Re: [Qemu-devel] [QEMU][RFC V2 06/10] xen-pci: register PCI device in Xen and handle IOREQ_TYPE_PCI_CONFIG

[Julien Grall <julien.grall@citrix.com>]

On 08/23/2012 03:41 PM, Stefano Stabellini wrote:
> On Wed, 22 Aug 2012, Julien Grall wrote:
>    
>> With QEMU disaggregation QEMU needs to specify which PCI device it's able to
>> handle. It will use the device place in the topology (domain, bus, device,
>> function).
>> When Xen will trap an access for the config space, it will forge a new
>> ioreq and forward it to the right QEMU.
>>
>> Signed-off-by: Julien Grall<julien.grall@citrix.com>
>> ---
>>   hw/pci.c   |    6 ++++++
>>   hw/xen.h   |    1 +
>>   xen-all.c  |   38 ++++++++++++++++++++++++++++++++++++++
>>   xen-stub.c |    5 +++++
>>   4 files changed, 50 insertions(+), 0 deletions(-)
>>
>> diff --git a/hw/pci.c b/hw/pci.c
>> index 4d95984..0112edf 100644
>> --- a/hw/pci.c
>> +++ b/hw/pci.c
>> @@ -33,6 +33,7 @@
>>   #include "qmp-commands.h"
>>   #include "msi.h"
>>   #include "msix.h"
>> +#include "xen.h"
>>
>>   //#define DEBUG_PCI
>>   #ifdef DEBUG_PCI
>> @@ -781,6 +782,11 @@ static PCIDevice *do_pci_register_device(PCIDevice *pci_dev, PCIBus *bus,
>>       pci_dev->devfn = devfn;
>>       pstrcpy(pci_dev->name, sizeof(pci_dev->name), name);
>>       pci_dev->irq_state = 0;
>> +
>> +    if (xen_enabled()&&  xen_register_pcidev(pci_dev)) {
>> +        return NULL;
>>      
> Is this an error condition? If so we should print an error message,
> right?
>    

Yes, it means that the BDF is already registered by another QEMU.