[PATCH] Fix qemu traditional with PCI passthrough.

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

Hey Ian and Stefano,

This patch was posted way back last year in July 
(see http://lists.xen.org/archives/html/xen-devel/2013-07/msg00004.html) and
was mentioned to:

">From what I understand following the conversation, I think this is
> probably the right way to solve the problem, but given that it's only
> really a problem when you load and unload drivers, which is the
> uncommon case, I think at this point we should probably hold off on
> this one until 4.3.1.
> 
> Stefano, thoughts?

I think that's OK. I'll wait to apply the qemu-xen patch until after the
release."

I think this patch just got lost in the Xen 4.4 release. Dusting it
off and reposting.

The issue at hand is simple - you boot an PVHVM guest with a PCI
passthrough device and in a loop do:

#!/bin/bash

while (true)
do
 rmmod igbvf
 killall dhclient
 modprobe igbvf
 dhclient eth1
done


and you find yourself in distressed to see that after a while it
cannot allocate any IRQs. I've tested it and it fixes the issue.


Now there was also an qemu-xen version of this patch posted:
http://lists.xen.org/archives/html/xen-devel/2013-07/msg00008.html

and I just reposted it, see:
http://mid.gmane.org/1396975053-16435-1-git-send-email-konrad@kernel.org
(or "[PATCH] Fix qemu-xen with PCI passthrough.")


 hw/pass-through.c | 8 +++++++-
 hw/pt-msi.c       | 5 +++--
 2 files changed, 10 insertions(+), 3 deletions(-)

Zhenzhong Duan (1):
      qemu-xen-trad: free all the pirqs for msi/msix when driver unloads

Thanks!