Bit of follow up for this issue.

By tracking down the function that caused the crash, it was in 
'crashlog_init_fs'.

This function uses and 'address' that seem to be initialized if the config 
symbol:

CONFIG_NO_BOOTMEM

is NOT defined...

Apparently it is defined, and the crashlog_addr variable apparently is used 
with an initialization of 0...

By deselecting 'crash log' kernel config option, that allowed the kernel to 
boot to a command line prompt.

John Clark.

On Oct 16, 2012, at 11:03 AM, John Clark wrote:

> 
> On Oct 15, 2012, at 8:32 AM, Rainer 'Rei' Schuth wrote:
> 
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA256
>> 
>> Hi
>> 
>> On 10/13/2012 12:21 AM, John Clark wrote:
>>> I have been looking and eventually found some set of patches which I have 
>>> been using that were posted in March.
>>> 
>>> However, what was not clear from the various posts at that time was the 
>>> following:
>>> 
>>> 1) what SVN revision of OpenWRT that was used.
>>> 2) whether anyone has done any further work.
>>> 
>> 
>> We have been using the patches since March.
>> You can find working patches under [0].
>> 
> 
> 
> Thanks for the URL. I fixed up my code set. which allows for building a 
> kernel and ramdisk. However, a couple of items.
> 
> 1) the "build" process always seems to ask about a WATCHDOG timer driver.
> 2) The built kernel crashes right after processing the initramfs image.
> 
> I have built a working Linux-3.6 kernel with the LTIB environment, so I'll be 
> looking at differences there. But if anyone has seen this sort of problem and 
> has a fix/hint I'd appreciate it.
> 
> Thanks,
> John Clark.
> 
> 
> _______________________________________________
> openwrt-devel mailing list
> [email protected]
> https://lists.openwrt.org/mailman/listinfo/openwrt-devel

_______________________________________________
openwrt-devel mailing list
[email protected]
https://lists.openwrt.org/mailman/listinfo/openwrt-devel

Reply via email to