[PATCH 2/2] ti: Do not remove ft_board_setup for CONFIG_NO_NET

Tom Rini trini at konsulko.com
Fri Sep 5 23:43:43 CEST 2025


On Fri, Sep 05, 2025 at 10:06:10AM +0000, Maarten Brock wrote:
> From: Tom Rini
> > On Thu, Sep 04, 2025 at 01:19:26PM +0000, Maarten Brock wrote:
> > > From: Tom Rini
> > >
> > > > On Mon, Sep 01, 2025 at 04:06:10PM +0000, Maarten Brock wrote:
> > > >
> > > > > Even when CONFIG_NO_NET is selected there is still need for a
> > > > > definition of ft_board_setup(). Thus move the checks for
> > > > > CONFIG_DM_ETH and CONFIG_DRIVER_TI_CPSW inside the function.
> > > > >
> > > > > Signed-off-by: Maarten Brock <maarten.brock at sttls.nl>
> > > >
> > > > Shouldn't OF_BOARD_SETUP be disabled in that case?
> > > >
> > > > --
> > > > Tom
> > >
> > > Had there also been non-NET-related fixups in this ft_board_setup(),
> > > would that also have been your answer?
> > > I don't see why disabling NET drivers would have to impact anything else.
> > 
> > Yes it would have been my answer. And because not all dependencies are
> > easily expressible in Kconfig.
> 
> If this is your opinion then I propose to remove the checks for DM_ETH and
> DRIVER_TI_CPSW alltogether. They obviously serve no purpose as one is
> required to deselect OF_BOARD_SETUP when NO_NET is selected.
> 
> -#if defined(CONFIG_OF_BOARD_SETUP) && defined(CONFIG_OF_CONTROL) && \
> -	defined(CONFIG_DM_ETH) && defined(CONFIG_DRIVER_TI_CPSW)
> +#if defined(CONFIG_OF_BOARD_SETUP) && defined(CONFIG_OF_CONTROL)
> 
> I must hope that there will never be the need for other fixups for e.g.
> the RTC or the user is left with either choosing both RTC and NET or neither.

Well, those checks (or similar ones) were first put in 13-14 years ago
maybe? So yes, some restructuring might well make sense at this point,
given what the codebase looks like now. We can deal with what the future
brings, in the future. Your logs in the other patch show that the kernel
has again changed its mind about where the MAC must come from as I would
have sworn that there was a window of time where no, the kernel wouldn't
read the fuses.

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 228 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20250905/3f0fdeff/attachment.sig>


More information about the U-Boot mailing list