rk3399-gru-kevin: issues on bringup

[Simon Glass]

Hi Marty,

On Mon, 3 Aug 2020 at 07:49, Simon Glass <sjg at chromium.org> wrote:
>
> Hi Marty,
>
> On Sun, 2 Aug 2020 at 21:02, Simon Glass <sjg at chromium.org> wrote:
> >
> > Hi Marty,
> >
> > On Fri, 31 Jul 2020 at 12:30, Simon Glass <sjg at chromium.org> wrote:
> > >
> > > Hi Marty,
> > >
> > > On Fri, 31 Jul 2020 at 05:19, Marty E. Plummer <hanetzer at startmail.com> wrote:
> > > >
> > > > On Tue, Jul 28, 2020 at 12:58:30PM -0600, Simon Glass wrote:
> > > > > Hi Marty,
> > > > >
> > > > > On Tue, 21 Jul 2020 at 21:07, Marty E. Plummer <hanetzer at startmail.com> wrote:
> > > > > >
> > > > > > On Tue, Jul 21, 2020 at 10:21:52AM -0600, Simon Glass wrote:
> > > > > > > Hi Marty,
> > > > > > >
> > > > > > > Did you check spl_boot_device()?
> > > > > > >
> > > > > > After sending the initial email I noticed your binman work, which does
> > > > > > some of the stuff I think I need. My current setup is as follows:
> > > > > >
> > > > > >
> > > > > > > Also take a look at the CONFIG_TARGET stuff in the code as it might
> > > > > > > speciy BOB but not KEVIN.
> > > > > > Yeah. I worked that in.
> > > > > >
> > > > > > Currently, a rom which is built with these changes (assuming u-boot.rom
> > > > > > is what I want for SPI booting; strange its only 4mb, aren't these
> > > > > > devices 8mb flash?) I get no output at all over the servo, aside from
> > > > > > the EC.
> > > > >
> > > > > I think it is only 4MB.
> > > > >
> > > > Nah, kevin is deffo 8mb flash chip. Otherwise I wouldn't have been able
> > > > to shove a 7.xmb kernel+initramfs into a coreboot image and flash it.
> > >
> > > Well that's odd. Maybe Kevin got a larger device than gru?
> > >
> > > >
> > > > I have to pad the u-boot.rom with dd to flash it.
> > > > > I am not sure that I have a kevin. Did you try using the debug UART?
> > > > >
> > > > Yeah, assuming you mean `dut-control cpu_uart_pty` with a servo hooked
> > > > up, and using `socat READLINE /dev/pts/something`. I get no output, but
> > > > the same chromiumos chroot with vanilla coreboot+depthcharge and
> > > > hardware config does do output as expected.
> > > >
> > > > Perhaps I'm missing something simple.
> > >
> > > Probably, it often is. But what?!
> > >
> > > I actually just found some old Chromebooks I didn't know i had, so I
> > > have a Kevin. If you push your tree somewhere I might be able to take
> > > a look this weekend.
> >
> > I dug this out but I cannot get it to boot with the em100 emulator.
> > I'll see if I can figure that out at some point. Sometimes I get
> > serial garbage as well.
> >
> > It is surprising that the flash is 8MB when Gru is 4MB. Must have run
> > out of room.
>
> Also I pushed the tree with your changes and one of mine here:
>
> https://github.com/sjg20/u-boot/tree/kevin

Well I don't know why the em100 doesn't work, but apparently it never did.

Anyway the image seems to jump into ATF first, then to U-Boot SPL. So
perhaps the problem is in ATF. From what I can tell, it prints about
10 characters of junk before it starts setting up SDRAM. Perhaps it
has a clock wrong.

In any case, it should be possible to jump directly into U-Boot SPL
and hopefully the debug UART works. Can you try that, perhaps by
building an image without ATF?

Regards,
SImon