Hi Laszlo,
On 28 March 2018 at 11:27, Laszlo Ersek lersek@redhat.com wrote:
On 03/28/18 03:24, Emil Velikov wrote:
From: Emil Velikov emil.velikov@collabora.com
Ealier commit a325725633c26aa66ab940f762a6b0778edf76c0 did not attribute that virtio can be either PCI or a platform device and removed the .set_busid hook. Whereas only the "platform" instance should have been removed.
Since then, two things have happened:
- the driver manually calls drm_dev_set_unique, addressing the Xserver
regression - see commit 9785b4321b0bd701f8d21d3d3c676a7739a5cf22
- core itself calls drm_pci_set_busid, on drm_set_busid IOCTL setting
the busid, so we don't need to fallback to dev->unique - see commit 5c484cee7ef9c4fd29fa0ba09640d55960977145
With that in place we can remove the local workaround.
Cc: Daniel Vetter daniel.vetter@ffwll.ch Cc: Gerd Hoffmann kraxel@redhat.com Cc: Hans de Goede hdegoede@redhat.com Cc: Laszlo Ersek lersek@redhat.com Signed-off-by: Emil Velikov emil.velikov@collabora.com
Gents, can someone double-check this please? Just in case.
I guess I should test whether this series regresses the use case described in c2cbc38b97; is that correct?
Precisely.
I can add that to my todo list, but it won't be quick. I get into building kernels maybe once per quarter :)
What repo and branch should I apply this series on top of? ... Actually, I was only CC'd on patch 3/4, so I don't even have the full series. Do you have a public branch somewhere?
In practise only 3/4 is needed applied on top of any tree that has the commits mentioned in the summary.
FWIW here is a patchwork series [1], just this patch [2] and a tree [3].
[1] https://patchwork.freedesktop.org/series/40778/ [2] https://patchwork.freedesktop.org/patch/213335/ [3] https://github.com/evelikov/linux/commits/ioctl-cleanups
(If this would mean too much work for you, esp. considering that I'm going to be slow, feel free to go ahead with the patch. Should I find later that something broke, I'll whine then. :) )
The original patch that caused breakage was tested, yet seemingly in a different manner. Hence, I'd love it, if anyone who had the original problem can can give this a try.
I guess it can wait 1-2 weeks before getting merged, since it won't cause any conflicts ;-)
Thanks! Emil