From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tejun Heo Subject: Re: [PATCH 2/2] block: Avoid deadlocks with bio allocation by stacking drivers Date: Mon, 10 Sep 2012 16:13:26 -0700 Message-ID: <20120910231326.GG7677@google.com> References: <1347055973-11581-3-git-send-email-koverstreet@google.com> <20120908193641.GB12773@dhcp-172-17-108-109.mtv.corp.google.com> <20120910002810.GA23241@moria.home.lan> <20120910172210.GC14103@google.com> <20120910202435.GG16360@google.com> <20120910204010.GA32310@google.com> <20120910213349.GH16360@google.com> <20120910213710.GA7677@google.com> <20120910215633.GA19739@google.com> <20120910220910.GB7677@google.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20120910220910.GB7677-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org> Sender: linux-bcache-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Kent Overstreet Cc: linux-bcache-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, dm-devel-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org, axboe-tSWWG44O7X1aa/9Udqfwiw@public.gmane.org, Vivek Goyal , Mikulas Patocka , bharrosh-C4P08NqkoRlBDgjK7y7TUQ@public.gmane.org, david-FqsqvQoI3Ljby3iVrkZq2A@public.gmane.org List-Id: linux-bcache@vger.kernel.org Hello, again. On Mon, Sep 10, 2012 at 03:09:10PM -0700, Tejun Heo wrote: > I'm still a bit scared but think this is correct. > > Acked-by: Tejun Heo > > One last thing is that we may want to add @name on bioset creation so > that we can name the workqueue properly but that's for another patch. Yet another thing that I noticed in a different discussion. http://thread.gmane.org/gmane.linux.network.drbd.devel/2130 Before this, I think bios didn't get reordered while they're traveling down the stacked bio drivers. After this, I don't think that's true anymore. Currently, IIRC, we don't have any ordering functionality at bio interface, so I don't think this breaks anything but this can lead to stupidly subtle bugs if the upper layer is making assumption on ordering somehow. It's something which at least should be noted, I think. Whether we want to update the code so that ordering is maintained, I don't know. I hope not. It's already crazy complex. :( Thanks. -- tejun