Comment # 6 on bug 1162590 from Michal Kube��ek

(In reply to Michal Kube��ek from comment #4)
> I'll try to add some tracing printk() calls into the function and see if it
> isn't called at some earlier time (e.g. while booting the system). It could
> leave some mess behind which would only cause trouble later when the system
> gets under load.

And the only result is that tomoyo_realpath_from_path() is probably never
called. I added some printk() to see what is going on, one of them right at
the beginning of the function - and there was nothing in the log while the
build still failed with random ICEs.

Maybe a bit of hint (even if less than I hoped for) is that when I booted the
same kernel with "slub_debug" or "slab_nomerge", the issue did not manifest.