Date: Wed, 3 Mar 2004 22:05:54 +0100
From: Andrea Arcangeli <andrea@suse.de>
To: Dave McCracken <dmccr@us.ibm.com>
Cc: "Martin J. Bligh" <mbligh@aracnet.com>, linux-kernel@vger.kernel.org
Subject: Re: 230-objrmap fixes for 2.6.3-mjb2
Message-ID: <20040303210554.GW4922@dualathlon.random>
References: <20040303070933.GB4922@dualathlon.random> <20040303025820.2cf6078a.akpm@osdl.org> <7440000.1078328791@[10.10.2.4]> <20040303165746.GO4922@dualathlon.random> <10500000.1078333658@[10.1.1.4]> <20040303183901.GU4922@dualathlon.random> <14140000.1078339447@[10.1.1.4]> <20040303185122.GV4922@dualathlon.random> <14830000.1078340481@[10.1.1.4]>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <14830000.1078340481@[10.1.1.4]>
User-Agent: Mutt/1.4.1i
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1056
Lines: 20

On Wed, Mar 03, 2004 at 01:01:21PM -0600, Dave McCracken wrote:
> The point would be to have a way of finding and skipping the locked page
> before we go look up the vmas for it.  2.4 doesn't have that problem
> because it's working from the vma.

it has the same problem because we scan those pages always at the
physical side and always at the pagetable side. we could avoid scanning
them at the pagetable side, but we scan them anyways exactly to be able
to call mark_page_accessed that will move the page in the active list,
so the physical side doesn't waste an excessive amount of cpu on
unfreeable mlocked pages. 

the trick is to keep marking those mlocked pages as accessed (so they go
into the active list) every time we encounter them and we find them
mlocked, so they're not in our way all the time.
-
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/