[U-Boot] [PATCH 0/5] db410c: updates for grub + gfxterm

[Rob Clark]

On Wed, Jul 19, 2017 at 12:49 PM, Rob Clark <robdclark at gmail.com> wrote:
> On Thu, Jul 6, 2017 at 8:02 AM, Mateusz Kulikowski
> <mateusz.kulikowski at gmail.com> wrote:
>> Hi Rob,
>>
>> On 25.06.2017 01:05, Rob Clark wrote:
>>> In particular, support for display setup by lk.  This introduces a
>>> simplefb display driver that uses the framebuffer fdt node populated
>>> by the firmware[1] for u-boot display support (and, at least for
>>> what I am working on, more interestingly, EFI GOP support).
>> [...]
>>
>>> A few related patches so that on db410c we actually use the fdt
>>> passed by lk (and so that dm/core is clever enough to notice fdt
>>> nodes under "chosen"), config updates, and related fixes.
>> [...]
>>
>> I finally had time to look at patches (this and the earlier series).
>>
>> Enabling the output is pretty cool - I thought of that myself some time ago.
>>
>> Question: Do you have non-standard partition layout?
>> I have flashed most recent 96boards debian image, but my device doesn't
>> have splash partition. (It initializes screen properly, just doesn't display anything)
>
> yes, I have a modified version of bootloader package from linaro:
>
> https://people.freedesktop.org/~robclark/tmp/dragonboard410c_bootloader_emmc_linux-79-u-boot.tgz
>
> I split the boot partition (which was originally big enough for
> kernel+initrd which is somewhat larger than u-boot.img) to make a
> splash partition.  I probably should write up instructions for how to
> do that and generate the splash.img (but it isn't too hard to find
> with a bit of googling)..
>
> It isn't the latest lk + u-boot, on the lk side I have some fixes I
> still need to push to make it deal properly with #address/size-cells
> and a few other misc things.  (And on the u-boot side, adding a
> reserved-memory node in the dragonboard410c.dts)..
>
>> Question2: Are you sure you have included all the patches?
>>
>> I've applied this series (also included patches from 2017-06-20), but u-boot
>> doesn't parse device tree properly and things like usb cease to work -
>> both with your and official lk:
>>
>> <log>
>>
>> U-Boot 2017.07-rc3-00015-ga2592ee (Jul 06 2017 - 13:48:40 +0200)
>> Qualcomm-DragonBoard 410C
>>
>> DRAM:  986 MiB
>> MMC:   sdhci at 07824000: 0, sdhci at 07864000: 1
>> Using default environment
>>
>> In:    serial
>> Out:   serial
>> Err:   serial
>> Failed to find PMIC pon node. Check device tree
>> Net:   Net Initialization Skipped
>> No ethernet found.
>> Hit any key to stop autoboot:  0
>>
>> </log>
>>
>>
>> On the other hand it works properly with your pre-built binary [1].
>>
>> I assume the difference may be that you have added proper (instead of fake) device
>> tree into mkbootimg - is that the case?
>>
>
> I've (mostly) been using the dragonboard410c.dtb from u-boot build,
> since there seem to be some small problems in the other direction..
> u-boot isn't completely compatible with upstream linux dtb.
>
> I did find a few things that were only working after I shifted to
> CONFIG_OF_BOARD because I hadn't done a clean build (ie. still picking
> up old .dtb)
>
> I'm working on cleaning up things, and I'll try to write up some
> better instructions and push things to github branch this afternoon,
> to make it a bit more reproducible
>

jfyi, I've pushed latest of what I'm using (both with lk ->
kernel(simplefb) and with lk -> u-boot -> grub -> kernel(efifb) here:

  https://github.com/robclark/lk/commits/db410c-display
  https://github.com/robclark/u-boot/commits/wip

BR,
-R