From mboxrd@z Thu Jan 1 00:00:00 1970 From: "David S. Miller" Date: Sun, 27 Mar 2005 18:23:13 +0000 Subject: Re: [PATCH] add a clear_pages function to clear pages of higher Message-Id: <20050327102313.04498cdc.davem@davemloft.net> List-Id: References: <200503111008.12134.vda@port.imtp.ilyichevsk.odessa.ua> <200503181154.37414.vda@port.imtp.ilyichevsk.odessa.ua> <20050318192808.GB38053@muc.de> <16963.2075.713737.485070@napali.hpl.hp.com> <20050327171220.GA18506@muc.de> In-Reply-To: <20050327171220.GA18506@muc.de> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Andi Kleen Cc: davidm@hpl.hp.com, clameter@sgi.com, vda@port.imtp.ilyichevsk.odessa.ua, haveblue@us.ibm.com, akpm@osdl.org, linux-kernel@vger.kernel.org, mel@csn.ul.ie, linux-ia64@vger.kernel.org, Jens.Maurer@gmx.net On 27 Mar 2005 19:12:20 +0200 Andi Kleen wrote: > With non temporal stores > you guarantee at least one hard cache miss directly after > the return to user space. This is true if the cacheline were not present already at the time of the non-temporal store. I know what you're trying to say, I'm just clarifying. The real question is if a large enough ratio of those cachelines in the page get similarly accessed. I happen to think the answer to that for any real example is yes. Yet, I have no way to prove this. It would be cool to do some hacks under Xen or user-mode Linux to get some real statistics about this. Actually, this could be done also with hacks to valgrind or other similar tools. QEMU could also be used.