Re: [Xen-devel] Re: [Xen-users] Passthrough support ?

[Mark Williamson <mark.williamson@xxxxxxxxxxxx>]

>  We also mentioned about xen_pci_sharedinfo in previous mails
>  Well As I read in the linux code , xen_pci_sharedinfo contains xen_pci_op
> where front end specifies which operation is to be done on which pci
> device. And then it keeps checking the status of this device through
> another flag in xen_pci_sharedinfo. I infer here that dom0's drivers are
> being used.

Yes, but not dom0's drivers /for that device/.  Config space ops submitted 
through that interface will use dom0's drivers for the PCI /bus/, but that's 
independent of the /specific device/ being used.

Remember there's two levels in a PCI subsystem: the driver for the bus (which 
will tell you what's plugged into the bus and how to configure it) and the 
driver for the device (which relies on the driver for the bus to help it talk 
to the device)

pcifront is a PCI bus driver.  The PCI core and the PCI device drivers in the 
domU talk to it in order to do some of their work.  pciback is a service that 
runs in dom0 and lets pcifront instances exchange information with the /real/ 
PCI bus driver.

> Can you point me towards a code path where domU actually uses 
> its own drivers having owned the PCI device ?

pcifront uses pciback to query what PCI devices are available.  pcifront 
appears to the domU as if it's a real PCI bus, but actually it's talking to 
the /real/ host PCI bus via bciback.  Config space operations go through the 
PCI bus driver, so in the domU any config space operations go through 
pcifront, to pciback and from there to the real hardware (if appropriate).

Once the domU has got the details of which devices are available (this process 
is done by Linux's generic PCI probing code, based on information it gets 
from pcifront) it'll load the driver for the device it's been given.  This is 
basically just the standard Linux device driver - same as you'd use on native 
Linux - buy compiled against the XenLinux kernel.

The device driver may do a number of things:

1) it might access PCI config space.  It does this through a generic kernel 
API, and the requests are fed into pcifront and satisfied by pciback.  The 
device driver code is the same path as would be run on a native install, 
except that it's serviced by pcifront not a real PCI bus.

2) it might want to access IO memory regions or IO ports.  It can do this 
directly using the normal APIs (e.g. inb / outb for port IO) etc that Linux 
exports to drivers.  These work *directly*, without going through dom0.

3) it might do DMA to the device.  This will likely be set up using commands 
written to IO memory regions / ports.  It uses standard Linux APIs to set up 
the DMA-able memory; XenLinux includes Xen-aware functions to make sure the 
memory addressing works right (this is the same as in dom0).  The normal 
driver path can be followed here without dom0 interaction.

If the device receives interrupts, I believe they currently come direct from 
Xen without dom0 being involved.

Does that help clarify things?

