[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH 0 of 4] Add commands to automatically prep devices for pass-through

On Thu, 2012-05-10 at 15:12 +0100, Sander Eikelenboom wrote:
> Hello Ian,
> Thursday, May 10, 2012, 12:38:40 PM, you wrote:
> > On Thu, 2012-05-10 at 11:17 +0100, George Dunlap wrote:
> >> On Wed, May 9, 2012 at 2:45 PM, George Dunlap
> >> <george.dunlap@xxxxxxxxxxxxx> wrote:
> >> > On 09/05/12 12:59, Ian Campbell wrote:
> >> >>
> >> >> Right, however it is strictly speaking a new feature which is not
> >> >> mentioned on the TODO list and has not previously been posted (AFAIK,
> >> >> please correct me if not) and we are currently supposed to be in feature
> >> >> freeze (and have been for several weeks, if not a month).
> >> >>
> >> >> IIRC this functionality was mooted when the pci permissive patch was
> >> >> being done as something which would be a 4.3 feature.
> >> >> We need to decide if we want to make an exception for this new feature
> >> >> or not. Although I'm sure this feature is very nice and handy, we've
> >> >> lived without it for years and people seem to be able to use the
> >> >> existing scheme.
> >> 
> >> And, I realize that at some point all of the deadlines are going to be
> >> arbitrary; but I have always felt this is important enough to get an
> >> exception.  I consider having to muck about with sysfs to be basically
> >> a UI bug that really needs fixing.  I have a lot of other things that
> >> I would like to get done for the 4.2 release; but I thought this was
> >> important enough to get priority (above the PoD patch series, for
> >> instance).  NB I'm not saying that you should accept it because I
> >> worked on it; I only bring it up to demonstrate how important I think
> >> the feature is.
> > OK, given that the code is basically self contained and shouldn't effect
> > anything unless a user "opts-in" to using it I think you've convinced
> > me. Lets take this (I'll review the actual patches shortly).
> > BTW, IMHO it is preferable for actual deployments to use the kernel
> > command line options to hide devices rather than either this feature or
> > sysfs.
> > Fully hiding the device from dom0 drivers generally seems like it is
> > always better. That way the first driver to try and touch the hardware
> > is the one inside the domU. This avoids issues with dom0 drivers setting
> > stuff up but not tearing it down in a way that domU can cope with and
> > makes the use of hardware which doesn't support FLR more reliable etc.
> Haven't checked it (haven't got the time right now) but:
> Is using wildcards in the BDF's on the commandline supported already
> (like the ones supported in the config files for domains)

Based on a quick scan of the code it doesn't appear so, I don't maintain
PCI backthough so there might be something in the pipeline.

> I tend to have quite long commandlines to hide a pci-e card with 8
> functions (needed to specify 8 BDF's seperatly) for pci passthrough,
> would be nice if one could just specify 09:00.* for example.

It sounds useful to me.

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.