From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Lameter Subject: Re: [rfc] approach to pull writepage out of reclaim Date: Thu, 09 Oct 2008 15:45:00 -0500 Message-ID: <48EE6D4C.7080901@linux-foundation.org> References: <20081009144103.GE9941@wotan.suse.de> <48EE3A07.9060205@linux-foundation.org> <20081009194434.GB25780@parisc-linux.org> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: Nick Piggin , Linux Memory Management List , linux-fsdevel@vger.kernel.org To: Matthew Wilcox Return-path: Received: from smtp1.linux-foundation.org ([140.211.169.13]:49295 "EHLO smtp1.linux-foundation.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751416AbYJIUpp (ORCPT ); Thu, 9 Oct 2008 16:45:45 -0400 In-Reply-To: <20081009194434.GB25780@parisc-linux.org> Sender: linux-fsdevel-owner@vger.kernel.org List-ID: Matthew Wilcox wrote: > On Thu, Oct 09, 2008 at 12:06:15PM -0500, Christoph Lameter wrote: >> Nick Piggin wrote: >> >>> So. Firstly, what I'm looking at is doing swap writeout from pdflush. This >>> patch does that (working in concept, but pdflush and background writeout >>> from dirty inode list isn't really up to the task, might scrap it and do the >>> writeout from kswap). But writeout from radix-tree should actually be able to >>> give better swapout pattern than LRU writepage as well. >> Patch is missing from the message. > > It's no longer acceptable to post descriptions of what you're about to > do? You have to invest lots of time into creating a patch and testing that > it works before posting it (only to have it shot down because someone > disagrees with the design of your solution)? Really? The text says that a patch was included.... So I was expecting it.... But the problem you mention is real. Tried numerous times to get a conceptual discussion going without a patch. Usually that does not lead to anything.