[0/2] PCI: rockchip: Fix PCIe probing in 5.9

Message

Marc Zyngier Aug. 15, 2020, 12:51 p.m. UTC

Recent changes to the way PCI DT nodes are parsed are now enforcing
the presence of a "device_type" property, which has been mandated
since... forever. This has the unfortunate effect of breaking
non-compliant systems, and those using the Rockchip PCIe driver are
amongst the victims. Oh well.

In order to keep users happy as well as my own machines up and
running, let's paper over the problem by detecting a broken DT from
the driver itself, and inserting the missing property at runtime.

For a good measure, a second patch fixes the DT, but the chances of
such a fix being deployed at this stage are pretty slim, so the above
hack is in for the long run.

Marc Zyngier (2):
  PCI: rockchip: Work around missing device_type property in DT
  arm64: dts: rockchip: Fix PCIe DT properties

 arch/arm64/boot/dts/rockchip/rk3399.dtsi    |  2 +-
 drivers/pci/controller/pcie-rockchip-host.c | 29 +++++++++++++++++++++
 2 files changed, 30 insertions(+), 1 deletion(-)

Comments

Heiko Stübner Jan. 9, 2021, 3:39 p.m. UTC | #1

On Sat, 15 Aug 2020 13:51:10 +0100, Marc Zyngier wrote:
> Recent changes to the way PCI DT nodes are parsed are now enforcing
> the presence of a "device_type" property, which has been mandated
> since... forever. This has the unfortunate effect of breaking
> non-compliant systems, and those using the Rockchip PCIe driver are
> amongst the victims. Oh well.
> 
> In order to keep users happy as well as my own machines up and
> running, let's paper over the problem by detecting a broken DT from
> the driver itself, and inserting the missing property at runtime.
> 
> [...]

Applied, thanks!

[2/2] arm64: dts: rockchip: Fix PCIe DT properties
      commit: 43f20b1c6140896916f4e91aacc166830a7ba849

Best regards,