From mboxrd@z Thu Jan 1 00:00:00 1970 From: Anton Blanchard Subject: Re: [PATCH] serialize bus scanning Date: Thu, 28 Aug 2003 06:29:08 +1000 Sender: linux-scsi-owner@vger.kernel.org Message-ID: <20030827202908.GB7852@krispykreme> References: <20030825122416.GE15506@lst.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from dp.samba.org ([66.70.73.150]:15274 "EHLO lists.samba.org") by vger.kernel.org with ESMTP id S262288AbTH0UdU (ORCPT ); Wed, 27 Aug 2003 16:33:20 -0400 Content-Disposition: inline In-Reply-To: <20030825122416.GE15506@lst.de> List-Id: linux-scsi@vger.kernel.org To: Christoph Hellwig Cc: James Bottomley , linux-scsi@vger.kernel.org Hi, > Synchronize all scanning activity, this fixes long-standing races > vs /proc/scsi/scsi and sysfs addition and deletion of devices. > > Note that this does not serialize removing, the lists will get > their own locking soon. Ive started playing with parallel probe in the lab (on some large ppc64 boxes). I know we have a lot of work to do before this will all work, but would it be possible to make this semaphore per host? Even with a per host semaphore, it annoys me how long it takes to spin up 7 disks especially if a few of them are dead. Longer term are we planning to overlap spinup of disks on the same host? (We have a property in our device tree that says how quickly you can spin disks up on the same host, eg 1sec apart. We should also be able to do this in parallel on all hosts. We could easily read this property in a small initramfs program and then issue probe commands via /sys) Anton