On Tue, 2 May 2000 frankeh@us.ibm.com wrote:
>The smart things that I see has to happen is to allow a set of processes to
>be attached to a set of memory pools and the OS basically enforcing
>allocation in those constraints. I brought this up before and I think
>Andrea proposed something similar. Allocation should take place in those
Yes, that's why I think we need to be able to know the state of the cache
in a single pg_data_t. If 99% of the pg_data_t is _freeable_ cache it
worth to shrink a bit from the cache of _such_ pg_data_t instead of
risking shrinking and then allocating the memory from a foregin pg_data_t
because we respect a global LRU). This can't hurt at all the common non
NUMA case since in the common case of 99% of IA32 boxes out there we have
_one_ only pg_data_t thus the lru keeps to be effectively system-global
for them.
Andrea
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sun May 07 2000 - 21:00:10 EST