From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:42415) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YXlps-0003Ac-Qi for qemu-devel@nongnu.org; Tue, 17 Mar 2015 03:22:45 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1YXlpp-0005Db-Iq for qemu-devel@nongnu.org; Tue, 17 Mar 2015 03:22:44 -0400 Received: from mx1.redhat.com ([209.132.183.28]:40632) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YXlpp-0005DR-Ay for qemu-devel@nongnu.org; Tue, 17 Mar 2015 03:22:41 -0400 From: Markus Armbruster References: <55071113.9010406@redhat.com> Date: Tue, 17 Mar 2015 08:22:38 +0100 In-Reply-To: <55071113.9010406@redhat.com> (Paolo Bonzini's message of "Mon, 16 Mar 2015 18:21:23 +0100") Message-ID: <87fv94m6kh.fsf@blackfin.pond.sub.org> MIME-Version: 1.0 Content-Type: text/plain Subject: Re: [Qemu-devel] lsi53c895a lsi_do_dma assertion failure List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Paolo Bonzini Cc: Chris Takemura , qemu-devel Paolo Bonzini writes: > On 16/03/2015 16:59, Chris Takemura wrote: >> Hi everyone, >> >> We're using qemu as a device model for xen 4.4. Lately we've been >> having qemu crashes with this output: >> >> lsi_scsi: error: Unhandled writeb 0xff = 0x0 >> >> lsi_scsi: error: Unhandled writeb 0x100 = 0x0 >> >> lsi_scsi: error: Unhandled writeb 0x101 = 0x0 >> >> lsi_scsi: error: Unhandled writeb 0x102 = 0x0 >> >> qemu-system-i386: /home/chris/qemu/hw/scsi/lsi53c895a.c:541: lsi_do_dma: >> Assertion `s->current' failed. >> >> >> This is with a clean qemu-2.1.0 tree, built for Xen. >> >> I confirmed that commit d1d74664ea99cdc571afee12e31c8625595765b0 is >> present in our code. (As described in the thread that begins >> with >> http://qemu-devel.nongnu.narkive.com/LtGj28sZ/win2k8-32-bit-mix-of-ide-and-scsi-assertion >> ). > > What guest is this? The answer is basically "do not use this device > model", it is old and buggy and newer guests do not even ship the driver > anymore. I agree with the advice, but I can't help to notice we're not taking it ourselves: -device if=scsi still creates lsi53c895a HBAs for many machines. I think we should fix that for new machine models.