we run out of ZONE_NORMAL but until we run oom it doesn't mean we shrink
it frequently, those are pretty static allocations (the mem_map is the
most static and in turn the biggest one ;), but it doesn't mean, we
allocate 500M and release 500M in a few seconds of zone normal. Also
many like selects are released and reallocated before you've a chance to
need a shrink.
The shrinkers are the ones allocating huge chunks in a loop and never
releasing it except through the VM (an updatedb would do it with the
lots of metadata overhead, but it's not really as common as the
highmem ones).
So your argument about ZONE_NORMAL being uner much more pressure doesn't
make sense to me, in line with my answer to Andrew that the frequency
of allocations (NOTE: without later explicit deallocations but relaying
on the cache collecting in the vm).
Andrea
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/