On tis, jan 02, 2024 at 10:12, "Russell King (Oracle)" <linux@xxxxxxxxxxxxxxx> wrote:
> On Tue, Dec 19, 2023 at 11:15:41AM +0100, Tobias Waldekranz wrote:
>> On tis, dec 19, 2023 at 10:22, Marek Behún <kabel@xxxxxxxxxx> wrote:
>> > On Thu, 14 Dec 2023 21:14:39 +0100
>> > Tobias Waldekranz <tobias@xxxxxxxxxxxxxx> wrote:
>> >
>> >> +MODULE_FIRMWARE("mrvl/x3310fw.hdr");
>> >
>> > And do you have permission to publish this firmware into linux-firmware?
>>
>> No, I do not.
>>
>> > Because when we tried this with Marvell, their lawyer guy said we can't
>> > do that...
>>
>> I don't even have good enough access to ask the question, much less get
>> rejected by Marvell :) I just used that path so that it would line up
>> with linux-firmware if Marvell was to publish it in the future.
>>
>> Should MODULE_FIRMWARE be avoided for things that are not in
>> linux-firmware?
>
> Without the firmware being published, what use is having this code in
> mainline kernels?
Personally, I primarily want this merged so that future contributions to
the driver are easier to develop, since I'll be able test them on top of
a clean net-next base.
More broadly, I suppose it will help others who are looking to support
similar boards to run the latest kernel, without the need to juggle
external patches which are likely to break as the driver evolves.
Having a single, canonical, implementation of firmware loading, instead
of multiple slightly-broken-in-different-ways ones floating around, also
seems like a net positive.
