Re: Unstable Kernel behavior on an ARM based board

[Clemens Koller <clemens.ml@gmx.net>]

On 02/03/2019 13.39, Russell King - ARM Linux admin wrote:
> On Sat, Mar 02, 2019 at 05:20:53PM +0500, Embedded Engineer wrote:
>> On Sat, Mar 2, 2019 at 4:57 PM Russell King - ARM Linux admin
>> <linux@armlinux.org.uk> wrote:
>>>
>>> Okay, please try enabling DEBUG_LL, SERIAL_EARLYCON and select the
>>> correct serial port. Then arrange for u-boot to pass "earlycon" in
>>> addition to your other kernel arguments (shown in the "append" line
>>> above.)
>>
>> That's great :). I didn't make any change to kernel config as
>> DEBUG_LL, SERIAL_EARLYCON were already enabled. Just added 'earlycon'
>> to kernel arguments and boom, it booted without any crash. However
>> there were some errors mounting the rootfs. Please find below link to
>> complete boot logs:
>>
>> https://pastebin.com/7CfN6wLQ
> 
> I think you're lucky it got that far:
> 
> tegra-ehci 7d008000.usb: dma_pool_alloc ehci_qtd, f08de000 (corrupted)
> 00000000: 60 00 00 00 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7  `...............
> 00000010: a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7  ................
> 00000020: 40 02 00 00 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7  @...............
> 00000030: a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7  ................
> 00000040: 60 03 00 00 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7  `...............
> 00000050: a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7 a7  ................
> 
> There is a definite pattern to that corruption - first 32-bit word in
> every 32-bytes - I suspect the first 32-bit transfer into the cache
> ends up being corrupted.
> 
> IMHO, this points to RAM timing or wiring issues.  I don't know much
> about Tegra, but could it be a too-old u-boot (it reports that it's
> 2014, but apparently extlinux support was only merged in 2017...)?
> As you've said that these are a new board design, it could also be a
> design error with the RAM wiring.

The behaviour looks familiar to me if RAM timing is off. The memory initialization might not be suitable for your PCB design.
You can try to exercise some in-depth memory tests from within u-boot up to the temperature extremes of your hardware, before you switch over to linux. Checkout "mtest".

Regards,
Clemens

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel