于 2011年06月21日 03:43, Andrea Arcangeli 写道:
On Tue, Jun 21, 2011 at 01:01:17AM +0800, Cong Wang wrote:
Without this patch, THP is still initialized (although khugepaged is not started),
that is what I don't want to see when I pass "transparent_hugepage=never",
because "never" for me means THP is totally unseen, even not initialized.
The ram saving by not registering in sysfs is not worth the loss of
generic functionality. You can try to make the hash and slab
khugepaged allocations more dynamic if you want to microoptimize for
RAM usage, that I wouldn't be against if you find a way to do it
simply and without much complexity (and .text) added. But likely there
are other places to optimize that may introduce less tricks and would
give you a bigger saving than ~8kbytes, it's up to you.
But the THP functionality is not going to be used.
Yeah, sounds reasonable, I will try to check if I can make it.
Thanks for pointing this out!
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>