[U-Boot] WaRP7 nok on master

[Auer, Lukas]

Hi Stefano,

On Thu, 2019-05-02 at 10:55 +0200, Stefano Babic wrote:
> Hi Piere, Lukasz,
> 
> On 01/05/19 20:49, Pierre-Jean Texier wrote:
> > Hi Fabio, Stefano,
> > 
> > Just FYI, I just tested the U-Boot master branch (with u-boot-imx merges).
> > And I have some problems when the WaRP7 boot-up.
> > In fact, no output...
> > 
> > However, on u-boot-imx, everything works well (tags/u-boot-imx-20190426).
> > 
> > After some manipulation with git bisect, it appears that the problem
> > comes from commit [1].
> > So, with a git revert 3a7c45f6a7725808e2e82908be4bc90d4d78e737,
> > everything works again.
> > 
> > Without this revert, the WaRP7 is not
> > functional (also the pico-pi i.MX7 if DM_SERIAL is implemented, tested
> > by Joris in CC)
> 
> This is painful - Lukasz, can you comment this ? Your patch is small and
> it seems to fix just qemu, it is difficult to understand why it has side
> effects on real SOCs like i.MX.
> 

I was able to reproduce the issue on my side. With the patch, U-Boot
probes the drivers for devices under simple-bus device tree nodes in
the pre-relocation device model. The default value of
CONFIG_SYS_MALLOC_LEN (0x400) leaves U-Boot with not enough memory to
do this, causing it to hang. If it is increased, for example to 0x1000,
everything works again.

Let me know, how you want to fix this. If you want, I can send a patch
to increase CONFIG_SYS_MALLOC_LEN for the i.MX parts.

Thanks,
Lukas

> > 
> > [1]
> > https://github.com/trini/u-boot/commit/3a7c45f6a7725808e2e82908be4bc90d4d78e737
> >