Cheers,
Mark

> Mark Williamson wrote:
> >> 1. When PCI Passthrough support  'is not enabled' , how does domU access
> >> PCI device ?
> >>     I suppose they continue to communicate using pcifront/back split
> >> drivers and now dom0's
> >>     drivers are used ( right ? )
> >
> > If you don't have PCI passthrough (as provided by pcifront/back) then the
> > domU can't really have direct access to a PCI device.  The only accesses
> > it'll be able to make are by using the virtual block / net interfaces to
> > request dom0 do some IO for it.
> >
> >> 2. PCI-passthrough enables domU to own the PCI device. Now dom0 can no
> >> longer use that device .
> >
> > Yep.
> >
> >>     If domU has mapped address space of the required  PCI device , why
> >> need it talk to dom0 any  further ?
> >
> > DomU owns the PCI device, but dom0 owns the PCI *bus* hardware.  To query
> > (and potentially set) information in "PCI configuration space", the domU
> > will have to talk to dom0 somehow, because that information comes from
> > the PCI bus hardware, not from the device itself.  The PCI config space
> > information includes the location of the mmio and port io regions used by
> > the device, so without this communication the domU wouldn't know how to
> > talk to its device.
> >
> > Come to that, the PCI config space information is needed to determine
> > what the device *is* so the domU knows what driver to use to talk to it
> > ;-)
> >
> > The PCI bus control hardware can only be owned by dom0, so it is
> > necessary to indirect these operations through dom0.  pciback doesn't
> > necessarily pass through config space data unmodified to pcifront (there
> > used to be a number of different options on how it passed stuff through)
> > and it doesn't necessarily allow pcifront to alter config space (for
> > safety reasons).  But in the end it provides enough of the functionality
> > of PCI config space to enable the domU to find and identify the PCI
> > device(s) it owns, and load the appropriate drivers.
> >
> >>     What kind of hand shake is involved ( setup and tear down as you
> >> have mentioned ) between domU and
> >>     domU  " in passthrough case" ?
> >
> > Well, the operations on PCI config space will go through dom0.  That's
> > the main thing I can think of right now, but there may be other
> > operations too.
> >
> >> 3. DomU discovers all PCI devices through  Xenstore.
> >
> > I'm not actually sure on this point, but you should be able to figure it
> > out with a scan through the code.  Post here if you have trouble
> > understanding sections!
> >
> >>>  requires to talk to its assigned device(s).
> >>
> >> How are devices "assigned"  to domU ?  I am specifically talking about
> >> late binding.
> >
> > pciback needs to "claim" the device in dom0 so that no "real" device
> > drivers in dom0 will try to use it.  We don't want the domains fighting
> > over it ;-) After dom0 boot time, you can move devices to the control of
> > pciback by writing values to sysfs to tell any current driver to release
> > its hold on the device, then tell pciback to grab it.
> >
> > Once pciback has grabbed a device, the rest of dom0 Linux doesn't know
> > that a domU is using the device.  As far as Linux is concerned, the
> > device is owned by pciback.
> >
> > To give the domain access to the device, dom0 needs to issue some
> > hypercalls giving the domU the rights to map the appropriate mmio
> > regions, and access the appropriate IO regions.  It'll also need to have
> > setup a connection between pcifront and pciback.  I guess it might stick
> > something in Xenstore - maybe you can see what happens in the code?  Some
> > of this will be done in pciback, some is probably administered in the
> > tools code.
> >
> > I'm sorry to be a bit vague.  It's a while since I looked at this code,
> > so please bare in mind that I'm a little rusty ;-)
> >
> > Cheers,
> > Mark
> >
> >>>  Having
> >>> obtained this information, communication with the device is possible
> >>> directly using IO ports, memory IO regions, and DMA.
> >>>
> >>>> 2. But, in case of emulation drivers of dom0 are used  where as in
> >>>> case of passthrough ( as the name suggests ) native drivers in domU
> >>>> are used .
> >>>
> >>> For true emulation (qemu device model), a userspace process in dom0
> >>> handles modelling a "real" device and then issues IO using normal
> >>> userspace APIs. These get serviced by the dom0 kernel using the normal
> >>> device driver.
> >>>
> >>> For PV drivers, the frontend driver in the domU kernel issues requests
> >>> which are picked up by the backend driver in dom0's kernel, which then
> >>> issues requests into the IO stack.  Again this uses the normal device
> >>> driver in dom0 to talk to the actual device, it's just that the request
