Re: Regression PCI passthrough from 4.5.5 to 4.6.0-rc1

From: "Roger Pau Monné" <roger.pau@citrix.com>
To: Jan Beulich <JBeulich@suse.com>
Cc: Andreas Kinzler <hfp@posteo.de>, xen-devel@lists.xen.org
Subject: Re: Regression PCI passthrough from 4.5.5 to 4.6.0-rc1
Date: Mon, 14 Aug 2017 12:56:58 +0100	[thread overview]
Message-ID: <20170814115658.qcwk42lyx4ihsvlb@dhcp-3-128.uk.xensource.com> (raw)
In-Reply-To: <5991A920020000780016F550@prv-mh.provo.novell.com>

On Mon, Aug 14, 2017 at 05:44:00AM -0600, Jan Beulich wrote:
> >>> On 14.08.17 at 13:31, <hfp@posteo.de> wrote:
> > On Mon, 31 Jul 2017 12:12:45 +0200, Jan Beulich <jbeulich@suse.com> wrote:
> >>>>> "Andreas Kinzler" <ml-ak@hfp.de> 07/17/17 6:32 PM >>>
> >>>>> Jan, I still have access to the hardware so perhaps we can finally  
> >>>>> solve
> >>>>> this problem.
> >>>> Feel free to go ahead; I'll be on vacation for the next three weeks.
> >>> Perhaps we can shortcut debugging a bit because I looked through the
> >>> patches of XenServer 7.2 and found the attached patch. Now I tried it  
> >>> and
> >>> it seems to solve all the problems. Does that patch look good to you,  
> >>> too?
> >> Iirc the patch had even been submitted once, and rejected as being not
> >> generally correct (i.e. it cures a symptom rather than the cause). What
> >> we'd need to know is the order of actions the guest takes which ought to
> >> result in the vector getting unmasked, but doesn't in reality.
> > 
> > I defined XEN_PT_LOGGING_ENABLED in xen_pt.h as requested without the  
> > "hack" patch. Log is attached. Does it help?
> 
> It tells me that there's nothing unexpected on that side. As I think I
> had indicated before, we really need to see both sides (qemu and
> hypervisor), as part of the MSI-X handling lives in Xen. And for the
> hypervisor side it is unlikely that we'll be able to get away without
> a debugging patch. I am intending to make such available to you in
> case you can't do so yourself, but I can't currently predict when I'll
> get to it.

I think the problem is that pci_msi_conf_write_intercept is failing to
unmask the entries when MSI-X is enabled with entries already
configured, but this will require some debugging patch as Jan said.

Following the MSI-X code is quite complicated, this split brain
between Xen and QEMU makes it quite hard. I can try to come up with a
patch later.

Roger.

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