From mboxrd@z Thu Jan 1 00:00:00 1970 Date: Mon, 29 Jan 2007 12:18:29 -0800 (PST) From: Linus Torvalds Subject: Re: [patch] mm: mremap correct rmap accounting In-Reply-To: <20070129120325.26707d26.akpm@osdl.org> Message-ID: References: <45B61967.5000302@yahoo.com.au> <45BD6A7B.7070501@yahoo.com.au> <20070129120325.26707d26.akpm@osdl.org> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-mm@kvack.org Return-Path: To: Andrew Morton Cc: Hugh Dickins , Nick Piggin , Linux Memory Management , Ralf Baechle List-ID: On Mon, 29 Jan 2007, Andrew Morton wrote: > > Can we convert those bits of mips to just have a single zero-page, like > everyone else? > > Is that trick a correctness thing, or a performance thing? If the latter, > how useful is it, and how common are the chips which use it? It was a performance thing, iirc. Apparently a fairly big deal: pages kicking each other out of the cache due to idiotic cache design. But I forget any details. MIPS in general is a f*cking pain in the *ss. They have a few chips that are sane, but just an incredible amount of totally braindamaged ones. They're not the only ones with virtual caches, but they're certainly well-represented there. Sad. Linus -- 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