mbox series

[v2,0/3] RK3588 PCIe2 support

Message ID 20230713171851.73052-1-sebastian.reichel@collabora.com (mailing list archive)
Headers show
Series RK3588 PCIe2 support | expand

Message

Sebastian Reichel July 13, 2023, 5:18 p.m. UTC
Hi,

This adds PCIe v2 support for RK3588. The series has been tested with
the onboard RTL8125 network card on Rockchip RK3588 EVB1 (&pcie2x1l1)
and Radxa Rock 5B (&pcie2x1l2).

Changes since v1:
 * https://lore.kernel.org/all/20230616170022.76107-1-sebastian.reichel@collabora.com/
 * Dropped patch adding 'RK3588' (queued by Rob)
 * Updated patch adding legacy-interrupt-controller according to comments
   from Rob and Serge
    - added missing additionalProperties: false
    - added all properties to new required property
    - removed useless quotes around interrupt-controller
    - added Rob's Reviewed-by
 * Updated patch adding the missing RK356x/RK3588 interrupt names, so that it
   provides more details about the interrupts
 * Updated the DT patch according to the comment from Jonas Karlman, so that
   the addresses are in 32 bit address space starting at 0x40000000

[0] https://lore.kernel.org/all/20230612171337.74576-1-sebastian.reichel@collabora.com/

Thanks,

-- Sebastian

Sebastian Reichel (3):
  dt-bindings: PCI: dwc: rockchip: Fix interrupt-names issue
  dt-bindings: PCI: dwc: rockchip: Add missing
    legacy-interrupt-controller
  arm64: dts: rockchip: rk3588: add PCIe2 support

 .../bindings/pci/rockchip-dw-pcie.yaml        |  48 ++++++++
 .../devicetree/bindings/pci/snps,dw-pcie.yaml |  76 +++++++++++-
 arch/arm64/boot/dts/rockchip/rk3588.dtsi      |  54 +++++++++
 arch/arm64/boot/dts/rockchip/rk3588s.dtsi     | 108 ++++++++++++++++++
 4 files changed, 285 insertions(+), 1 deletion(-)

Comments

Heiko Stuebner July 13, 2023, 8:09 p.m. UTC | #1
Hi Sebastian,

Am Donnerstag, 13. Juli 2023, 19:18:48 CEST schrieb Sebastian Reichel:
> Hi,
> 
> This adds PCIe v2 support for RK3588. The series has been tested with
> the onboard RTL8125 network card on Rockchip RK3588 EVB1 (&pcie2x1l1)
> and Radxa Rock 5B (&pcie2x1l2).
> 
> Changes since v1:

Didn't Rob write that he applied patches 1-3 of the v1 series? [0]
or did I miss further communication somehow?

Thanks
Heiko

[0] https://lore.kernel.org/r/20230626193238.GA3553158-robh@kernel.org

>  * https://lore.kernel.org/all/20230616170022.76107-1-sebastian.reichel@collabora.com/
>  * Dropped patch adding 'RK3588' (queued by Rob)
>  * Updated patch adding legacy-interrupt-controller according to comments
>    from Rob and Serge
>     - added missing additionalProperties: false
>     - added all properties to new required property
>     - removed useless quotes around interrupt-controller
>     - added Rob's Reviewed-by
>  * Updated patch adding the missing RK356x/RK3588 interrupt names, so that it
>    provides more details about the interrupts
>  * Updated the DT patch according to the comment from Jonas Karlman, so that
>    the addresses are in 32 bit address space starting at 0x40000000
> 
> [0] https://lore.kernel.org/all/20230612171337.74576-1-sebastian.reichel@collabora.com/
> 
> Thanks,
> 
> -- Sebastian
> 
> Sebastian Reichel (3):
>   dt-bindings: PCI: dwc: rockchip: Fix interrupt-names issue
>   dt-bindings: PCI: dwc: rockchip: Add missing
>     legacy-interrupt-controller
>   arm64: dts: rockchip: rk3588: add PCIe2 support
> 
>  .../bindings/pci/rockchip-dw-pcie.yaml        |  48 ++++++++
>  .../devicetree/bindings/pci/snps,dw-pcie.yaml |  76 +++++++++++-
>  arch/arm64/boot/dts/rockchip/rk3588.dtsi      |  54 +++++++++
>  arch/arm64/boot/dts/rockchip/rk3588s.dtsi     | 108 ++++++++++++++++++
>  4 files changed, 285 insertions(+), 1 deletion(-)
> 
>
Sebastian Reichel July 13, 2023, 8:48 p.m. UTC | #2
Hi Heiko,

On Thu, Jul 13, 2023 at 10:09:29PM +0200, Heiko Stuebner wrote:
> Am Donnerstag, 13. Juli 2023, 19:18:48 CEST schrieb Sebastian Reichel:
> > This adds PCIe v2 support for RK3588. The series has been tested with
> > the onboard RTL8125 network card on Rockchip RK3588 EVB1 (&pcie2x1l1)
> > and Radxa Rock 5B (&pcie2x1l2).
> 
> Didn't Rob write that he applied patches 1-3 of the v1 series? [0]
> or did I miss further communication somehow?
> [0] https://lore.kernel.org/r/20230626193238.GA3553158-robh@kernel.org

Rob wrote, that he de-applied the first patch:

https://lore.kernel.org/all/CAL_Jsq+=kBrujhLW_KNRWpj+DQJbnXrA=RS3La5ekbJtji+xiQ@mail.gmail.com/

It seems the second one was also dropped, since I rebased on top of
6.5-rc1, which only had patch 3/4.

FWIW the remaining dt-bindings fix issues that also exist for
RK356x, so I guess there is no strict dependency. It might be
acceptable to merge the DTS patch already, so that we finally
get working network on Rock 5B. That would temporarily introduce
DT warnings though (RK3588 is currently warning-free).

-- Sebastian
Heiko Stuebner July 14, 2023, 11:54 a.m. UTC | #3
Hi Sebastian,

Am Donnerstag, 13. Juli 2023, 22:48:44 CEST schrieb Sebastian Reichel:
> On Thu, Jul 13, 2023 at 10:09:29PM +0200, Heiko Stuebner wrote:
> > Am Donnerstag, 13. Juli 2023, 19:18:48 CEST schrieb Sebastian Reichel:
> > > This adds PCIe v2 support for RK3588. The series has been tested with
> > > the onboard RTL8125 network card on Rockchip RK3588 EVB1 (&pcie2x1l1)
> > > and Radxa Rock 5B (&pcie2x1l2).
> > 
> > Didn't Rob write that he applied patches 1-3 of the v1 series? [0]
> > or did I miss further communication somehow?
> > [0] https://lore.kernel.org/r/20230626193238.GA3553158-robh@kernel.org
> 
> Rob wrote, that he de-applied the first patch:
> 
> https://lore.kernel.org/all/CAL_Jsq+=kBrujhLW_KNRWpj+DQJbnXrA=RS3La5ekbJtji+xiQ@mail.gmail.com/
> 
> It seems the second one was also dropped, since I rebased on top of
> 6.5-rc1, which only had patch 3/4.

thanks for the clarification :-) .

> FWIW the remaining dt-bindings fix issues that also exist for
> RK356x, so I guess there is no strict dependency. It might be
> acceptable to merge the DTS patch already, so that we finally
> get working network on Rock 5B. That would temporarily introduce
> DT warnings though (RK3588 is currently warning-free).

We're still early in the release cycle, so I guess we should
just give Rob(?) time to pick up the binding patches if they
look ok to him. And not introduce new warnings ;-)


Heiko