diff mbox

[v5] virtio-pci: enable bus master for old guests

Message ID 1410459530-29832-1-git-send-email-mst@redhat.com
State New
Headers show

Commit Message

Michael S. Tsirkin Sept. 11, 2014, 6:20 p.m. UTC 
commit cc943c36faa192cd4b32af8fe5edb31894017d35
    pci: Use bus master address space for delivering MSI/MSI-X messages
breaks virtio-net for rhel6.[56] x86 guests because they don't
enable bus mastering for virtio PCI devices. For the same reason,
rhel6.[56] ppc64 guests cannot boot on a virtio-blk disk anymore.

Old guests forgot to enable bus mastering, enable it automatically on
DRIVER (guests use some devices before DRIVER_OK).

Reported-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
---
Sorry, v4 was broken - forgot to commit before git format-patch.
Here's the correct version.
Still not tested - will do next week.
Testing reports appreciated meanwhile.

 hw/virtio/virtio-pci.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

Comments

Greg Kurz Sept. 11, 2014, 6:42 p.m. UTC | #1 
On Thu, 11 Sep 2014 21:20:03 +0300
"Michael S. Tsirkin" <mst@redhat.com> wrote:

> commit cc943c36faa192cd4b32af8fe5edb31894017d35
>     pci: Use bus master address space for delivering MSI/MSI-X messages
> breaks virtio-net for rhel6.[56] x86 guests because they don't
> enable bus mastering for virtio PCI devices. For the same reason,
> rhel6.[56] ppc64 guests cannot boot on a virtio-blk disk anymore.
> 
> Old guests forgot to enable bus mastering, enable it automatically on
> DRIVER (guests use some devices before DRIVER_OK).
> 
> Reported-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> ---
> Sorry, v4 was broken - forgot to commit before git format-patch.
> Here's the correct version.
> Still not tested - will do next week.
> Testing reports appreciated meanwhile.
> 

Nice patch indeed ! :)

Reviewed-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Tested-by: Greg Kurz <gkurz@linux.vnet.ibm.com>

Cc'ing qemu-stable as Michael said he would queue the initial
bus master patch for 2.1.2.

>  hw/virtio/virtio-pci.c | 10 ++++++++++
>  1 file changed, 10 insertions(+)
> 
> diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
> index ddb5da1..a827cd4 100644
> --- a/hw/virtio/virtio-pci.c
> +++ b/hw/virtio/virtio-pci.c
> @@ -314,6 +314,16 @@ static void virtio_ioport_write(void *opaque, uint32_t addr, uint32_t val)
>              msix_unuse_all_vectors(&proxy->pci_dev);
>          }
> 
> +        /* Linux before 2.6.34 drives the device without enabling
> +           the PCI device bus master bit. Enable it automatically
> +           for the guest. This is a PCI spec violation but so is
> +           initiating DMA with bus master bit clear. */
> +        if (val == (VIRTIO_CONFIG_S_ACKNOWLEDGE | VIRTIO_CONFIG_S_DRIVER)) {
> +            pci_default_write_config(&proxy->pci_dev, PCI_COMMAND,
> +                                     proxy->pci_dev.config[PCI_COMMAND] |
> +                                     PCI_COMMAND_MASTER, 1);
> +        }
> +
>          /* Linux before 2.6.34 sets the device as OK without enabling
>             the PCI device bus master bit. In this case we need to disable
>             some safety checks. */
Michael S. Tsirkin Sept. 14, 2014, 10:44 a.m. UTC | #2 
On Thu, Sep 11, 2014 at 08:42:06PM +0200, Greg Kurz wrote:
> On Thu, 11 Sep 2014 21:20:03 +0300
> "Michael S. Tsirkin" <mst@redhat.com> wrote:
> 
> > commit cc943c36faa192cd4b32af8fe5edb31894017d35
> >     pci: Use bus master address space for delivering MSI/MSI-X messages
> > breaks virtio-net for rhel6.[56] x86 guests because they don't
> > enable bus mastering for virtio PCI devices. For the same reason,
> > rhel6.[56] ppc64 guests cannot boot on a virtio-blk disk anymore.
> > 
> > Old guests forgot to enable bus mastering, enable it automatically on
> > DRIVER (guests use some devices before DRIVER_OK).
> > 
> > Reported-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
> > Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> > ---
> > Sorry, v4 was broken - forgot to commit before git format-patch.
> > Here's the correct version.
> > Still not tested - will do next week.
> > Testing reports appreciated meanwhile.
> > 
> 
> Nice patch indeed ! :)
> 
> Reviewed-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
> Tested-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
> 
> Cc'ing qemu-stable as Michael said he would queue the initial
> bus master patch for 2.1.2.

In my opinion it's best to keep the initial patch off stable.


