 PM Mark Brown <broonie@xxxxxxxxxx> wrote:
> > > > > > On Mon, Jan 13, 2020 at 01:01:06PM +0000, John Garry wrote:
> > > > > > > On 13/01/2020 11:42, Mark Brown wrote:
> > > >
> > > > > > > > The idiomatic approach appears to be for individual board vendors
> > > > > > > > to allocate IDs, you do end up with multiple IDs from multiple
> > > > > > > > vendors for the same thing.
> > > >
> > > > > > > But I am not sure how appropriate that same approach would be for some 3rd
> > > > > > > party memory part which we're simply wiring up on our board. Maybe it is.
> > > >
> > > > > > It seems to be quite common for Intel reference designs to assign
> > > > > > Intel IDs to non-Intel parts on the board (which is where I
> > > > > > became aware of this practice).
> > > >
> > > > > Basically vendor of component in question is responsible for ID, but
> > > > > it seems they simple don't care.
> > > >
> > > > AFAICT a lot of the time it seems to be that whoever is writing
> > > > the software ends up assigning an ID, that may not be the silicon
> > > > vendor.
> > >
> > > ...which is effectively abusing the ACPI ID allocation procedure.
> > >
> > > (And yes, Intel itself did it in the past â see badly created ACPI IDs
> > >    in the drivers)
> >
> > Hi Mark,

Hi Andy,

> > About this topic of ACPI having no method to describe device buswidth in
> > the resource descriptor, it may be an idea for me to raise a Tianocore
> > feature request @ https://bugzilla.tianocore.org/
>
> The 19.6.126 describes the SPI resource, in particular:
>
> ---8<---8<---
> DataBitLength is the size, in bits, of the smallest transfer unit for
> this connection. _LEN is automatically
> created to refer to this portion of the resource descriptor.
> ---8<---8<---
>
> Is it what you are looking for? (As far as I know most of the
> firmwares simple abuse this field among others)

I didn't think so - I thought that there was a distinction between width 
and length in SPI terms.

So how do you find that most firmwares abuse this field? AFAICS, linux 
kernel doesn't interpret this field at all.

> > There seems to be an avenue there for raising new features for the spec.
> > I (or my org) can't participate in AWSG.
>
> But have you read 19.6.126?

Maybe some clarification at least could be achieved :)

Cheers,
John