[Pdbg] [PATCH 1/2] Add checkstop FIR support via DTS

Balbir Singh bsingharora at gmail.com
Wed Apr 11 23:22:37 AEST 2018


On Wed, Feb 14, 2018 at 5:07 PM, Alistair Popple <alistair at popple.id.au> wrote:
> Hi Balbir,
>
>> +mc {
>> +     fir_mask=<0x1800000 0x0>;
>> +     instances=<0x0 0x1>;
>> +     compatible="ibm,chiplet-fir";
>> +};
>
> There are plans to add each chiplet as it's own node in the device-tree with an
> associated ibm,chiplet-id = <0x??> to idenitfy it. Rather than having to have a
> seperate chiplet representation for a chiplets top-level FIR register I think it
> makes more sense to link the bits set in global_fir with device tree nodes based
> on the chiplet-id property. So for example the above would become:
>
> eg:
>
> mc01 {
>      reg = <0xc0ffee>;
>      ibm,chiplet-id = <0x7>;
> };
>
> mc23 {
>     reg = <0xdeadbeef>;
>     ibm,chiplet-id = <0x8>;
> };
>
> Admittedly this results in each chiplet having to explicitly exist in the
> device-tree, but that will be required anyway for the "reg = <...>" properties.
> Anyway let me know your thoughts.

OK, since I've not seen this yet. I am backing off and will revisit when all
of that is added.

Thanks,
Balbir Singh.


More information about the Pdbg mailing list