[U-Boot] [PATCH 1/2] dm: pci: make ranges dt property optional

Bin Meng bmeng.cn at gmail.com
Wed Jun 6 13:19:41 UTC 2018


Hi Christian,

On Wed, May 23, 2018 at 9:39 PM, Christian Gmeiner
<christian.gmeiner at gmail.com> wrote:
> Hi Bin,
>
> Am Mi., 23. Mai 2018 um 15:10 Uhr schrieb Bin Meng <bmeng.cn at gmail.com>:
>
>> Hi Christian,
>
>> On Wed, May 23, 2018 at 8:00 PM, Christian Gmeiner
>> <christian.gmeiner at gmail.com> wrote:
>> > If we use u-boot as coreboot payload with a generic dts without
>> > any ranges specified we fail in pci pre_probe and our pci bus
>> > is not usable.
>> >
>
>> What do you mean by "a generic dts"?
>
>
>> The coreboot payload needs to specify a dedicated dts for that board,
>> for example to build a coreboot payload for minnowmax, we need specify
>> minnowmax.dts in the Kconfig. README.x86 documents this.
>
>
> Yeah.. so in my eyes a "generic dts" contains only this part regarding pci:
>
>          pci {
>                  compatible = "pci-x86";
>                  #address-cells = <0x3>;
>                  #size-cells = <0x2>;
>          };
>
> The important part is that it does not contain any ranges. Coreboot is the
> one
> who does the pci bus setup (assigning memory windows for devices etc). So I
> do not want to know in u-boot at build time (ranges thing for pci) how the
> pci
> bus looks regarding addresses. My end goal is one generic u-boot payload
> that
> gets used for different FSP 2.0 based boards.
>

I see your point. Then they way we support coreboot might be changed.
We may create a generic U-Boot payload for coreboot targets.
Can you respin the patch to address these comments, like u-boot ->
U-Boot, adding detailed info to the commit message about this, and
update README.x86?

Regards,
Bin


More information about the U-Boot mailing list