[U-Boot] SoCFPGA PL330 DMA driver and ECC scrubbing

[Marek Vasut]

On 07/11/2018 03:56 PM, Jason Rush wrote:
> On 7/11/2018 8:48 AM, Marek Vasut wrote:
>> On 07/11/2018 03:49 PM, Jason Rush wrote:
>>> On 7/11/2018 3:55 AM, Marek Vasut wrote:
>>>> On 07/11/2018 05:11 AM, Jason Rush wrote:
>>>> [...]
>>>>
>>>>>>>>> However, if I press the HPS_RST push button on the SoCKit (which is connected
>>>>>>>>> to power on reset), occasionally U-Boot will lock up while booting.  It always
>>>>>>>>> boots and operates correctly from the initial power on, but it almost always
>>>>>>>>> fails to boot after pressing the HPS_RST button.
>>>>>>>>>
>>>>>>>>> Usually after pressing the HPS_RST button, U-Boot makes it past the SPL, and
>>>>>>>>> hangs somewhere after the call to setup_reloc() in ./common/board_f.c.  Once
>>>>>>>>> it hangs there, pressing the HPS_RST button again usually causes the SPL to
>>>>>>>>> hang while setting up the MMU (before my call to memset).  Eventually the
>>>>>>>>> WDT kicks in, and it just keeps hanging up in the same place.  Once it gets in
>>>>>>>>> this mode, the only way to recover it is by toggling power on the board.
>>>>>>>>>
>>>>>>>>> I spent a bunch of time today trying to track down where it was hanging, but
>>>>>>>>> I couldn't pin point anything.  The MMU tables looked correct.  The MMU
>>>>>>>>> registers looked good.  I'm not sure the best way to debug what's going on.
>>>>>>>> Try triggering warm reset and cold reset via the reset register:
>>>>>>>>
>>>>>>>> mw 0xffd05004 1
>>>>>>>> mw 0xffd05004 2
>>>>>>>>
>>>>>>>> Does it hang in one case and not in the other ?
>>>>>>>>
>>>>>>> It hangs in both cases.
>>>>>>>
>>>>>>> I did find that if I do not metset the last 1MiB of DRAM with the cache on,
>>>>>>> both warm and cold resets work.
>>>>>>>
>>>>>>> I changed the ecc scrubbing to zero out the first 0x8000 bytes and the last
>>>>>>> 0x10000 bytes before the MMU is setup and I enable dcache.  Then with
>>>>>>> the dcache enabled, I zero out the rest of memory.  The resets work in this
>>>>>>> case as well.  So there seems to be some side effect of clearing out the
>>>>>>> relocate address space with the cache on.
>>>>>> Can you investigate ?
>>>>>>
>>>>> I'd be happy to investigate more, but I'm not really sure what
>>>>> my next step should be.
>>>>>
>>>>> Something appears to be happening differently when U-Boot
>>>>> relocates if the dcache is on.  But don't know how to track it
>>>>> down.
>>>> IIRC I disabled cache after scrubbing.
>>> My mistake.  I did disable the dcache after scrubbing too.  The
>>> code is almost identical to the Arria10 code where after
>>> scrubbing it flushes the dcache, then turns it off.
>>>
>>> The weird reset problems happens if I scrub the area where
>>> u-boot relocates to with the dcache on, then turn dcache off.
>>>
>>> I tried to also tried turning the MMU off, but that didn't help.
>> Maybe there are some data used by the SPL there ? I think the SPL has
>> malloc area in RAM at some point.
>>
> I thought something similar, so I narrowed it down to clearing
> just from where U-Boot relocates to the end of DRAM.  If I'm
> correct, that includes where U-Boot relocates and where the
> MMU tables are normally stored.

Hm, although, I think if you do the scrubbing right after the DRAM
controller gets inited, there isn't anything in the DRAM yet. You might
need to keep digging, since this is odd.

-- 
Best regards,
Marek Vasut