Re: [Xen-devel] Q77 IGD instantly crashes on xen-pciback bind.

[Sander Eikelenboom <linux@xxxxxxxxxxxxxx>]

Thursday, November 27, 2014, 11:23:24 AM, you wrote:

> On Nov 24,  1:28pm, Pasi =?iso-8859-1?Q?K=E4rkk=E4inen?= wrote:

>> Hello,

> Hi, hope the week is going well for everyone.

>> > >> As I was walking out the door I remembered I had been delinquent
>> > >> with information.  The dom0 kernel is 32-bit 3.14.22 straight from
>> > >> kernel.org under a 64-bit hypervisor compiled from 4.4.1 sources.
>> > 
>> > > Wow, quite an old thread :)
>> > >  
>> > > So you're still seeing the same problem with recent Xen/Linux
>> > > versions.. 
>> > 
>> > Yes, the perils of platforming for 7 year field deployments... :-)
>> > 
>> > I can certainly build up a toolchain against the HEAD of XEN git and
>> > the most recent release of the kernel if everyone feels that would be
>> > beneficial.
>> > 
>> > > This might be a stupid question, but here goes anyway: Do you have
>> > > serial console set up? And all the debug/verbose options specified
>> > > for Xen and Linux?
>> > 
>> > The platform in question doesn't have any serial ports, at least not
>> > surfaced.  We will need to do a bit of wiring if we need to go in that
>> > direction.

>> You mentioned it's Intel Q77 chipset based motherboard..  which
>> means it should have Intel AMT functionality, which provides SOL
>> (Serial-over-LAN), which you can use as a serial console for Xen.
>>
>> There are tools (at least amtterm) that you can use on another box
>> to connect to the AMT SOL remotely..

> So we wired up serial console connectivity to the test box and
> repeated the VGA device binding with loglvl=all.  We lost the box
> immediately without anything being written to the logs.

> So we went hunting.

> Interestingly the problem appears to be secondary to a BIOS
> configuration option.  This may be specific to this platform but we
> wanted to get it documented in the thread in case anyone else runs
> into this.

> The DQ77KB BIOS we are using has an option for 'IGD flat panel
> display'.  The default option is LVDS, setting this to 'disabled'
> clears the problem.

> I haven't run down where things go wrong in pci_stub but I assume it
> does something to the hardware which causes a problem when the video
> controller is reset and then shutdown.

>> > Now that I have the machine in a harness in the lab I will stick a
>> > '#define DEBUG 1' in the top of drivers/xen/xen-pciback/pci_stub.c
>> > since that is where the action seems to be going on.
>> > 
>> > The platform is headed for a measured computing environment so I
>> > thought there may be some type of conflict with tboot holding a
>> > reference to the VGA driver but I verified the issue in a straight
>> > hypervisor boot.
>> > 
>> > I see that Tiejun Chen from Intel is sorting out issues with respect
>> > to the need to export the ISA bridge into the device emulator in order
>> > to support passthrough on these IGD devices.  I bound the 00:1f.0 ISA
>> > bridge device to pciback and that worked but it did not change the
>> > behavior of the regression.  When the 00:02.0 device is bound to
>> > pciback the display is cleared and the machine dies in its tracks.

>> Yeah, Tiejun is working on upstreaming the IGD passthru patches to
>> Qemu-upstream.
>>
>> Qemu-dm-traditional already has (most of) the IGD passthru patches. 
>> 
>> Hope that helps,

> So we are obviously working with qemu-dm-traditional and with the
> IGD/LVDS BIOS configuration issue fixed the adapater passthrough is
> working and Windows7 is coming up and detecting the IGD as a standard
> VGA display adapter.  Additional invocations of the VM after the first
> one result in failed passthrough with a garbled display.

This is probably due to the current lack of slot/bus reset in xen-pciback,
Konrad has a preliminary kernel patch for xen-pciback that does this.
I have attached the patch, though it has some rough edges in the design :-)
 
I'm currently running with his 3.19 xen-pciback patches series + the preliminary
patch for slot/bus reset and rebooting a guest with vga/pci passthrough now
works. (i'm running with a radeon card, passed through as a secondary card
to the emulated qemu one, in a linux guest using qemu-xen, so i can't help
you with your other questions and problems).

--
Sander

> I spent an afternoon wandering through the mailing lists and found
> what I think are the two patches which are needed to map the 00:1f.0
> ISA bridge device into the guest.  From the discussions surrounding
> those patches it appears as if the Windows HD driver needs addresses
> managed by that bridge to recognize the IGD device.

> I will get those patches wired into qemu-dm-traditional and tested in
> between whisky, wine, turkey and napping today.... :-)

> I'm hoping that this positively impacts the ability to execute
> multiple sessions.  I will report back the results so we have all of
> this in the mailing list record.

>> -- Pasi

> Thanks for offering the pointers, have a good day.

> Greg

> }-- End of excerpt from Pasi =?iso-8859-1?Q?K=E4rkk=E4inen?=

> As always,
> Dr. G.W. Wettstein, Ph.D.   Enjellic Systems Development, LLC.
> 4206 N. 19th Ave.           Specializing in information infra-structure
> Fargo, ND  58102            development.
> PH: 701-281-1686
> FAX: 701-281-3949           EMAIL: greg@xxxxxxxxxxxx
> ------------------------------------------------------------------------------
> "Immortality is an adequate definition of high availability for me."
>                                 -- Gregory F. Pfister

Attachment: 0001-xen-pciback-Implement-PCI-reset-slot-or-bus-with-do_.patch
Description: Binary data

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel