Bad page state BUG triggered by init process (kernel 3.1.2)

[Bruce Stenning]

On Tue, Dec 13, 2011 at 12:22 PM, Bruce Stenning <b.a.stenning at gmail.com> wrote:
>> BUG: Bad page state in process busybox  pfn:00000
>> page:c0291000 count:0 mapcount:0 mapping:  (null) index:0x0
>> page flags: 0x200(arch_1)
>
> After some experimentation, the diagnostics provided by CONFIG_DMA_API_DEBUG
> have shown up problems in the IOP ADMA self-tests (inconsistencies between
> map and unmap of DMA pages) and disabling these self-tests removes the
> DMA API use warnings and prevents the bad page state bug from triggering
> when init runs.

These problems (including the kernel memory corruption issues we have also been
experiencing) have been tracked to the following issue:

http://article.gmane.org/gmane.linux.raid/33553

Bruce.