From mboxrd@z Thu Jan 1 00:00:00 1970 Date: Tue, 18 Jul 2006 07:03:12 -0700 (PDT) From: Christoph Lameter Subject: Re: [PATCH] mm: inactive-clean list In-Reply-To: <44BCE86A.4030602@mbligh.org> Message-ID: References: <1153167857.31891.78.camel@lappy> <1153224998.2041.15.camel@lappy> <44BCE86A.4030602@mbligh.org> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-mm@kvack.org Return-Path: To: "Martin J. Bligh" Cc: Peter Zijlstra , linux-mm , Linus Torvalds , Andrew Morton , linux-kernel List-ID: On Tue, 18 Jul 2006, Martin J. Bligh wrote: > > Adding logic to determine the number of clean pages is not necessary. The > > number of clean pages in the pagecache can be determined by: > > > > global_page_state(NR_FILE_PAGES) - global_page_state(NR_FILE_DIRTY) > > It's not that simple. We also need to deal with other types of non-freeable > pages, such as memlocked. mlocked is an exceptional case. The problem is that the information if a page is mlocked is only available via the vma. One has to scan the reverse list and check all the vmas for the flag. Is mlock that important? What other types of non freeable pages could exist? Maybe slab allocations and direct kernel allocations? We have only limited means to reclaim those pages. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org