On Wed, 2017-02-22 at 10:14 -0500, Boris Ostrovsky wrote:
> On 02/22/2017 09:28 AM, Bjorn Helgaas wrote:
> > 
> > On Tue, Feb 21, 2017 at 10:58:39AM -0500, Boris Ostrovsky wrote:
> > > 
> > > On 02/21/2017 10:45 AM, Juergen Gross wrote:
> > > > 
> > > > On 21/02/17 16:31, Dan Streetman wrote:
> > > > > 
> > > > > On Fri, Jan 13, 2017 at 5:30 PM, Konrad Rzeszutek Wilk
> > > > > <konrad.wilk@oracle.com> wrote:
> > > > > > 
> > > > > > On Fri, Jan 13, 2017 at 03:07:51PM -0500, Dan Streetman wrote:
> > > > > > > 
> > > > > > > Revert the main part of commit:
> > > > > > > af42b8d12f8a ("xen: fix MSI setup and teardown for PV on HVM guests")
> > > > > > > 
> > > > > > > That commit introduced reading the pci device's msi message data to see
> > > > > > > if a pirq was previously configured for the device's msi/msix, and re-use
> > > > > > > that pirq.  At the time, that was the correct behavior.  However, a
> > > > > > > later change to Qemu caused it to call into the Xen hypervisor to unmap
> > > > > > > all pirqs for a pci device, when the pci device disables its MSI/MSIX
> > > > > > > vectors; specifically the Qemu commit:
> > > > > > > c976437c7dba9c7444fb41df45468968aaa326ad
> > > > > > > ("qemu-xen: free all the pirqs for msi/msix when driver unload")
> > > > > > > 
> > > > > > > Once Qemu added this pirq unmapping, it was no longer correct for the
> > > > > > > kernel to re-use the pirq number cached in the pci device msi message
> > > > > > > data.  All Qemu releases since 2.1.0 contain the patch that unmaps the
> > > > > > > pirqs when the pci device disables its MSI/MSIX vectors.
> > > > > > > 
> > > > > > > This bug is causing failures to initialize multiple NVMe controllers
> > > > > > > under Xen, because the NVMe driver sets up a single MSIX vector for
> > > > > > > each controller (concurrently), and then after using that to talk to
> > > > > > > the controller for some configuration data, it disables the single MSIX
> > > > > > > vector and re-configures all the MSIX vectors it needs.  So the MSIX
> > > > > > > setup code tries to re-use the cached pirq from the first vector
> > > > > > > for each controller, but the hypervisor has already given away that
> > > > > > > pirq to another controller, and its initialization fails.
> > > > > > > 
> > > > > > > This is discussed in more detail at:
> > > > > > > https://lists.xen.org/archives/html/xen-devel/2017-01/msg00447.html
> > > > > > > 
> > > > > > > Fixes: af42b8d12f8a ("xen: fix MSI setup and teardown for PV on HVM guests")
> > > > > > > Signed-off-by: Dan Streetman <dan.streetman@canonical.com>
> > > > > > Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
> > > > > This doesn't seem to be applied yet, is it still waiting on another
> > > > > ack?  Or maybe I'm looking at the wrong git tree...
> > > > Am I wrong or shouldn't this go through the PCI tree? Konrad?
> > > Konrad is away this week but since pull request for Xen tree just went
> > > out we should probably wait until rc1 anyway (unless something big comes
> > > up before that).
> > I assume this should go via the Xen or x86 tree, since that's how most
> > arch/x86/pci/xen.c patches have been handled, including af42b8d12f8a.
> > If you think otherwise, let me know.
>
> OK, I applied it to Xen tree's for-linus-4.11.

Hm, we want this (c74fd80f2f4) in stable too, don't we?