From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeff Garzik Subject: Re: Core scsi layer crashes in 2.6.8.1 Date: Tue, 5 Oct 2004 12:53:23 -0400 Sender: linux-scsi-owner@vger.kernel.org Message-ID: <20041005165323.GA24922@havoc.gtf.org> References: <1096401785.13936.5.camel@localhost.localdomain> <4162B345.9000806@rtr.ca> <1096988167.2064.7.camel@mulgrave> <200410051749.22245.oliver@neukum.org> <1096991666.2064.25.camel@mulgrave> <4162C474.8010505@rtr.ca> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from havoc.gtf.org ([69.28.190.101]:6079 "EHLO havoc.gtf.org") by vger.kernel.org with ESMTP id S269116AbUJEQ7N (ORCPT ); Tue, 5 Oct 2004 12:59:13 -0400 Content-Disposition: inline In-Reply-To: <4162C474.8010505@rtr.ca> List-Id: linux-scsi@vger.kernel.org To: Mark Lord Cc: James Bottomley , Oliver Neukum , Anton Blanchard , Alan Cox , Linux Kernel Mailing List , SCSI Mailing List On Tue, Oct 05, 2004 at 11:57:40AM -0400, Mark Lord wrote: > James Bottomley wrote: > > > >It would add quite a bit of complexity to the reference counted > >aynchronous model to try and force synchronicity between queuecommand > >and scsi_remove_host in the mid-layer. Therefore it's much easier to > >let the LLD decide what to do with the command. > > Presumably the same is also true for scsi_remove_device() ? What I do in my local hotplug code is assume that the SCSI layer will be stupid and send commands after I call scsi_remove_device(), for an indeterminant but short period of time. Jeff