Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 13 May 2013 14:01:44 -0400
From:      John Baldwin <jhb@FreeBSD.org>
To:        Attilio Rao <attilio@FreeBSD.org>
Cc:        svn-src-head@freebsd.org, svn-src-all@freebsd.org, src-committers@freebsd.org
Subject:   Re: svn commit: r250601 - in head/sys: sys vm x86/acpica
Message-ID:  <51912A88.8040609@FreeBSD.org>
In-Reply-To: <201305131540.r4DFepou074563@svn.freebsd.org>
References:  <201305131540.r4DFepou074563@svn.freebsd.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On 5/13/13 11:40 AM, Attilio Rao wrote:
> Author: attilio
> Date: Mon May 13 15:40:51 2013
> New Revision: 250601
> URL: http://svnweb.freebsd.org/changeset/base/250601
> 
> Log:
>   o Add accessor functions to add and remove pages from a specific
>     freelist.
>   o Split the pool of free pages queues really by domain and not rely on
>     definition of VM_RAW_NFREELIST.
>   o For MAXMEMDOM > 1, wrap the RR allocation logic into a specific
>     function that is called when calculating the allocation domain.
>     The RR counter is kept, currently, per-thread.
>     In the future it is expected that such function evolves in a real
>     policy decision referee, based on specific informations retrieved by
>     per-thread and per-vm_object attributes.
>   o Add the concept of "probed domains" under the form of vm_ndomains.
>     It is responsibility for every architecture willing to support multiple
>     memory domains to correctly probe vm_ndomains along with mem_affinity
>     segments attributes.  Those two values are supposed to remain always
>     consistent.
>     Please also note that vm_ndomains and td_dom_rr_idx are both int
>     because segments already store domains as int.  Ideally u_int would
>     have much more sense. Probabilly this should be cleaned up in the
>     future.
>   o Apply RR domain selection also to vm_phys_zero_pages_idle().

My other comments aside: to be clear (this log message doesn't make it
clear I think): this changes our NUMA allocation policy to be
round-robin instead of first touch (which is probably better for the
general case).

Also, if you want to honor SLIT, you will need to put back something
like the lookup lists from before, though in this case it can just be
a simple list of domains for each domain.

-- 
John Baldwin



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?51912A88.8040609>