Writing Linux kernel to MTD

[Ricard Wanderlof]

On Mon, 9 Feb 2009, Michal Ludvig wrote:

>> My instinctive reaction is that there may be a difference in the way the
>> Linux and u-boot drivers access the NAND device in terms of bus timing
>> etc, so that the Linux driver breaks the timing causing bad
>> communication with the device. I.e., the timing setup of the Linux
>> driver is at fault here.
>
> Interesting idea, however I happily use UBIFS filesystem at the same
> time in the same NAND Flash and observe no problems. If it's broken it's
> broken in a very reliable and stable way...

Sounds like it's not that then.

>> Another similar condition I have experienced is that due to a
>> misconfiguration one of the control bits to the NAND flash was
>> inadvertently mapped to a network indicator LED, causing NAND read/write
>> to fail if network traffic occurred at the same time.
>
> Just in case I disabled network altogether and did the write test from
> UBIFSroot instead of from development NFSroot and it's still failing.
>
> Any other ideas?

Well, chances that's a contention with the network led is probably small, 
but it could be some sort of other hardware interdependency.

/Ricard
--
Ricard Wolf Wanderlöf                           ricardw(at)axis.com
Axis Communications AB, Lund, Sweden            www.axis.com
Phone +46 46 272 2016                           Fax +46 46 13 61 30