From owner-freebsd-arch@FreeBSD.ORG Fri Oct 7 09:47:15 2005 Return-Path: X-Original-To: arch@FreeBSD.org Delivered-To: freebsd-arch@FreeBSD.ORG Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 2E97A16A41F for ; Fri, 7 Oct 2005 09:47:15 +0000 (GMT) (envelope-from glebius@FreeBSD.org) Received: from cell.sick.ru (cell.sick.ru [217.72.144.68]) by mx1.FreeBSD.org (Postfix) with ESMTP id 9E52843D46 for ; Fri, 7 Oct 2005 09:47:14 +0000 (GMT) (envelope-from glebius@FreeBSD.org) Received: from cell.sick.ru (glebius@localhost [127.0.0.1]) by cell.sick.ru (8.13.3/8.13.3) with ESMTP id j979lCJS038501 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Fri, 7 Oct 2005 13:47:13 +0400 (MSD) (envelope-from glebius@FreeBSD.org) Received: (from glebius@localhost) by cell.sick.ru (8.13.3/8.13.1/Submit) id j979lCuh038500; Fri, 7 Oct 2005 13:47:12 +0400 (MSD) (envelope-from glebius@FreeBSD.org) X-Authentication-Warning: cell.sick.ru: glebius set sender to glebius@FreeBSD.org using -f Date: Fri, 7 Oct 2005 13:47:12 +0400 From: Gleb Smirnoff To: dima <_pppp@mail.ru> Message-ID: <20051007094712.GK14542@cell.sick.ru> References: <20051006183413.GH14542@cell.sick.ru> Mime-Version: 1.0 Content-Type: text/plain; charset=koi8-r Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.6i Cc: arch@FreeBSD.org Subject: Re: [REVIEW/TEST] polling(4) changes X-BeenThere: freebsd-arch@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Discussion related to FreeBSD architecture List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 07 Oct 2005 09:47:15 -0000 dima, On Fri, Oct 07, 2005 at 01:28:58PM +0400, dima wrote: d> > d> Seems to be a first considerable step regarding the ideas discussed in March :) d> > d> But, my idea about the separate locking of each interface dissappeared from this implementation. mtx_poll is good to protect the pollrec array and other sensitive variables. But we could get advantage of SMP machines writing polling loops like this: d> > d> d> > d> for( i = 0; i < poll_handlers; ++i ) { d> > d> mtx_lock( &iface_lock[i] ); d> > d> pr[i].handler(pr[i].ifp, POLL_ONLY, count); d> > d> mtx_unlock( &iface_lock[i] ); d> > d> } d> > d> > What is the benefit here? The driver must have its own lock. d> d> Well, consider the absense of the mtx_poll lock: d> d> - mtx_lock( &mtx_poll ); d> for( i = 0; i < poll_handlers; ++i ) { d> + mtx_lock( &iface_lock[i] ); d> pr[i].handler( pr[i].ifp, POLL_ONLY, count ); d> + mtx_unlock( &iface_lock[i] ); d> } d> - mtx_unlock( &mtx_poll ); d> d> So, several kernel threads in an SMP machine can poll different interfaces simultaneously. And mtx_lock should only be used in ether_poll_[de]register(). Imagining that we will have several polling threads in future, the above design has some disadvantages, I think: First, we still need to protect the array pr[], with some mutex while traversing it, and while editing it in ether_poll_[de]register. May be like it was done in kern_poll.c, rev 1.21. Second, the approach above won't give a nice parallelization. Imagine two threads, both working in a cycle shown above. They will contest on the lock of each interface: - t1 starts - t1 locks iface_lock[1] - t2 starts - t1 polls pr[1]... - t2 blocks on iface_lock[1] - t1 polls pr[1]... - t1 polls pr[1]... - t1 polls pr[1]... - t1 polls pr[1]... - t1 unlocks iface_lock[1] - t2 locks iface_lock[1] - t1 locks iface_lock[2] - t2 polls empty pr[1], quickly returns - t1 polls pr[2]... - t2 unlocks iface_lock[1] - t1 polls pr[2]... - t2 blocks on iface_lock[2] - t1 polls pr[2]... - t1 polls pr[2]... - t1 polls pr[2]... - t1 polls pr[2]... - t1 unlocks iface_lock[2] - t2 locks iface_lock[2] - t1 locks iface_lock[3] - t2 polls empty pr[2], quickly returns - t1 polls pr[3]... - t2 unlocks iface_lock[2] So, one thread works, and other just goes after the first one, and picks only a small number of packets, or even just wastes CPU cycles. Really we do not have several kernel threads in polling. netisr_poll() is always run by one thread - swi1:net. Well, we have also idle_poll thread, but it is very special case. Frankly speaking, it can't work without help from netisr_poll(). The current polling is designed for a single threaded kernel, for RELENG_4. We can't achieve parallelization with strong redesign. The future plans are to create per-interface CPU bound threads. The plans can change. You are welcome to help. -- Totus tuus, Glebius. GLEBIUS-RIPN GLEB-RIPE