diff mbox

arm: dts: update for 4.2-rcx

Message ID 55A7F6F6.40202@ti.com (mailing list archive)
State New, archived
Headers show

Commit Message

Murali Karicheri July 16, 2015, 6:24 p.m. UTC
On 07/16/2015 12:55 PM, santosh shilimkar wrote:
> On 7/16/2015 9:50 AM, Murali Karicheri wrote:
>> On 07/16/2015 12:40 PM, santosh shilimkar wrote:
>>> On 7/16/2015 7:53 AM, Murali Karicheri wrote:
>>>> Santosh,
>>>>
>>>> Do you have my patch for https://lkml.org/lkml/2015/7/15/879 ?
>>>> This one should be applied to v4.2-rc kernel with priority.
>>>> Please review and do the needful.
>>>>
>>> Yes I have it for next merge window. It won't be qualified as a
>>> fix rather more of enhancement. Do you have also patches for
>>> board DTS files which enables PCIE. I can queue them together
>>> for next merge window.
>> No this is a fix for v4.2-rc. PCIe driver is not complete without having
>> a SerDes driver. Currently the version we send is not accepted since the
>> config blobs are used. On K2E EVM, this causes problem since the EVM has
>> Marvell SATA controller present and with default values in the SerDes
>> register, it seems to pass the PCIe link check, but causes issues since
>> the configuration is not correct. The manifestation is that when EVM is
>> booted with NFS rootfs, the boot hangs. We can't enable PCIe on this EVM
>> since to work, SerDes driver has to be present as well. So by default,
>> the PCIe DT binding should be disabled in SoC specific DTS. It can be
>> enabled in the board specific DTS when the SerDes device driver is also
>> present. We plan to upstream a common SerDes driver soon. But at this
>> point this has to be disabled by default.
>>
> Please add this information and respin the patch. Thanks for explaining
> it.
>
>> I can provide more details if needed in the patch commit description,
>> but I though it is not correct to keep the PCIe DT bindings to be
>> enabled in the SoC specific DTS to begin with. So this is a fix for
>> v4.2-rc.
>>
> Please do. It wasn't clear from the patch description. With above added
> description its a legitimate fix which I can queue it up for rc's.
>
Ok. Will do. Need an additional patch to rename the PCI nodes such that 
status can be overridden in a board specific DTS.

+
+&pcie1 {
+       status = "okay";
+};


Same for pcie0 in keystone.dtsi. Do you prefer to put both in the same 
patch or a separate one for this?

Murali
> Regards,
> Santosh
>
>

Comments

Santosh Shilimkar July 16, 2015, 8:35 p.m. UTC | #1
On 7/16/2015 11:24 AM, Murali Karicheri wrote:
> On 07/16/2015 12:55 PM, santosh shilimkar wrote:
>> On 7/16/2015 9:50 AM, Murali Karicheri wrote:
>>> On 07/16/2015 12:40 PM, santosh shilimkar wrote:
>>>> On 7/16/2015 7:53 AM, Murali Karicheri wrote:
>>>>> Santosh,
>>>>>
>>>>> Do you have my patch for https://lkml.org/lkml/2015/7/15/879 ?
>>>>> This one should be applied to v4.2-rc kernel with priority.
>>>>> Please review and do the needful.
>>>>>
>>>> Yes I have it for next merge window. It won't be qualified as a
>>>> fix rather more of enhancement. Do you have also patches for
>>>> board DTS files which enables PCIE. I can queue them together
>>>> for next merge window.
>>> No this is a fix for v4.2-rc. PCIe driver is not complete without having
>>> a SerDes driver. Currently the version we send is not accepted since the
>>> config blobs are used. On K2E EVM, this causes problem since the EVM has
>>> Marvell SATA controller present and with default values in the SerDes
>>> register, it seems to pass the PCIe link check, but causes issues since
>>> the configuration is not correct. The manifestation is that when EVM is
>>> booted with NFS rootfs, the boot hangs. We can't enable PCIe on this EVM
>>> since to work, SerDes driver has to be present as well. So by default,
>>> the PCIe DT binding should be disabled in SoC specific DTS. It can be
>>> enabled in the board specific DTS when the SerDes device driver is also
>>> present. We plan to upstream a common SerDes driver soon. But at this
>>> point this has to be disabled by default.
>>>
>> Please add this information and respin the patch. Thanks for explaining
>> it.
>>
>>> I can provide more details if needed in the patch commit description,
>>> but I though it is not correct to keep the PCIe DT bindings to be
>>> enabled in the SoC specific DTS to begin with. So this is a fix for
>>> v4.2-rc.
>>>
>> Please do. It wasn't clear from the patch description. With above added
>> description its a legitimate fix which I can queue it up for rc's.
>>
> Ok. Will do. Need an additional patch to rename the PCI nodes such that
> status can be overridden in a board specific DTS.
>
> +
> +&pcie1 {
> +       status = "okay";
> +};
> diff --git a/arch/arm/boot/dts/k2e.dtsi b/arch/arm/boot/dts/k2e.dtsi
> index 71e2b1c..0055477 100644
> --- a/arch/arm/boot/dts/k2e.dtsi
> +++ b/arch/arm/boot/dts/k2e.dtsi
> @@ -94,7 +94,7 @@
>                          status = "disabled";
>                  };
>
> -               pcie@21020000 {
> +               pcie1: pcie@21020000 {
>                          compatible = "ti,keystone-pcie","snps,dw-pcie";
>                          clocks = <&clkpcie1>;
>                          clock-names = "pcie";
>
>
> Same for pcie0 in keystone.dtsi. Do you prefer to put both in the same
> patch or a separate one for this?
>
Two patches please.

Regards,
Santosh
diff mbox

Patch

diff --git a/arch/arm/boot/dts/k2e.dtsi b/arch/arm/boot/dts/k2e.dtsi
index 71e2b1c..0055477 100644
--- a/arch/arm/boot/dts/k2e.dtsi
+++ b/arch/arm/boot/dts/k2e.dtsi
@@ -94,7 +94,7 @@ 
                         status = "disabled";
                 };

-               pcie@21020000 {
+               pcie1: pcie@21020000 {
                         compatible = "ti,keystone-pcie","snps,dw-pcie";
                         clocks = <&clkpcie1>;
                         clock-names = "pcie";