From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jason Gunthorpe Subject: Re: [PATCH v2 03/12] IB/hfi1: Fix an Oops on pci device force remove Date: Tue, 25 Oct 2016 12:03:45 -0600 Message-ID: <20161025180345.GA16794@obsidianresearch.com> References: <20161017160731.GA5679@obsidianresearch.com> <20161025155754.4950.23412.stgit@scvm10.sc.intel.com> <20161025164851.GA28096@obsidianresearch.com> <7f4cbe0c-0c83-48b2-9901-4a5e27b306b4@intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <7f4cbe0c-0c83-48b2-9901-4a5e27b306b4-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org> Sender: linux-rdma-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Tadeusz Struk Cc: Dennis Dalessandro , dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org, linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, Dean Luick , Ira Weiny List-Id: linux-rdma@vger.kernel.org On Tue, Oct 25, 2016 at 10:38:55AM -0700, Tadeusz Struk wrote: > On 10/25/2016 09:48 AM, Jason Gunthorpe wrote: > > nit: This isn't similar to uverbs or other IB core interfaces - they > > all allow unbind while the FD remains open, this forces the FD to > > become closed - which means hot plug will not be supported by hfi1. > > > > That's correct, the reason, as I said before, is that PIO processes > don't go via FD, but talk directly to MMIO PCI regions. Most of the IB drivers do this - Mellanox implemented a forced remapping of the user space VMA on unplug to deal with it, you can use that technique in HFI1 too. Jason -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html