It appears that the "sequential" argument to free_kpages, as called by
the kmalloc code, can have one of two values:
1, meaning the address being freed is a single page kmalloc was
using for small allocations; or
2, meaning the address being freed is either a large allocation
done with alloc_kpages, or a completely invalid free attempt.
In the first case the block being freed should be exactly one page
long; in the second case, however, it may also be exactly one page
long (or it may be longer).
It is thus questionable whether this argument provides any useful data
at all; you may find it simplest to ignore it.
In any event, I apologize for the undocumented and icky use of magic
signalling values.
--
- David A. Holland / dholland(a)eecs.harvard.edu
Show replies by date