[U-Boot] arm: sunxi: Bananapi_M2_Ultra not working with DM_MMC

[Jagan Teki]

On Tue, Apr 9, 2019 at 12:43 AM Tom Rini <trini at konsulko.com> wrote:
>
> On Tue, Apr 09, 2019 at 12:41:48AM +0530, Jagan Teki wrote:
> > On Tue, Apr 9, 2019 at 12:39 AM Tom Rini <trini at konsulko.com> wrote:
> > >
> > > On Tue, Apr 09, 2019 at 12:37:04AM +0530, Jagan Teki wrote:
> > > > Hi Tom,
> > > >
> > > > On Tue, Apr 9, 2019 at 12:19 AM Tom Rini <trini at konsulko.com> wrote:
> > > > >
> > > > > On Tue, Apr 09, 2019 at 12:17:15AM +0530, Jagan Teki wrote:
> > > > > > On Tue, Apr 9, 2019 at 12:02 AM Michael Nazzareno Trimarchi
> > > > > > <michael at amarulasolutions.com> wrote:
> > > > > > >
> > > > > > > Hi jagan
> > > > > > >
> > > > > > >
> > > > > > > On Mon., 8 Apr. 2019, 8:26 pm Tom Rini, <trini at konsulko.com> wrote:
> > > > > > >>
> > > > > > >> On Mon, Apr 08, 2019 at 06:23:29PM +0530, Jagan Teki wrote:
> > > > > > >> > Hi Paul,
> > > > > > >> >
> > > > > > >> > On Mon, Apr 8, 2019 at 6:00 PM Paul Kocialkowski
> > > > > > >> > <paul.kocialkowski at bootlin.com> wrote:
> > > > > > >> > >
> > > > > > >> > > Hi,
> > > > > > >> > >
> > > > > > >> > > On Thu, 2019-04-04 at 05:51 -0300, Pablo Sebastián Greco wrote:
> > > > > > >> > > > A few days ago I tried to boot my Bananapi_M2_Ultra with 2019.04rc, I
> > > > > > >> > > > found that it wasn't booting, 2019.01 was working ok.
> > > > > > >> > > > Bisecting indicated that the problem was after
> > > > > > >> > > > http://git.denx.de/?p=u-boot.git;a=commitdiff;h=a7cca5793774ee139b75a704d6efaa4d29f09f93
> > > > > > >> > >
> > > > > > >> > > I think the patch should be reverted ASAP since it obviously breaks
> > > > > > >> > > some supported configs. Sadly, the offending commit doesn't say
> > > > > > >> > > anything about the test coverage for the change and what the status is
> > > > > > >> > > after it. There is probably a reason why it was enabled for sun4i only
> > > > > > >> > > before and there must have been a motivation for doing this on all
> > > > > > >> > > sunxi platforms, but then again, the commit message says nothing about
> > > > > > >> > > those underlying reasons.
> > > > > > >> > >
> > > > > > >> > > I believe we should be more strict on patch review and not let any
> > > > > > >> > > change bringing such a major change get applied with a commit message
> > > > > > >> > > that provides no context about why the change is okay and how it was
> > > > > > >> > > tested.
> > > > > > >> >
> > > > > > >> > Appropriate your concern.
> > > > > > >> >
> > > > > > >> > If you please list what all boards are not working with this effect,
> > > > > > >> > please write back. we will defiantly look into it. All these changes
> > > > > > >> > were merged in MW which is 2.5 months back, commenting in final stage
> > > > > > >> > like this is not the professional way.
> > > > > > >>
> > > > > > >> Being release day, here's my big concern.  How bad is this?  Is it a
> > > > > > >> single platform?  Later in the thread Jagan did enumerate the SoCs he
> > > > > > >> tested the overall series on.  But there's a lot of Allwinner SoCs and
> > > > > > >> boards.  I have a pine64 somewhere around here, but that's already been
> > > > > > >> checked off.  My other allwinner platform I took out of my testing loop
> > > > > > >> due to it not being a reliable piece of hardware.  So, does anyone have
> > > > > > >> a feel for how many platforms may or may not be broken right now?
> > > > > > >> Thanks!
> > > > > > >
> > > > > > >
> > > > > > > You have 13 to 15 boards. Can you just report all of them?
> > > > > >
> > > > > > Issue, seems to be on SCSI side. MMC is able to probe and boot (you
> > > > > > may see initial logs on the thread). Enabling DM_MMC breaking SCSI
> > > > > > reads, debugging same with Pablo will get back.
> > > > >
> > > > > OK, thanks, please keep us up to date!
> > > >
> > > > As expected, SCSI reads failing. DM_MMC is forcing BLK so which
> > > > expecting boards to enable DM_SCSI. So the migration plan for DM_SCSI
> > > > and DM_MMC should be in same line, as of now it's not.
> > >
> > > Ah, is the scsi driver for Allwinner DM-ified yet?  And for the release
> > > I'd really like to do today can we just switch off DM_MMC on the DM_SCSI
> > > Allwinner platforms?
> >
> > Since the migration plan still there for DM_SCSI (v2019.07) we haven't
> > enabled DM_SCSI yet.
>
> Except on sun4i.  But different driver then?  But I guess for today we
> need to use:
>         select DM_MMC if MMC && !(DM_SCSI && SCSI)
> ?

Yes, A10 has DM_SCSI


Above solutions seems to be 'recursive dependency' issue.

drivers/scsi/Kconfig:11:error: recursive dependency detected!
drivers/scsi/Kconfig:11:    symbol DM_SCSI depends on BLK
drivers/block/Kconfig:4:    symbol BLK depends on DM_MMC
drivers/mmc/Kconfig:26:    symbol DM_MMC is selected by DM_SCSI
For a resolution refer to Documentation/kbuild/kconfig-language.txt
subsection "Kconfig recursive dependency limitations"