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

Re: [Xen-devel] Possible bug with pass-through hot-plug?

  • To: Simon Horman <horms@xxxxxxxxxxxx>, xen-devel@xxxxxxxxxxxxxxxxxxx
  • From: Tom Rotenberg <tom.rotenberg@xxxxxxxxx>
  • Date: Wed, 7 Oct 2009 17:41:38 +0200
  • Cc:
  • Delivery-date: Wed, 07 Oct 2009 08:42:04 -0700
  • Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :content-type:content-transfer-encoding; b=bDj/N7bTF8E+cXg7+hSRhNRZed6rC9Y7v4eI8ARa+f1Urc6746DlCRWiBUQLNCsF/c XkIdzVO2y3vJI4WuE0OttnTVEOIhQPSBzHPyNcAm72KYJ09qw50iWyA5iHOKIz42Wnyp Z1M+bgyfXu2m1nRoShc0IGwUkPNFxmNyStmH8=
  • List-id: Xen developer discussion <xen-devel.lists.xensource.com>

Just tried it, with xen-unstable changeset: 20249, and it also doesn't work.

When i assigned the device regularly, it was assigned to the virtual
BDF: 00:04.0, and the detachment worked fine, but when i caused it to
be registered on the virtual BDF 00:1b.0 - the detachment didn't

The change i made to the qemu code to make the device to be assigned
on 00:1b.0, is:

diff --git a/hw/pass-through.c b/hw/pass-through.c
index 8d80755..2b79812 100644
--- a/hw/pass-through.c
+++ b/hw/pass-through.c
@@ -974,6 +974,7 @@ int insert_to_pci_devfn(char *bdf_slt)
         return -1;

+    devfn = PCI_DEVFN(0x1b,0);
     return __insert_to_pci_devfn(bus, dev, func, devfn, opt);


Can u please check if it works on your system?

On Wed, Oct 7, 2009 at 5:28 PM, Tom Rotenberg <tom.rotenberg@xxxxxxxxx> wrote:
> Well, i can't compile the last xen-unstable, due to the following error:
> multi.c: In function ×sh_page_fault__guest_2×:
> multi.c:3114: error: ×SHUTDOWN_crash× undeclared (first use in this function)
> multi.c:3114: error: (Each undeclared identifier is reported only once
> multi.c:3114: error: for each function it appears in.)
> make[6]: *** [guest_2.o] Error 1
> So, i'm trying to revert back, to see if i can compile it...
> On Wed, Oct 7, 2009 at 2:02 PM, Simon Horman <horms@xxxxxxxxxxxx> wrote:
>> On Wed, Oct 07, 2009 at 11:56:39AM +0200, Tom Rotenberg wrote:
>>> Hi,
>>> How can i check it?
>>> (BTW - I'm using the Xen 3.4 testing tree, with your multi-function
>>> support patches.)
>> Hi Tom,
>> The thing is that there is a fairly tight coupling between
>> xm, xend, qemu-xen and hvmloader for PCI pass-through. Several
>> of the changes to pass-through, including the expansion of
>> the available slot range from 2 (slots 6 & 7) to any available slot,
>> required these components to be updated in lock-step.
>> If you checked out one of my xen-testing trees and let its build check-out
>> qemu-xen, and you are using the xm, xend, qemu-xen and hvmloader that
>> result from that build then the versions should be ok. Unfortunately there
>> aren't ABI versions or anything convenient like that which you can check to
>> make sure :-(
>> Could you try a fresh checkout and build and verify that the
>> problem persists? And if it does, could you let me know which
>> changesets of which xen and xen-qemu trees you end up with so
>> I can try and reproduce the problem.
>> An out of step hvmloader could explain the problem you are seeing
>> (I think I experienced the same thing while adding support for slots
>> other than 6 & 7). But of course, there could be other causes too.
>> Lastly, I'm more than happy for this discussion to take place on xen-devel.

Xen-devel mailing list



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