> Under high stress (`make -j' in two kernel trees, a `tar tfvz BIGFILE.gz >
> /dev/null', and starting grok with a real big data base) I got from the
> tar job once
>
> Jul 22 21:07:08 boole kernel: release: gzip kernel stack corruption. Aiee
>
> at a load higher than 42 (ehh). The debug message in try_to_free_message was
> never triggered.
this one should be considered harmless in this context, try to find out
which kernel function in your kernel uses up too much stack space. [i've
sent a script to linux-kernel a few months ago that does this
automatically]
[or if it happens repeatedly and you cant find out why it happens, try the
kernel stack overflow detector patch]
-- mingo