Specifically, I added the appropriate entries to the size_descriptor
table of kmalloc.c and
modified the NR_MEM_LISTS used by the page allocation routines. So
far, I haven't experienced any problems, and the memory allocation
seems to be working fine.
However, since the change seemed so trivial, I am concerned that
there are factors of which I am unaware. The changes seem perfectly
in line with the nature of the 'buddy system', but then why has
the upper bound of kmalloc been limited to what seems a relatively
small value? Are there other dependencies, etc., which I overlooked?
I would greatly appreciate any insight into the matter.
Thanks,
Matt Sexton
eldiablo@mit.edu