[Qemu-devel] [PATCH v3] xen/pt: allow passthrough of devices with bogus interrupt pin

[Qemu-devel] [PATCH v3] xen/pt: allow passthrough of devices with bogus interrupt pin

[Zhao Yan]

For some pci device, even its PCI_INTERRUPT_PIN is not 0, it actually
doesn't support INTx mode, so its machine irq read from host sysfs is 0.
In that case, report PCI_INTERRUPT_PIN as 0 to guest and let passthrough
continue.

Cc: Roger Pau Monné <roger.pau@citrix.com>
Cc: Jan Beulich <JBeulich@suse.com>
Signed-off-by: Zhao Yan <yan.y.zhao@intel.com>

---
v2: fix some coding style issue
v3:
   1. let subject be more descriptive (roger)
   2. disable INTx assertion if machine irq is 0.(roger)
   3. in xen_pt_irqpin_reg_init(), drop the else branch as the default
value for *data is 0. (roger)
---
 hw/xen/xen_pt.c             | 7 +++++++
 hw/xen/xen_pt_config_init.c | 4 +++-
 2 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/hw/xen/xen_pt.c b/hw/xen/xen_pt.c
index e5a6eff..b563837 100644
--- a/hw/xen/xen_pt.c
+++ b/hw/xen/xen_pt.c
@@ -847,6 +847,13 @@ static void xen_pt_realize(PCIDevice *d, Error **errp)
     }
 
     machine_irq = s->real_device.irq;
+    if (machine_irq == 0) {
+        XEN_PT_LOG(d, "machine irq is 0\n");
+        cmd |= PCI_COMMAND_INTX_DISABLE;
+        s->machine_irq = 0;
+        goto out;
+    }
+
     rc = xc_physdev_map_pirq(xen_xc, xen_domid, machine_irq, &pirq);
     if (rc < 0) {
         error_setg_errno(errp, errno, "Mapping machine irq %u to"
diff --git a/hw/xen/xen_pt_config_init.c b/hw/xen/xen_pt_config_init.c
index aee31c6..12f71c1 100644
--- a/hw/xen/xen_pt_config_init.c
+++ b/hw/xen/xen_pt_config_init.c
@@ -300,7 +300,9 @@ static int xen_pt_irqpin_reg_init(XenPCIPassthroughState *s,
                                   XenPTRegInfo *reg, uint32_t real_offset,
                                   uint32_t *data)
 {
-    *data = xen_pt_pci_read_intx(s);
+    if (s->real_device.irq) {
+        *data = xen_pt_pci_read_intx(s);
+    }
     return 0;
 }
 
-- 
2.7.4

Re: [Qemu-devel] [PATCH v3] xen/pt: allow passthrough of devices with bogus interrupt pin

[Roger Pau Monné]

On Tue, Dec 04, 2018 at 02:50:49AM -0500, Zhao Yan wrote:
> For some pci device, even its PCI_INTERRUPT_PIN is not 0, it actually
> doesn't support INTx mode, so its machine irq read from host sysfs is 0.
> In that case, report PCI_INTERRUPT_PIN as 0 to guest and let passthrough
> continue.
> 
> Cc: Roger Pau Monné <roger.pau@citrix.com>
> Cc: Jan Beulich <JBeulich@suse.com>
> Signed-off-by: Zhao Yan <yan.y.zhao@intel.com>

Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>

One nit below.

> 
> ---
> v2: fix some coding style issue
> v3:
>    1. let subject be more descriptive (roger)
>    2. disable INTx assertion if machine irq is 0.(roger)
>    3. in xen_pt_irqpin_reg_init(), drop the else branch as the default
> value for *data is 0. (roger)
> ---
>  hw/xen/xen_pt.c             | 7 +++++++
>  hw/xen/xen_pt_config_init.c | 4 +++-
>  2 files changed, 10 insertions(+), 1 deletion(-)
> 
> diff --git a/hw/xen/xen_pt.c b/hw/xen/xen_pt.c
> index e5a6eff..b563837 100644
> --- a/hw/xen/xen_pt.c
> +++ b/hw/xen/xen_pt.c
> @@ -847,6 +847,13 @@ static void xen_pt_realize(PCIDevice *d, Error **errp)
>      }
>  
>      machine_irq = s->real_device.irq;
> +    if (machine_irq == 0) {
> +        XEN_PT_LOG(d, "machine irq is 0\n");
> +        cmd |= PCI_COMMAND_INTX_DISABLE;
> +        s->machine_irq = 0;

AFAICT this is already initialized to 0, so you can drop setting
machine_irq.

Thanks, Roger.

Re: [Qemu-devel] [PATCH v3] xen/pt: allow passthrough of devices with bogus interrupt pin

[Zhao Yan]

On Tue, Dec 04, 2018 at 10:30:18AM +0100, Roger Pau Monné wrote:
> On Tue, Dec 04, 2018 at 02:50:49AM -0500, Zhao Yan wrote:
> > For some pci device, even its PCI_INTERRUPT_PIN is not 0, it actually
> > doesn't support INTx mode, so its machine irq read from host sysfs is 0.
> > In that case, report PCI_INTERRUPT_PIN as 0 to guest and let passthrough
> > continue.
> > 
> > Cc: Roger Pau Monné <roger.pau@citrix.com>
> > Cc: Jan Beulich <JBeulich@suse.com>
> > Signed-off-by: Zhao Yan <yan.y.zhao@intel.com>
> 
> Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>
> 
> One nit below.
Got it. Thank you!
:)
> 
> > 
> > ---
> > v2: fix some coding style issue
> > v3:
> >    1. let subject be more descriptive (roger)
> >    2. disable INTx assertion if machine irq is 0.(roger)
> >    3. in xen_pt_irqpin_reg_init(), drop the else branch as the default
> > value for *data is 0. (roger)
> > ---
> >  hw/xen/xen_pt.c             | 7 +++++++
> >  hw/xen/xen_pt_config_init.c | 4 +++-
> >  2 files changed, 10 insertions(+), 1 deletion(-)
> > 
> > diff --git a/hw/xen/xen_pt.c b/hw/xen/xen_pt.c
> > index e5a6eff..b563837 100644
> > --- a/hw/xen/xen_pt.c
> > +++ b/hw/xen/xen_pt.c
> > @@ -847,6 +847,13 @@ static void xen_pt_realize(PCIDevice *d, Error **errp)
> >      }
> >  
> >      machine_irq = s->real_device.irq;
> > +    if (machine_irq == 0) {
> > +        XEN_PT_LOG(d, "machine irq is 0\n");
> > +        cmd |= PCI_COMMAND_INTX_DISABLE;
> > +        s->machine_irq = 0;
> 
> AFAICT this is already initialized to 0, so you can drop setting
> machine_irq.
> 
> Thanks, Roger.