[U-Boot] u-boot for Snow problem

[Michal Suchanek]

On 15 April 2015 at 17:00, Simon Glass <sjg at chromium.org> wrote:
> Hi Michal,
>
> On 14 April 2015 at 16:10, Michal Suchanek <hramrach at gmail.com> wrote:
>> Hello,
>>
>> On 4 March 2015 at 00:46, Simon Glass <sjg at chromium.org> wrote:
>>> Hi Michal,
>>>
>>> On 2 March 2015 at 04:25, Michal Suchanek <hramrach at gmail.com> wrote:
>>>> Hello,
>>>>
>>>> On 18 February 2015 at 06:24, Michal Suchanek <hramrach at gmail.com> wrote:
>>>>> On 18 February 2015 at 03:27, Simon Glass <sjg at chromium.org> wrote:
>>>>>> Hi Michal,
>>>>>>
>>>>>> On 16 February 2015 at 04:41, Michal Suchanek <hramrach at gmail.com> wrote:
>>>>>>> On 13 February 2015 at 05:51, Simon Glass <sjg at chromium.org> wrote:
>>>>>>>> Hi Michal,
>>>>>>>>
>>>>>>>> On 11 February 2015 at 10:16, Michal Suchanek <hramrach at gmail.com> wrote:
>>>>>>>>>
>>>>>>>>> Hello,
>>>>>>>>>
>>>>>>>>> I changed the SYS_START to work around the bug in the manufacturer
>>>>>>>>> firmware, applied snow_defconfig, built u-boot.bin, packed it into
>>>>>>>>> kernel uimage, signed it, copied it to a kernel partition, bumped
>>>>>>>>> priority of the partition, and rebooted.
>>>>>>>>>
>>>>>>>>
>>>>>>>> Do you mean u-boot-dtb.bin? If not you won't get a device tree and it
>>>>>>>> won't work.
>>>>>>>
>>>>>>> No, u-boot.bin. With u-boot-dtb.bin I get a snow # prompt on the
>>>>>>> built-in LCD, and working keyboard.
>>>>>>
>>>>>> OK sounds like it is working, good! I wonder if we should have a page
>>>>>> on elinux.org?
>>>>
>>>> It is working to some extent.
>>>>
>>>> I managed to load kernel from the emmc which works fine but the kernel
>>>> cannot read the emmc after it boots because it does not properly parse
>>>> the partitioning scheme. This should be trivially fixable in the
>>>> kernel and might actually work if I updated my sources but rebasing
>>>> the extra patches required for Snow is not automatically handled.
>>>>
>>>> On the other hand, the linux kernel has no problem with the SDXC card
>>>> in the SD slot and can read it just fine. Unfortunately, u-boot
>>>> complains about EFI partition errors and won't load anything from the
>>>> card. I tried two different GPT partitioning tools on the card and
>>>> both say that the partition layout is fine and that I have the default
>>>> 128 entries.
>>>>
>>>> How can I tell why u-boot does not like my GPT label?
>>>
>>> You could debug it in U-Boot and see what is going wrong.
>>>
>>
>> Apparently the problem is that the dw mmc controller fails and the
>> sectors are unreadable. U-boot sometimes reports timeout, sometimes
>> data error, and sometimes nothing because the result has already been
>> cached probably. And the reason the dw mmc controller fails is that
>> port 1 is disabled. Not using disabled controllers would probably
>> solve the problem.
>
> Seems strange. Can you post details of which commit you are building
> and the console output? I could try it too.


Sorry, I don't have the tree anymore.

None of my trees has the commit to relocate away from the address
clobbered by the Snow ro firmware.

I tried looking again what actually happens but none of the mmc
commands shows which exact mmc interfaces I have. U-boot somehow picks
2 of the 4 dwmmc interfaces the Exynos chip has and I suspect the
second one is picked or multiplexed wrong. The mmc0 to which emmc is
connected works fine but mmc2 connected to the external slot always
fails (on Snow mmc1 is unused and mmc3 is WiFi).

I tried building u-boot master but none of the boot commands I have in
the currently installed unknown u-boot snapshot likes the
u-boot-dtb.bin image and I definitely do not want to replace the
semi-working snapshot I have with non-relocated u-boot master.

Thanks

Michal