Re: [Xen-devel] [PATCH 1/2] xen/x86: Patch re-factors MSI/X config code from, drivers/passthrough/pci.c to x86 specific

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 13.04.15 at 12:34, <mjaggi@xxxxxxxxxxxxxxxxxx> wrote:
> On Monday 13 April 2015 03:45 PM, Stefano Stabellini wrote:
>> On Mon, 13 Apr 2015, Manish Jaggi wrote:
>>> @@ -282,22 +266,10 @@ static struct pci_dev *alloc_pdev(struct pci_seg 
>>> *pseg,
>>> u8 bus, u8 devfn)
>>>       *((u8*) &pdev->bus) = bus;
>>>       *((u8*) &pdev->devfn) = devfn;
>>>       pdev->domain = NULL;
>>> -    INIT_LIST_HEAD(&pdev->msi_list);
>>> -
>>> -    if ( pci_find_cap_offset(pseg->nr, bus, PCI_SLOT(devfn),
>>> PCI_FUNC(devfn),
>>> -                             PCI_CAP_ID_MSIX) )
>>> +    if (!pci_alloc_msix (pdev, pseg, bus, devfn))
>>>       {
>>> -        struct arch_msix *msix = xzalloc(struct arch_msix);
>>> -
>>> -        if ( !msix )
>>> -        {
>>> -            xfree(pdev);
>>> -            return NULL;
>>> -        }
>>> -        spin_lock_init(&msix->table_lock);
>>> -        pdev->msix = msix;
>>> +        return NULL;
>>>       }
>>> -
>>  From the look of it, this code doesn't seem x86 specific
> MSIX is only used for x86, dom0/U handles MSI/X in ARM. I think If ARM 
> is not using MSI/X then code has to be in x86 specific file.

No, you need to take a more general perspective here: MSI and
MSI-X are integral parts of PCI, and hence if ARM _really_ chooses
to hand control of it to DomU-s, the code here shouldn't become
x86-specific, but non-ARM-specific.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel