> On Dec 16, 2021, at 9:10 AM, Stefan Hajnoczi <stefa...@redhat.com> wrote:
> 
> On Wed, Dec 15, 2021 at 10:35:34AM -0500, Jagannathan Raman wrote:
>> +static ssize_t vfu_object_bar_rw(PCIDevice *pci_dev, hwaddr addr, size_t 
>> count,
>> +                                 char * const buf, const bool is_write,
>> +                                 bool is_io)
>> +{
>> +    AddressSpace *as = NULL;
>> +    MemTxResult res;
>> +
>> +    if (is_io) {
>> +        as = &address_space_io;
>> +    } else {
>> +        as = pci_device_iommu_address_space(pci_dev);
> 
> This access is not initiated by the device, it's coming from the CPU. It
> shouldn't go through the IOMMU address space.

Got it, thank you!
--
Jag


Reply via email to