Re: Regression: drm: Lobotomize set_busid nonsense for !pci drivers (a325725633c2)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, Oct 05, 2016 at 02:43:37PM +0200, Gerd Hoffmann wrote:
> On Mi, 2016-10-05 at 12:30 +0200, Daniel Vetter wrote:
> > On Wed, Oct 05, 2016 at 08:34:57AM +0200, Gerd Hoffmann wrote:
> > > On Di, 2016-10-04 at 09:43 +0200, Gerd Hoffmann wrote:
> > > >   Hi,
> > > > 
> > > > > diff --git a/drivers/gpu/drm/virtio/virtgpu_drm_bus.c
> > > > > b/drivers/gpu/drm/virtio/virtgpu_drm_bus.c
> > > > > index a59d0e309bfc..1fcf739bf509 100644
> > > > > --- a/drivers/gpu/drm/virtio/virtgpu_drm_bus.c
> > > > > +++ b/drivers/gpu/drm/virtio/virtgpu_drm_bus.c
> > > > > @@ -68,6 +68,10 @@ int drm_virtio_init(struct drm_driver *driver,
> > > > > struct virtio_device *vdev)
> > > > >                 dev->pdev = pdev;
> > > > >                 if (vga)
> > > > >                         virtio_pci_kick_out_firmware_fb(pdev);
> > > > > +
> > > > > +               ret = drm_dev_set_unique(dev, dev_name(dev->pdev));
> > > > > +               if (ret)
> > > > > +                       goto err_free;
> > > > >         }
> > > > 
> > > > Approach looks sensible to me, I'll give it a try ...
> > > 
> > > Well, dev_name() returns a string without the "pci:" prefix, we have to
> > > add that to make things actually work, like this:
> > 
> > Hm right, I missed that in digging through the piles of layers of struct
> > device naming.
> > 
> > > 
> > > --- a/drivers/gpu/drm/virtio/virtgpu_drm_bus.c
> > > +++ b/drivers/gpu/drm/virtio/virtgpu_drm_bus.c
> > > @@ -60,13 +60,22 @@ int drm_virtio_init(struct drm_driver *driver,
> > > struct virtio_device *vdev)
> > >  
> > >         if (strcmp(vdev->dev.parent->bus->name, "pci") == 0) {
> > >                 struct pci_dev *pdev = to_pci_dev(vdev->dev.parent);
> > > +               const char *pname = dev_name(&pdev->dev);
> > >                 bool vga = (pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA;
> > > +               char unique[20];
> > 
> > Iirc we kfree this on unload. Needs to be a kasprintf I think.
> 
> drm_dev_set_unique() does this:
> 
> dev->unique = kstrdup(name, GFP_KERNEL);
> 
> So it should not matter at all where the caller stores name.
> Or do you mean something else?

Oh, missed that. r-b stands as-is.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/dri-devel




[Index of Archives]     [Linux DRI Users]     [Linux Intel Graphics]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]
  Powered by Linux