On 05/26/2017 09:28 AM, Jorge Ramirez wrote:
[1] https://github.com/ldts/poplar-u-boot/blob/upstream/arch/arm/dts/hi3798cv200...
Yes, sorry. [1] needs to be updated to disable uart0 so that you can use platform data, at least for now. I do want to talk more with Rob about the general problem this exposes.
so you want me to
- keep the node in
https://github.com/ldts/poplar-u-boot/blob/upstream/arch/arm/dts/hi3798cv200...
Well, a uart0 node, but no "clock" property as that just needs to go away.
Sounds good to me. now see below
- add status=disable
- then add the platform_data
BUT for the pl011 driver to take the platform_data dont I also have to disable CONFIG_OF?
but if I disable CONFIG_OF then I lose USB_DM
No, the status = "disable" on uart0 should remove it from the dtb, or at least we should see it and go "Oh, no, we don't have uart0 via DT".
- Since the UART0 is enabled in the kernel's DTS I will have to
modify the device tree that I am importing from kernel.org to disable it.
- But even doing this is not enough: I have to completely remove the
uart0 node from the tree.
So to sum up:
In order to get the platform data for pl01x I have to either disable OF (so I lose the USB node as I said earlier) or *completely* remove the UART0 node from from the kernel dts. I personally would rather not modify the kernel's DTS trees that I am importing into uboot but I am really confused about the policy now.
ie: to be clear from my side, doing the following is not enough:
diff --git a/arch/arm/dts/hi3798cv200-poplar.dts b/arch/arm/dts/hi3798cv200-poplar.dts index b914287..d4ce16d 100644 --- a/arch/arm/dts/hi3798cv200-poplar.dts +++ b/arch/arm/dts/hi3798cv200-poplar.dts @@ -152,7 +152,7 @@ };
&uart0 { - status = "okay"; + status = "disabled"; };
I have to actually do
diff --git a/arch/arm/dts/hi3798cv200-poplar.dts b/arch/arm/dts/hi3798cv200-poplar.dts index b914287..028013f 100644 --- a/arch/arm/dts/hi3798cv200-poplar.dts +++ b/arch/arm/dts/hi3798cv200-poplar.dts @@ -17,7 +17,6 @@ compatible = "hisilicon,hi3798cv200-poplar", "hisilicon,hi3798cv200";
aliases { - serial0 = &uart0; serial2 = &uart2; };
@@ -151,12 +150,9 @@ label = "LS-SPI0"; };
-&uart0 { - status = "okay"; -};
&uart2 { status = "okay"; label = "LS-UART0"; };
OR
[jramirez@titan git.u-boot (upstream *$)]$ git diff diff --git a/arch/arm/dts/hi3798cv200-poplar.dts b/arch/arm/dts/hi3798cv200-poplar.dts index b914287..5d909b83 100644 --- a/arch/arm/dts/hi3798cv200-poplar.dts +++ b/arch/arm/dts/hi3798cv200-poplar.dts @@ -21,10 +21,6 @@ serial2 = &uart2; };
- chosen { - stdout-path = "serial0:115200n8"; - }; - memory@0 { device_type = "memory"; reg = <0x0 0x0 0x0 0x80000000>; @@ -152,7 +148,7 @@ };
&uart0 { - status = "okay"; + status = "disabled"; };
&uart2 {
Is this the right way to go then? alter the kernel DTS when merging into uboot?
please could you clarify?
I still think what I proposed when we started is the better way to go: a uboot specific hi3798cv200-u-boot.dtsifile that contains the two nodes that are giving trouble.
The timeline then goes:
- the usb node will disappear as soon as it lands in korg
- the uart node and the whole file will be removed during the cleanup
of all the pl01x uart offenders.
but if you think modifying the kernels dts is better I can do that as well.