[U-Boot] [BUG]Ｒｏckchip armv7 SoCs can not boot on latest U-Boot

[Tom Rini]

On Tue, May 23, 2017 at 04:43:56PM +0800, Kever Yang wrote:
> Tom, Simon,
> 
> 
> On 05/23/2017 04:38 AM, Tom Rini wrote:
> >On Mon, May 22, 2017 at 02:26:21PM -0600, Simon Glass wrote:
> >>+Tom
> >>
> >>Hi Kever,
> >>
> >>On 17 May 2017 at 03:56, Kever Yang <kever.yang at rock-chips.com> wrote:
> >>>Hi Bin,
> >>>
> >>>Thanks for your suggestion, I already find the root cause.
> >>>
> >>>This patch change the image package and lead to all Rockchip SoCs
> >>>
> >>>with BACK_TO_BROM enabled can not work.
> >>>
> >>>https://www.mail-archive.com/u-boot@lists.denx.de/msg245567.html
> >>>
> >>>Is it possible to do verify commits for different platform like kernelci
> >>>for
> >>>kernel patches in U-Boot?
> >>We do use buildman to verify that things build, and that tests pass on
> >>sandbox. But specific boards can still become broken.
> >>
> >>Tom has something like this set up and uses it to check the boards he
> >>has. I encourage you to set up some boards at your facility and do
> >>this!
> >Indeed, it's time for another reminder about both tbot (which Heiko
> >runs and instance of and maintains) and test/py/test.py +
> >https://github.com/swarren/uboot-test-hooks which is what we use to
> >control qemu targets in travis-ci and real hardware in various labs
> >(Stephen has one for tegra and I know there's a few other private
> >setups) including mine.  I don't however have any Rockchip targets, but
> >suggestions of a board that would be a good fit for a lab are welcome
> >(ie easy console and recovery from bad builds).
> 
> I can try to setup one although I have no any experience on this,
> is there a step by step document I can learn from?

Not exactly, but there's READMEs for it all.

> For kernel, it can use net boot with TFTP, but for U-Boot/SPL, we need
> to flash the eMMC or SD card, which make it more complicate and I don't
> have a solution for automatic test.

Is there some recovery method via USB or similar?  Does the ROM support
reading from say FAT?  Having to raw-write to an SD card is still a
tricky operation, in a recoverable way.

> I may need to verify for different SoCs, for example, I found that the
> rk3036 build also been broken, the SPL size is too large now(bigger than
> 4KB), I still don't have time to figure it out.

wrt sizes, you need to start using the existing linker script hooks that
make this fail to link :)

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20170523/5dbe4650/attachment.sig>