[U-Boot] changes in rk3288 code have made me unable to boot u-boot on veyron speedy

[Heiko Stuebner]

Am Samstag, 9. März 2019, 23:45:15 CET schrieb Marty E. Plummer:
> On Sat, Mar 09, 2019 at 05:43:24PM +0100, Heiko Stuebner wrote:
> > Hi Marty,
> > 
> > Am Samstag, 9. März 2019, 08:15:23 CET schrieb Marty E. Plummer:
> > > Was going to work on getting that usb->uart redirection code from the
> > > linux kernel into u-boot for rk3288, like we have for rk3188, but
> > > apparently there have been some changes which render chromebook_speedy_defconfig
> > > unable to produce a bootable image. Guidance and suggestions welcome.
> > > 
> > > Current chromebook_speedy_defconfig results:
> > > U-Boot SPL 2019.04-rc3-03639-ge8e3f2d2d4 (Mar 09 2019 - 00:59:05 -0600)                                                  
> > > Trying to boot from SPI
> > > SPI probe failed.
> > > SPL: failed to boot from all boot devices
> > > ### ERROR ### Please RESET the board ###
> > > 
> > > chromebook_speedy_defconfig with CONFIG_SPI_FLASH turned on (didn't get
> > > moved into the defconfig like the rest)
> > > U-Boot SPL 2019.04-rc3-03639-ge8e3f2d2d4 (Mar 09 2019 - 01:01:38 -0600)                                                  
> > > Trying to boot from SPI
> > > initcall sequence 001511f4 failed at call 00101ad5 (err=-38)
> > > ### ERROR ### Please RESET the board ###
> > > 
> > > And enabling the full pinctrl driver and the needed libfdt stuff results
> > > in no output at all.
> > 
> > Maybe you could try the in flight patch from David first:
> > http://patchwork.ozlabs.org/patch/1040541/
> > 
> Are you meaning I should apply said patch to hopefully fix the no-output
> when i add the pinctrl driver?

exactly ... the completely new pinctrl driver seems to have a small error
on rk3288, that this patch is fixing ... supposedly a v2 is to come shortly.


> > In general I noticed in recent tries that rk3288 scrapes really narrow
> > at the 32kb limit of the sram, so possibly we'll need TPL on all rk3288
> > boards similar to what the rk3288-vyasa board already does now.
> > 
> Yeah I was eyeballing that as well since adding some features to speedy
> from jerry in an attempt to get around the issue made the spl image too
> large.

yep, everything is pretty narrow there. Right now I'm playing with getting
ATF to work[*] on rk3288 and this obviously needs real mmc and fit image
support and other boards like the phycore-rk3288 cannot even build an
spl image right now due to it needing more code to work.


Heiko

[*] ATF got armv7 support in 2017, so it's interesting if I can make it
handle smp via PSCI (and maybe things like deeper suspend)
and I learned so much about ARM assembler these last days :-D