Re: [PATCH] balance inactive_dirty list

[Mark Hahn]

> while observing lots of different workloads (all I/O bound). Finally,

well, not all loads are IO-bound in the sense you're looking at.
in particular, the test I usually run (make -j2 with mem=48m)
is actually hurt by this patch.  but you're right, this change 
does improve streaming IO.

> We're trying to be too clever there, and that eventually hurts
> performance because inactive_dirty list is too small for typical

I certainly agree the code is dubious, but this is the reason 
inactive_target exists, afaikt.

> have tested. The patch simplifies code a lot and removes unnecessary
> complex calculation. Code is now completely autotuning. I have a

otoh, the "complex calculation" was always trivial, 
and *more* autotuning than your suggested fix...

> -	if (!target) {
> -		int inactive = nr_free_pages() + nr_inactive_clean_pages() +
> -						nr_inactive_dirty_pages;
> -		int active = MAX(nr_active_pages, num_physpages / 2);
> -		if (active > 10 * inactive)
> -			maxscan = nr_active_pages >> 4;
> -		else if (active > 3 * inactive)
> -			maxscan = nr_active_pages >> 8;
> -		else
> -			return 0;
> -	}
> +	if (!target)
> +		maxscan = nr_active_pages >> 4;