From: Markus Armbruster Subject: Re: [Patch] fix xenfb_update_screen bogus rect Date: Wed, 28 Jan 2009 15:31:38 +0100 Message-ID: <871vun33jp.fsf@pike.pond.sub.org> References: <497E784B.7060003@jp.fujitsu.com> <497ED10A.76E4.0078.0@novell.com> <498009CC.1000101@jp.fujitsu.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: In-Reply-To: <498009CC.1000101@jp.fujitsu.com> (Akio Takebe's message of "Wed\, 28 Jan 2009 16\:31\:24 +0900") List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Akio Takebe Cc: xen-devel List-Id: xen-devel@lists.xenproject.org Akio Takebe writes: > Hi, Jan > > Thank you for your review. > I remade it simpler. > But Just moving check-and-clear dirty flag to xenfb_update_screen(), > a guest with vcpu=1 could not boot. > It is caused by info->update_wanted = 0. Thanks for debugging this! > So I moved kthread_run() into XenbusStateConnected handling. > How about it? > > Signed-off-by: Akio Takebe > > Best Regards, > > Akio Takebe [...] > diff -r 83b71f4b5cb2 drivers/xen/fbfront/xenfb.c > --- a/drivers/xen/fbfront/xenfb.c Tue Jan 20 13:28:35 2009 +0000 > +++ b/drivers/xen/fbfront/xenfb.c Thu Jan 29 01:24:06 2009 +0900 > @@ -213,17 +213,23 @@ Please use -p with diff. > if (xenfb_queue_full(info)) > return; > > - mutex_lock(&info->mm_lock); > - > spin_lock_irqsave(&info->dirty_lock, flags); > - y1 = info->y1; > - y2 = info->y2; > - x1 = info->x1; > - x2 = info->x2; > - info->x1 = info->y1 = INT_MAX; > - info->x2 = info->y2 = 0; > + if (info->dirty){ > + info->dirty = 0; > + y1 = info->y1; > + y2 = info->y2; > + x1 = info->x1; > + x2 = info->x2; > + info->x1 = info->y1 = INT_MAX; > + info->x2 = info->y2 = 0; > + } else { > + spin_unlock_irqrestore(&info->dirty_lock, flags); > + return; > + } > spin_unlock_irqrestore(&info->dirty_lock, flags); > > + mutex_lock(&info->mm_lock); > + > list_for_each_entry(map, &info->mappings, link) { > if (!map->faults) > continue; Careful, locking is rather delicate here. Please read the big comment "There are three locks:", then explain why moving the locking of mm_lock is safe. [...]