Re: fw_devlink=on breaks probing devices when of_platform_populate() is used

[Saravana Kannan]

On Sun, Dec 11, 2022 at 12:46 AM Maksim Kiselev <bigunclemax@xxxxxxxxx> wrote:
>
>
> Hi, I have the same problem.
> https://lore.kernel.org/all/CALHCpMgEZjnR39upkR6iozSk-b5A_GHRo9rcDSPXzzQi6x_qCw@xxxxxxxxxxxxxx/
>
> I think the root of the problem was the choice of 'compatible'
> device tree property to marking mtd partition node as a nvmem provider.
>
> This property used only inside 'mtd_nvmem_add' function to setup
> 'no_of_node' flag.
>
> > config.no_of_node = !of_device_is_compatible(node, "nvmem-cells");
>
> This is how this flag processed by 'nvmem_register' function.
>
> >       if (config->of_node)
> >               nvmem->dev.of_node = config->of_node;
> >       else if (!config->no_of_node)
> >               nvmem->dev.of_node = config->dev->of_node;
>
> Thats all, there is no such driver which compatible with 'nvmem-cells'.
>
>
> So, maybe we should change the 'compatible' property to something else?

Sorry about the accidental HTML in my previous reply. Resending as plain text.

I have a patch series [1](v1 sent out a while back) that stops
depending on the existence of "compatible" property for fw_devlink to
work. I had a few issues that I have fixes for that were tested in the
thread. I've been meaning to send out a v2 with all those fixes rolled
in. I'll try to get that out this week. Hopefully that'll address the
issues assuming Maksim's analysis about "compatible" is correct. If
not, I can take a closer look after that.

[1] - https://lore.kernel.org/lkml/20220810060040.321697-1-saravanak@xxxxxxxxxx/

-Saravana