[PATCH 6/8] scsi: Drop CONFIG_SYS_SCSI_MAX_DEVICE

[Simon Glass]

Hi,

On Mon, 31 Jan 2022 at 09:04, Tom Rini <trini at konsulko.com> wrote:
>
> On Mon, Jan 31, 2022 at 04:58:46PM +0100, Heinrich Schuchardt wrote:
> > On 1/31/22 15:49, Simon Glass wrote:
> > > This is defined based on two other CONFIGs for all boards except sandbox
> > > and durian.
> > >
> > > For sandbox the value does not matter. For durian the value seems
> > > excessive.
> > >
> > > Drop the option completely, to simplify configuration and reduce the
> > > number of things we need to convert to Kconfig.
> > >
> > > Signed-off-by: Simon Glass <sjg at chromium.org>
> [snip]
> > > diff --git a/include/configs/helios4.h b/include/configs/helios4.h
> > > index de1ebbf3751..c1d42f17ca6 100644
> > > --- a/include/configs/helios4.h
> > > +++ b/include/configs/helios4.h
> > > @@ -40,8 +40,6 @@
> > >   #define CONFIG_SCSI_AHCI_PLAT
> > >   #define CONFIG_SYS_SCSI_MAX_SCSI_ID       1
> >
> > This should be Kconfig too. 1 seems to be a reasonable default value as it
> > covers most boards. The minimum should be enforced as 1 in Kconfig.
> >
> > >   #define CONFIG_SYS_SCSI_MAX_LUN           1
> >
> > Make this Kconfig, please. Let it default to 1 and enforce a minimum of 1
> > and a maximum of 256 in Kconfig.
>
> As the series title notes this is "some" options, so yes, a good
> follow-up would be to cover more of them.  I believe the end goal here
> is that we can drop dead code and more easily check/complain about
> unmigrated boards.

Yes.

Also I don't think any of these options should survive, since with
driver model we should a more dynamic approach, configured with device
tree as needed. See ahci_probe_scsi() where it sets the max_lun to a
fixed value at present, and max_id which is calculated after probing.

We just need someone to take on completion of the AHCI/SCSI/SATA
migration. The code could really use a look.

Regards,
Simon