From owner-freebsd-current Tue Dec 17 12:24:11 2002 Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 9397937B401 for ; Tue, 17 Dec 2002 12:24:10 -0800 (PST) Received: from ada.snu.ac.kr (ada.snu.ac.kr [147.46.106.49]) by mx1.FreeBSD.org (Postfix) with ESMTP id 7BD7043EE6 for ; Tue, 17 Dec 2002 12:24:09 -0800 (PST) (envelope-from redjade@ada.snu.ac.kr) Received: from ada.snu.ac.kr (ada.snu.ac.kr [147.46.106.49]) by ada.snu.ac.kr (8.12.6/8.12.6) with ESMTP id gBHKO3jC057407; Wed, 18 Dec 2002 05:24:03 +0900 (KST) (envelope-from redjade@ada.snu.ac.kr) Received: (from redjade@localhost) by ada.snu.ac.kr (8.12.6/8.12.6/Submit) id gBHKO3TS057406; Wed, 18 Dec 2002 05:24:03 +0900 (KST) Date: Wed, 18 Dec 2002 05:24:03 +0900 From: Kyunghwan Kim To: Andrew Gallatin Cc: current@FreeBSD.ORG Subject: Re: INTR_MPSAFE to network device drivers Message-ID: <20021217202402.GA57385@ada.snu.ac.kr> References: <20021217191841.GA57094@ada.snu.ac.kr> <15871.31635.962334.855790@grasshopper.cs.duke.edu> <20021217195300.GB57094@ada.snu.ac.kr> Mime-Version: 1.0 Content-Type: text/plain; charset=euc-kr Content-Disposition: inline In-Reply-To: <20021217195300.GB57094@ada.snu.ac.kr> User-Agent: Mutt/1.4i X-My-Present-Organization: Innuworks, Inc. Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG On Wed, Dec 18, 2002 at 04:53:00AM +0900, Kyunghwan Kim wrote: > On Tue, Dec 17, 2002 at 02:31:31PM -0500, Andrew Gallatin wrote: > > > mbuf and bpf routines are all mp-safe, so it seems that > > > it is safe to make network device drivers out of Giant lock. > > > Or is there any unresolved related issues? > > > > Yes, the mbuf allocator must occasionally call kmem_malloc(), which > > requires Giant. This means no net driver can be made INTR_MPSAFE, > > or it will eventually panic when kmem_malloc is called. > > I found and read the thread that you and Alan had discussed about this > problem just before. Then what about making updated version of mb_pop_cont() > that accepts occasionally acquiring Giant? Oh, sorry. Conclusion of the thread was preallocation. But it doesn't seem that preallocation is the correct way. -- Kyunghwan Kim redjade@ada.snu.ac.kr To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message