Booting on AR9331 based Easylink EL-M150

[Adrian Chadd]

Yeah, there's calibration data in the flash chip. You should be very careful. :)



-a


On 11 January 2015 at 12:17, Bernd Walter <ticso at cicely7.cicely.de> wrote:
> On Sun, Jan 11, 2015 at 10:16:49AM -0800, Adrian Chadd wrote:
>> Hi,
>>
>> I've had this issue reported by someone else - a tplink AR934x based
>> thing. I wonder if it's an updated uboot related thing.
>>
>> So the TL;DR is this.
>>
>> uboot "bootm" doesn't boot a kernel image - it boots a uboot
>> application. The application has a uboot header with details about
>> "where it should be unpacked", "where it should be executed", etc. The
>> tool that we're using is in ports but from an older uboot. It's
>> possible the header contents are now .. wrong.
>>
>> So, the flash address used by the "bootm" comand is 0x9f(blah), and
>> it'll unpack to wherever it says in the uboot header, and then execute
>> wherever it says. Everything above is wrong - Load, entry, size, etc -
>> and so that needs to be debugged.
>>
>> I have a tplink in a box that has the same issue, so I'll go see what
>> the story is.
>
> Great.
>
> My EL-M150 module has the same hynx memory chip as the carambola 2 I compared
> against, just the flash size ist different (winbond 25Q128 vs 25Q64).
> Don't know if the SPI flash is hardcoded somehow, otherwise the carambola 2
> u-boot should work IMHO.
>
> Is there anything important in the flash, like calibration data, which
> can be trashed by tampering with it?
> Or failed parsing by a different u-boot version?
> Guess I should dump the SPI data to tftp server anyway.
> Never programmed a 25Q myself, but can't be too hard in case I brick it.
>
> --
> B.Walter <bernd at bwct.de> http://www.bwct.de
> Modbus/TCP Ethernet I/O Baugruppen, ARM basierte FreeBSD Rechner uvm.