> >  hw/virtio/virtio-pci.c | 10 ++++++++++
> >  1 file changed, 10 insertions(+)
> > 
> > diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
> > index ddb5da1..a827cd4 100644
> > --- a/hw/virtio/virtio-pci.c
> > +++ b/hw/virtio/virtio-pci.c
> > @@ -314,6 +314,16 @@ static void virtio_ioport_write(void *opaque, uint32_t addr, uint32_t val)
> >              msix_unuse_all_vectors(&proxy->pci_dev);
> >          }
> > 
> > +        /* Linux before 2.6.34 drives the device without enabling
> > +           the PCI device bus master bit. Enable it automatically
> > +           for the guest. This is a PCI spec violation but so is
> > +           initiating DMA with bus master bit clear. */
> > +        if (val == (VIRTIO_CONFIG_S_ACKNOWLEDGE | VIRTIO_CONFIG_S_DRIVER)) {
> > +            pci_default_write_config(&proxy->pci_dev, PCI_COMMAND,
> > +                                     proxy->pci_dev.config[PCI_COMMAND] |
> > +                                     PCI_COMMAND_MASTER, 1);
> > +        }
> > +
> >          /* Linux before 2.6.34 sets the device as OK without enabling
> >             the PCI device bus master bit. In this case we need to disable
> >             some safety checks. */
> 
> 
> 
> -- 
> Gregory Kurz                                     kurzgreg@fr.ibm.com
>                                                  gkurz@linux.vnet.ibm.com
> Software Engineer @ IBM/Meiosys                  http://www.ibm.com
> Tel +33 (0)562 165 496
> 
> "Anarchy is about taking complete responsibility for yourself."
>         Alan Moore.
Jason Wang Sept. 15, 2014, 9:48 a.m. UTC | #3 
On 09/12/2014 02:20 AM, Michael S. Tsirkin wrote:
> commit cc943c36faa192cd4b32af8fe5edb31894017d35
>     pci: Use bus master address space for delivering MSI/MSI-X messages
> breaks virtio-net for rhel6.[56] x86 guests because they don't
> enable bus mastering for virtio PCI devices. For the same reason,
> rhel6.[56] ppc64 guests cannot boot on a virtio-blk disk anymore.
>
> Old guests forgot to enable bus mastering, enable it automatically on
> DRIVER (guests use some devices before DRIVER_OK).
>
> Reported-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> ---
> Sorry, v4 was broken - forgot to commit before git format-patch.
> Here's the correct version.
> Still not tested - will do next week.
> Testing reports appreciated meanwhile.
>
>  hw/virtio/virtio-pci.c | 10 ++++++++++
>  1 file changed, 10 insertions(+)
>
> diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
> index ddb5da1..a827cd4 100644
> --- a/hw/virtio/virtio-pci.c
> +++ b/hw/virtio/virtio-pci.c
> @@ -314,6 +314,16 @@ static void virtio_ioport_write(void *opaque, uint32_t addr, uint32_t val)
>              msix_unuse_all_vectors(&proxy->pci_dev);
>          }
>  
> +        /* Linux before 2.6.34 drives the device without enabling
> +           the PCI device bus master bit. Enable it automatically
> +           for the guest. This is a PCI spec violation but so is
> +           initiating DMA with bus master bit clear. */
> +        if (val == (VIRTIO_CONFIG_S_ACKNOWLEDGE | VIRTIO_CONFIG_S_DRIVER)) {
> +            pci_default_write_config(&proxy->pci_dev, PCI_COMMAND,
> +                                     proxy->pci_dev.config[PCI_COMMAND] |
> +                                     PCI_COMMAND_MASTER, 1);
> +        }
> +
>          /* Linux before 2.6.34 sets the device as OK without enabling
>             the PCI device bus master bit. In this case we need to disable
>             some safety checks. */

So with this patch, we can also remove the detecting and handling of
VIRTIO_PCI_FLAG_BUS_MATER_BUG?
Michael S. Tsirkin Sept. 15, 2014, 9:59 a.m. UTC | #4 
On Mon, Sep 15, 2014 at 05:48:17PM +0800, Jason Wang wrote:
> On 09/12/2014 02:20 AM, Michael S. Tsirkin wrote:
> > commit cc943c36faa192cd4b32af8fe5edb31894017d35
> >     pci: Use bus master address space for delivering MSI/MSI-X messages
> > breaks virtio-net for rhel6.[56] x86 guests because they don't
> > enable bus mastering for virtio PCI devices. For the same reason,
> > rhel6.[56] ppc64 guests cannot boot on a virtio-blk disk anymore.
> >
> > Old guests forgot to enable bus mastering, enable it automatically on
> > DRIVER (guests use some devices before DRIVER_OK).
> >
> > Reported-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
> > Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> > ---
> > Sorry, v4 was broken - forgot to commit before git format-patch.
> > Here's the correct version.
> > Still not tested - will do next week.
> > Testing reports appreciated meanwhile.
> >
> >  hw/virtio/virtio-pci.c | 10 ++++++++++
> >  1 file changed, 10 insertions(+)
> >
> > diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
> > index ddb5da1..a827cd4 100644
> > --- a/hw/virtio/virtio-pci.c
> > +++ b/hw/virtio/virtio-pci.c
> > @@ -314,6 +314,16 @@ static void virtio_ioport_write(void *opaque, uint32_t addr, uint32_t val)
> >              msix_unuse_all_vectors(&proxy->pci_dev);
> >          }
> >  
> > +        /* Linux before 2.6.34 drives the device without enabling
> > +           the PCI device bus master bit. Enable it automatically
> > +           for the guest. This is a PCI spec violation but so is
> > +           initiating DMA with bus master bit clear. */
> > +        if (val == (VIRTIO_CONFIG_S_ACKNOWLEDGE | VIRTIO_CONFIG_S_DRIVER)) {
> > +            pci_default_write_config(&proxy->pci_dev, PCI_COMMAND,
> > +                                     proxy->pci_dev.config[PCI_COMMAND] |
> > +                                     PCI_COMMAND_MASTER, 1);
> > +        }
> > +
> >          /* Linux before 2.6.34 sets the device as OK without enabling
> >             the PCI device bus master bit. In this case we need to disable
> >             some safety checks. */
> 
> So with this patch, we can also remove the detecting and handling of
> VIRTIO_PCI_FLAG_BUS_MATER_BUG?

