From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paolo Bonzini Subject: Re: [PATCH 1/2] tcm_vhost: Wait for pending requests in vhost_scsi_flush() Date: Mon, 11 Mar 2013 13:15:21 +0100 Message-ID: <513DCAD9.1020308@redhat.com> References: <1362978579-13322-1-git-send-email-asias@redhat.com> <1362978579-13322-2-git-send-email-asias@redhat.com> <513DC1C5.4070106@redhat.com> <20130311115356.GD23359@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20130311115356.GD23359@redhat.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: virtualization-bounces@lists.linux-foundation.org Errors-To: virtualization-bounces@lists.linux-foundation.org To: "Michael S. Tsirkin" Cc: kvm@vger.kernel.org, virtualization@lists.linux-foundation.org, target-devel@vger.kernel.org, Stefan Hajnoczi List-Id: virtualization@lists.linuxfoundation.org Il 11/03/2013 12:53, Michael S. Tsirkin ha scritto: > On Mon, Mar 11, 2013 at 12:36:37PM +0100, Paolo Bonzini wrote: >> Il 11/03/2013 06:09, Asias He ha scritto: >>> This patch makes vhost_scsi_flush() wait for all the pending requests >>> issued before the flush operation to be finished. >> >> There is no protection against issuing concurrent flush operations. If >> we later would like to make the flush a ioctl (for example for migration >> purposes), > > For migration I think we need to stop vhost, flush isn't useful > if we keep submitting requests afterwards. This ioctl would run when the vm is stopped. Paolo >> it would be confusing, and I'm not sure how you could extend >> the during_flush machinery. >> >> What about making vhost_scsi_flush() wait for all pending requests, >> including those issues during the flush operation? Then you can easily >> support concurrent flushes; just add a waitqueue and wake_up_all at the >> end of the flush operation. >> >> BTW, adding such a ioctl as part of this patch would probably be a good >> thing to do anyway. > > Please keep it separate, it's painful enough that we have a driver > with to established userspace, I don't want to add ioctls > until it's used by some upstream. >