From owner-freebsd-mobile Sun Nov 2 17:51:49 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id RAA26096 for mobile-outgoing; Sun, 2 Nov 1997 17:51:49 -0800 (PST) (envelope-from owner-freebsd-mobile) Received: from ns.mt.sri.com (SRI-56K-FR.mt.net [206.127.65.42]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id RAA26080 for ; Sun, 2 Nov 1997 17:51:40 -0800 (PST) (envelope-from nate@rocky.mt.sri.com) Received: from rocky.mt.sri.com (rocky.mt.sri.com [206.127.76.100]) by ns.mt.sri.com (8.8.7/8.8.7) with ESMTP id SAA12267; Sun, 2 Nov 1997 18:48:53 -0700 (MST) Received: (from nate@localhost) by rocky.mt.sri.com (8.7.5/8.7.3) id SAA04825; Sun, 2 Nov 1997 18:48:52 -0700 (MST) Date: Sun, 2 Nov 1997 18:48:52 -0700 (MST) Message-Id: <199711030148.SAA04825@rocky.mt.sri.com> From: Nate Williams MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit To: Mike Smith Cc: Guido van Rooij , guido@gvr.gvr.org (guido), freebsd-mobile@freebsd.org Subject: Re: card removal problems In-Reply-To: <199711030049.LAA00584@word.smith.net.au> References: <199711021414.PAA04580@gvr.gvr.org> <199711030049.LAA00584@word.smith.net.au> X-Mailer: VM 6.29 under 19.15 XEmacs Lucid Sender: owner-freebsd-mobile@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > > An update: I tried to remove 2 modem cards for a couple of times. > > The Nokia one (irq3) did lock up my system each time (I tried 3 times). > > The other one (angia) went ok for 2 times. Note that in the lockup > > case, nothing is printed on the console. > > If you feel like patching things, try putting a test in the sio > driver's interrupt handler that checks the 'gone' flag, and puke if it > gets an interrupt when the card is 'gone'. This is similar to the (off-line) advice I gave him. > It should also perform some > basic sanity checks on the data it's getting; the sio interrupt is > handled by a fast interrupt handler which will preclude (AFAIK) any > other interrupt handler from running. Actually, no. We don't register any card interrupts as 'fast', which could also be some cause for concern. (Bruce has mentioned it in the past, but things seem to work, so I haven't made any changes.) > If the modem generates an interrupt as it goes, it's possible that > it's spinning forever in the handler. That's my suspicion as well, but there are some sanity checks in there. As I pointed out in private email, we may be missing one or doing it in the wrong order, opening up a race condition. Nate