Re: lowmem_reserve (replaces protection)

[Andrea Arcangeli <andrea@novell.com>]

On Tue, Oct 26, 2004 at 08:31:32PM -0400, Rik van Riel wrote:
> On Tue, 26 Oct 2004, Nick Piggin wrote:
> 
> > OK that makes sense... it isn't the length of the name, but the fact
> > that that naming convention hasn't proliferated thoughout the 2.6 tree;
> 
> Speaking about not proliferating...
> 
> One thing we need to make sure of is that the lower zone
> protection stuff doesn't put the allocation threshold
> higher than kswapd's freeing threshold.

I agree. I didn't introduce that bug, the very same problem would happen
with the previous protection code. So this is not a regression, I'm far
from finished... I'm just trying to post orthogonal patches, since Hugh
had a much better merging success rate with small patches (though I find
very hard to produce small patches myself when there's more than one
thing to fix in the same file).

the per-classzone kswapd treshold was very well taken care of in 2.4,
thanks the watermarks embedding the low/min/high and the classzone being
passed up to the kswapd wakeup function.

> Otherwise on a 1GB system, we'll end up cycling most of
> userspace allocations through the 128MB highmem zone,
> instead of falling back to the other zones.

that's the side effect of the per-zone lru too (though I'm not going to
change the lru).