From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Michael S. Tsirkin" Subject: Re: [PATCH 12/22] virtio_pci: allow duplicate capabilities. Date: Thu, 21 Mar 2013 17:58:10 +0200 Message-ID: <20130321155810.GA1925@redhat.com> References: <1363854584-25795-1-git-send-email-rusty@rustcorp.com.au> <1363854584-25795-13-git-send-email-rusty@rustcorp.com.au> <20130321102814.GC30493@redhat.com> <514B188A.3030502@zytor.com> <20130321144330.GA1454@redhat.com> <514B1D25.9090206@zytor.com> <20130321151937.GB1454@redhat.com> <514B26BC.6010700@zytor.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Content-Disposition: inline In-Reply-To: <514B26BC.6010700@zytor.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: "H. Peter Anvin" Cc: virtualization@lists.linux-foundation.org List-Id: virtualization@lists.linuxfoundation.org On Thu, Mar 21, 2013 at 08:26:52AM -0700, H. Peter Anvin wrote: > On 03/21/2013 08:19 AM, Michael S. Tsirkin wrote: > >> > >> In BIOS (e.g. SeaBIOS), using MMIO is very difficult, so your boot ROM > >> probably wants to use I/O space even if MMIO is available. > > > > Is this a real concern? Modern cards seem to supply PXE ROMs even > > though they have no IO BARs. > > > > Most of them do really ugly hacks in hardware (like putting in a "back > door" in config space) to make that possible. config space register that let us access registers within BAR actually sounds pretty reasonable. Way better than an I/O BAR. > > > > Problem is, BIOS and OS normally assume failure to allocate > > any resources means card won't function and disable it. > > So it does not seem to be worth it to have such a > > device specific failover ability. > > > > That is a violation of the PCIe spec; the PCIe spec specifically states > that failure to allocate an I/O BAR should still allow the device to > function. Where does it say this? Also, if as you say BIOS is not prepared to handle MMIO, and the device is needed for boot, where does this leave us? > So we shouldn't rule it out going forward. > > -hpa I'm not against this as such I just think we need some other solution for BIOS. > -- > H. Peter Anvin, Intel Open Source Technology Center > I work for Intel. I don't speak on their behalf.