Re: [Xen-devel] PCI Passthrough bug with x86 HVM

[Stefano Stabellini <sstabellini@xxxxxxxxxx>]

On Wed, 26 Jun 2019, Stefano Stabellini wrote:
> On Wed, 26 Jun 2019, Juergen Gross wrote:
> > On 26.06.19 14:21, Chao Gao wrote:
> > > On Wed, Jun 26, 2019 at 08:17:50AM +0200, Juergen Gross wrote:
> > > > On 24.06.19 20:47, Stefano Stabellini wrote:
> > > > > + xen-devel
> > > > > 
> > > > > On Mon, 24 Jun 2019, Stefano Stabellini wrote:
> > > > > > Hi all,
> > > > > > 
> > > > > > I might have found a bug with PCI passthrough to a Linux HVM guest 
> > > > > > on
> > > > > > x86 with Xen 4.12. It is not easy for me to get access, and 
> > > > > > especially
> > > > > > change components, on this particular system, and I don't have 
> > > > > > access
> > > > > > to
> > > > > > other x86 boxes at the moment, so apologies for the partial
> > > > > > information
> > > > > > report. The setup is as follow:
> > > > > > 
> > > > > > - two PCI devices have been assigned to a HVM guest, everything is
> > > > > > fine
> > > > > > - reboot the guest from inside, i.e. `reboot' in Linux
> > > > > > - after the reboot completes, only one device is assigned
> > > > > > 
> > > > > > Before the reboot, I see all the appropriate xenstore entries for 
> > > > > > both
> > > > > > devices. Everything is fine. After the reboot, I can only see the
> > > > > > xenstore entries of one device. It is as if the other device
> > > > > > "disappeared" without throwing any errors.
> > > > > > 
> > > > > > Have you seen this before? Do you know if it has been fixed in
> > > > > > staging?
> > > > > > I noticed this fix which seems to be very relevant:
> > > > > > 
> > > > > > https://lists.xenproject.org/archives/html/xen-devel/2018-11/msg01616.html
> > > > > > 
> > > > > > but it is already included in 4.12.
> > > > 
> > > > Stefano, could you please try the attached patch? It is only compile
> > > > tested for now.
> > > > 
> > > > 
> > > > Juergen
> > > 
> > > > From ea95dcdfc60a895cc43baf34c8e0fb088e10008d Mon Sep 17 00:00:00 2001
> > > > From: Juergen Gross <jgross@xxxxxxxx>
> > > > To: xen-devel@xxxxxxxxxxxxxxxxxxxx
> > > > Cc: Ian Jackson <ian.jackson@xxxxxxxxxxxxx>
> > > > Cc: Wei Liu <wl@xxxxxxx>
> > > > Date: Wed, 26 Jun 2019 08:15:28 +0200
> > > > Subject: [PATCH] libxl: fix pci device re-assigning after domain reboot
> > > > 
> > > > After a reboot of a guest only the first pci device configuration will
> > > > be retrieved from Xenstore resulting in loss of any further assigned
> > > > passed through pci devices.
> > > > 
> > > > The main reason is that all passed through pci devices reside under a
> > > > common root device "0" in Xenstore. So when the device list is rebuilt
> > > > from Xenstore after a reboot the sub-devices below that root device
> > > > need to be selected instead of using the root device number as a
> > > > selector.
> > > > 
> > > > Fix that by adding a new member to struct libxl_device_type which when
> > > > set is used to get the number of devices. Add such a member for pci to
> > > > get the correct number of pci devices instead of implying it from the
> > > > number of pci root devices (which will always be 1).
> > > > 
> > > > While at it fix the type of libxl__device_pci_from_xs_be() to match
> > > > the one of the .from_xenstore member of struct libxl_device_type. This
> > > > fixes a latent bug checking the return value of a function returning
> > > > void.
> > > > 
> > > > Signed-off-by: Juergen Gross <jgross@xxxxxxxx>
> > > 
> > > Tested-by: Chao Gao <chao.gao@xxxxxxxxx>
> > 
> > Thanks!
> 
> Thank you very much both of you! I'll let you know if it works.

Tested-by: Stefano Stabellini <sstabellini@xxxxxxxxxx>

Let's get it in the tree, thanks!

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel