From owner-freebsd-hackers@FreeBSD.ORG Wed Jun 2 07:12:37 2004 Return-Path: Delivered-To: freebsd-hackers@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 5A3DF16A4D9; Wed, 2 Jun 2004 07:12:37 -0700 (PDT) Received: from freefall.freebsd.org (freefall.freebsd.org [216.136.204.21]) by mx1.FreeBSD.org (Postfix) with ESMTP id 2115343D2F; Wed, 2 Jun 2004 07:12:37 -0700 (PDT) (envelope-from bmilekic@FreeBSD.org) Received: from freefall.freebsd.org (bmilekic@localhost [127.0.0.1]) i52ECbrX037145; Wed, 2 Jun 2004 07:12:37 -0700 (PDT) (envelope-from bmilekic@freefall.freebsd.org) Received: (from bmilekic@localhost) by freefall.freebsd.org (8.12.11/8.12.11/Submit) id i52ECYrK037144; Wed, 2 Jun 2004 07:12:34 -0700 (PDT) (envelope-from bmilekic) Date: Wed, 2 Jun 2004 07:12:34 -0700 From: Bosko Milekic To: Gleb Smirnoff Message-ID: <20040602141234.GA33162@freefall.freebsd.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.4.1i cc: freebsd-hackers@freebsd.org cc: freebsd-current@freebsd.org Subject: Re: [HEADS-UP] mbuma is in the tree X-BeenThere: freebsd-hackers@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Technical Discussions relating to FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 02 Jun 2004 14:12:37 -0000 > Bosko, [deletia] > are you going to convert mbuf tag allocator to UMA? Now >tags are allocated with malloc(). AFAIK, tags are used heavily in pf, >and forthcoming ALTQ. Moving to UMA should affect their performance >positively. First off, malloc() *is* UMA. With mbuma in the tree, I don't believe we have any remaining custom-allocators in the tree. As for what to do with m_tags, it is still unclear to me. Personally, I'm conflicted about their use. On one hand, they offer a clean way to attach metadata to packets, but on the other hand they are quite expensive. If you read the paper on mbuma, you'll notice that I point out that it would be worth investigating whether, in scenarios where an m_tag is ALWAYS required per packet (e.g., MAC), providing a secondary zone with pre-allocated m_tags for packet headers might be worth it. Prior to this work, however, I suggest we investigate the possibility of using smaller mini-mbufs whenever clusters are used so that space wastage is reduced. -Bosko