Yes - I sent a follow-up patch to do that.
Amos Kong Sept. 15, 2014, 10:13 a.m. UTC | #5 
On Mon, Sep 15, 2014 at 12:59:44PM +0300, Michael S. Tsirkin wrote:
> On Mon, Sep 15, 2014 at 05:48:17PM +0800, Jason Wang wrote:
> > On 09/12/2014 02:20 AM, Michael S. Tsirkin wrote:
> > > commit cc943c36faa192cd4b32af8fe5edb31894017d35
> > >     pci: Use bus master address space for delivering MSI/MSI-X messages
> > > breaks virtio-net for rhel6.[56] x86 guests because they don't
> > > enable bus mastering for virtio PCI devices. For the same reason,
> > > rhel6.[56] ppc64 guests cannot boot on a virtio-blk disk anymore.
> > >
> > > Old guests forgot to enable bus mastering, enable it automatically on
> > > DRIVER (guests use some devices before DRIVER_OK).
> > >
> > > Reported-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
> > > Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> > > ---
> > > Sorry, v4 was broken - forgot to commit before git format-patch.
> > > Here's the correct version.
> > > Still not tested - will do next week.
> > > Testing reports appreciated meanwhile.
> > >
> > >  hw/virtio/virtio-pci.c | 10 ++++++++++
> > >  1 file changed, 10 insertions(+)
> > >
> > > diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
> > > index ddb5da1..a827cd4 100644
> > > --- a/hw/virtio/virtio-pci.c
> > > +++ b/hw/virtio/virtio-pci.c
> > > @@ -314,6 +314,16 @@ static void virtio_ioport_write(void *opaque, uint32_t addr, uint32_t val)
> > >              msix_unuse_all_vectors(&proxy->pci_dev);
> > >          }
> > >  
> > > +        /* Linux before 2.6.34 drives the device without enabling
> > > +           the PCI device bus master bit. Enable it automatically
> > > +           for the guest. This is a PCI spec violation but so is
> > > +           initiating DMA with bus master bit clear. */
> > > +        if (val == (VIRTIO_CONFIG_S_ACKNOWLEDGE | VIRTIO_CONFIG_S_DRIVER)) {
> > > +            pci_default_write_config(&proxy->pci_dev, PCI_COMMAND,
> > > +                                     proxy->pci_dev.config[PCI_COMMAND] |
> > > +                                     PCI_COMMAND_MASTER, 1);
> > > +        }
> > > +
> > >          /* Linux before 2.6.34 sets the device as OK without enabling
> > >             the PCI device bus master bit. In this case we need to disable
> > >             some safety checks. */
> > 
> > So with this patch, we can also remove the detecting and handling of
> > VIRTIO_PCI_FLAG_BUS_MATER_BUG?
> 
> Yes - I sent a follow-up patch to do that.

I will check if this patch fixed this bug (it seems WIN8r2 doesn't set
BUS_MASTER bit correctly): https://bugzilla.redhat.com/show_bug.cgi?id=1090237
diff mbox

Patch

diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
index ddb5da1..a827cd4 100644
--- a/hw/virtio/virtio-pci.c
+++ b/hw/virtio/virtio-pci.c
@@ -314,6 +314,16 @@  static void virtio_ioport_write(void *opaque, uint32_t addr, uint32_t val)
             msix_unuse_all_vectors(&proxy->pci_dev);
         }
 
+        /* Linux before 2.6.34 drives the device without enabling
+           the PCI device bus master bit. Enable it automatically
+           for the guest. This is a PCI spec violation but so is
+           initiating DMA with bus master bit clear. */
+        if (val == (VIRTIO_CONFIG_S_ACKNOWLEDGE | VIRTIO_CONFIG_S_DRIVER)) {
+            pci_default_write_config(&proxy->pci_dev, PCI_COMMAND,
+                                     proxy->pci_dev.config[PCI_COMMAND] |
+                                     PCI_COMMAND_MASTER, 1);
+        }
+
         /* Linux before 2.6.34 sets the device as OK without enabling
            the PCI device bus master bit. In this case we need to disable
            some safety checks. */