[PATCH 0/2] Allow non-PCI device as primary

Thierry Reding thierry.reding at gmail.com
Wed Jun 25 06:49:44 PDT 2014

From: Thierry Reding <treding at nvidia.com>


These patches are the same as I posted back in February[0], but rebased
on top of today's master branch. They solve the same issue that Rob's
patches address but are slightly less intrusive (no autoload support is
added in the series, that's the target for the OutputClass series). One
other difference is that this series uses the platform bus as a last
resort fallback if no other bus was determined to be the primary one. I
think that's a little safer than the fallback in xf86platformProbe()
from Rob's patches because it gives the PCI and SBUS implementations an
opportunity to probe first.

Another difference is that, because the platform bus is a last resort,
it isn't picky about the type of device that it uses. It will simply use
the first one irrespective of the underlying kernel bus (Rob's patches
work only for the kernel's platform devices, which doesn't work on Tegra
for example since it uses a virtual device for DRM).

Adding a couple of people on Cc who've shown interest in these patches.


[0]: http://lists.x.org/archives/xorg-devel/2014-February/040568.html

Thierry Reding (2):
  xfree86: Fallback to first platform device as primary
  xfree86: Allow non-PCI devices as primary

 hw/xfree86/common/xf86Bus.c         |  3 +++
 hw/xfree86/common/xf86platformBus.c | 24 ++++++++++++++++++++----
 hw/xfree86/common/xf86platformBus.h |  1 +
 3 files changed, 24 insertions(+), 4 deletions(-)


More information about the xorg-devel mailing list