> >>> is made using a kernel-internal API rather than a userspace API (which
> >>> results in slightly different actions being taken).
> >>>
> >>>> 3. dom0 provides a virtual PCI device { an interface for device-OPs
> >>>> and status  of this virtual device} to domU and  through associated
> >>>> event channel domU makes
> >>>>    "synchronous" use of this device.
> >>>
> >>> domU uses this for control plane operations, but for most work it can
> >>> talk to its PCI device directly without going through dom0.
> >>>
> >>>> ===
> >>>> Queries:
> >>>>
> >>>> 1. What i am really not so sure about is ... passthrough case
> >>>>     Will there be requirement to map the address space of this PCI
> >>>> device in domU ?  Will the page which was being shared  so-far
> >>>> {xen_pci_sharedinfo}
> >>>>      for emulation , be "flipped"  ( transferred ) into domU ?
> >>>
> >>> xen_pci_sharedinfo - is that the page used to talk to the PCI backend
> >>> from pcifront?  If so, then no, that's just used for dom0-domU
> >>> communications.
> >>
> >> Well As I read in the xen0linux code , xen_pci_sharedinfo contains
> >> xen_pci_op where front end specifies which
> >> operation is to be done on which pci device. And then it keeps checking
> >> the status of this device through another flag in
> >> xen_pci_sharedinfo. I infer here that dom0's drivers are being used. Can
> >> you point me towards a code path where domU actually
> >> uses its own drivers having owned the PCI device ?
> >>
> >>> In order to map the address space of the PCI device directly, the domU
> >>> is given permissions to map the IO memory regions of that device into
> >>> it's page tables.  I think this is now possible to do using a grant
> >>> table operation...
> >>>
> >>> It is also given permission to access certain IO port ranges so that it
> >>> can use the device's port IO interfaces.
> >>>
> >>>>      2. Well ,
> >>>> Having read the code for linux (dom0,domU) I see that there are split
> >>>> device  drivers for PCI. (pci front and PCIback). Which are normally
> >>>>    communicating over xenbus.  which looks almost like other split
> >>>> drivers.  How exactly then passthrough enables use of domU's drivers ?
> >>>
> >>> The key thing to understand is that the pcifront / pciback is basically
> >>> just used for setup and teardown, not for the actual IO.  The real IO
> >>> is done directly by the domU without going through dom0.  For the block
> >>> and net drivers, *all* IO goes through dom0.
> >>>
> >>>> 3. And if passthrough support isnt provided how will communication
> >>>> between pcifront-pciback  be different ? ( I guess netbsd , freebsd do
> >>>> not have passthrough support yet )
> >>>
> >>> I'm not entirely clear what you're asking here, but I'll take a stab at
> >>> it:
> >>>
> >>> if pcifront (in domU) and pciback (in dom0) aren't available then
> >>> passthrough won't work.  The dom0 has to support the backend functions
> >>> of PCI passthrough and the domU has to know how to talk to it.  It's
> >>> also implicit that they're using the same interface version to talk to
> >>> each other - I'm not sure whether that's frozen stable or not.
> >>>
> >>> So *if* NetBSD lacks pciback support, it can't pass PCI devices to
> >>> guests that do.  Similarly, *if* it lacks pcifront support, it can't
> >>> have devices passed to it.
> >>>
> >>>> 4. What  restricts other domUs from accessing PCI device given to
> >>>> other domU via passrthrough support.
> >>>
> >>> There are some restrictions on what can be done in PCI config space to
> >>> prevent a guest fouling things up.  These need to be relaxed for some
> >>> awkward devices, though.
> >>>
> >>> For the device IO itself, domUs are only allowed to map mmio regions
> >>> and access io ports that are relevant to their device.  It's possible
> >>> for these to overlap with those for other devices, in which case you're
> >>> trusting the domU to be well behaved.  More crucially, though, giving a
> >>> domain a device with DMA capabilities is equivalent to giving it the
> >>> ability to subvert the entire machine.  DMA can't be sandboxed on most
> >>> current hardware, so if you give DMA rights to a VM it's automatically
> >>> just as trusted as dom0 with respect to not fooling about with other
> >>> domains, hardware, etc.
> >>>
> >>> Cheers,
> >>> Mark
> >>
> >> Thanks,
> >> Sanket
> >>
> >> _______________________________________________
> >> Xen-devel mailing list
> >> Xen-devel@xxxxxxxxxxxxxxxxxxx
> >> http://lists.xensource.com/xen-devel



-- 
Dave: Just a question. What use is a unicyle with no seat?  And no pedals!
Mark: To answer a question with a question: What use is a skateboard?
Dave: Skateboards have wheels.
Mark: My wheel has a wheel!

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users