Re: [PATCH v2 2/6] vfio: Avoid inspecting option QDict for rombar

[Akihiko Odaki <akihiko.odaki@daynix.com>]

On 2024/02/12 17:25, Cédric Le Goater wrote:
> On 2/12/24 09:04, Philippe Mathieu-Daudé wrote:
>> On 10/2/24 11:24, Akihiko Odaki wrote:
>>> Use pci_rom_bar_explicitly_enabled() to determine if rombar is 
>>> explicitly
>>> enabled.
>>>
>>> Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
>>> ---
>>>   hw/vfio/pci.c | 3 +--
>>>   1 file changed, 1 insertion(+), 2 deletions(-)
>>>
>>> diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
>>> index d7fe06715c4b..44178ac9355f 100644
>>> --- a/hw/vfio/pci.c
>>> +++ b/hw/vfio/pci.c
>>> @@ -1010,7 +1010,6 @@ static void vfio_pci_size_rom(VFIOPCIDevice *vdev)
>>>   {
>>>       uint32_t orig, size = cpu_to_le32((uint32_t)PCI_ROM_ADDRESS_MASK);
>>>       off_t offset = vdev->config_offset + PCI_ROM_ADDRESS;
>>> -    DeviceState *dev = DEVICE(vdev);
>>>       char *name;
>>>       int fd = vdev->vbasedev.fd;
>>> @@ -1044,7 +1043,7 @@ static void vfio_pci_size_rom(VFIOPCIDevice *vdev)
>>>       }
>>>       if (vfio_opt_rom_in_denylist(vdev)) {
>>> -        if (dev->opts && qdict_haskey(dev->opts, "rombar")) {
>>> +        if (pci_rom_bar_explicitly_enabled(&vdev->pdev)) {
>>
>> "pdev" is considered internal field, please use the DEVICE() macro
>> to access it. 
> 
> Yes. I was just looking at  vfio_pci_size_rom(). There is a test at
> the beginning of this routine which should be changed to use DEVICE()

It should be fine since vdev is VFIOPCIDevice * (therefore vdev->pdev 
refers to PCIDevice instead of DeviceState) and this code is an internal 
implementation of vfio-pci.

Regards,
Akihiko Odaki