Re: [PATCH] serial: 8250_hub6: Fix double definition for hub6_match_port()

[Uwe Kleine-König]

Hello Hugo,

On Mon, Jul 20, 2026 at 06:53:26PM -0400, Hugo Villeneuve wrote:
> On Mon, 20 Jul 2026 23:18:47 +0200
> Uwe Kleine-König <u.kleine-koenig@xxxxxxxxxxxx> wrote:
> 
> > On Mon, Jul 20, 2026 at 10:07:08AM -0400, Hugo Villeneuve wrote:
> > > Hi Uwe,
> > > 
> > > On Mon, 20 Jul 2026 12:31:31 +0200
> > > Uwe Kleine-König (The Capable Hub) <u.kleine-koenig@xxxxxxxxxxxx> wrote:
> > > 
> > > > On Mon, Jul 20, 2026 at 10:08:12AM +0200, Uwe Kleine-König (The Capable Hub) wrote:
> > > > > With CONFIG_SERIAL_8250_HUB6 configured as module
> > > > > (CONFIG_SERIAL_8250_HUB6=m) the cpp symbol CONFIG_SERIAL_8250_HUB6 isn't
> > > > > defined (only CONFIG_SERIAL_8250_HUB6_MODULE is). In this case
> > > > > hub6_match_port() is defined in drivers/tty/serial/8250/8250.h as static
> > > > > inline and in drivers/tty/serial/8250/8250_hub6.c as normal function.
> > > > > 
> > > > > Use IS_ENABLED() to also handle the modular case correctly.
> > > > > 
> > > > > Fixes: 3d406299d882 ("serial: 8250_hub6: add hub6_match_port()")
> > > > > Signed-off-by: Uwe Kleine-König (The Capable Hub) <u.kleine-koenig@xxxxxxxxxxxx>
> > > > > ---
> > > > >  drivers/tty/serial/8250/8250.h | 2 +-
> > > > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > > > 
> > > > > diff --git a/drivers/tty/serial/8250/8250.h b/drivers/tty/serial/8250/8250.h
> > > > > index 9d1068d0489d..fb9a7117d145 100644
> > > > > --- a/drivers/tty/serial/8250/8250.h
> > > > > +++ b/drivers/tty/serial/8250/8250.h
> > > > > @@ -334,7 +334,7 @@ int fintek_8250_probe(struct uart_8250_port *uart);
> > > > >  static inline int fintek_8250_probe(struct uart_8250_port *uart) { return 0; }
> > > > >  #endif
> > > > >  
> > > > > -#ifdef CONFIG_SERIAL_8250_HUB6
> > > > > +#if IS_ENABLED(CONFIG_SERIAL_8250_HUB6)
> > > > >  bool hub6_match_port(const struct uart_port *port1, const struct uart_port *port2);
> > > > >  #else
> > > > >  static inline bool hub6_match_port(const struct uart_port *port1, const struct uart_port *port2)
> > > > > 
> > > > > base-commit: 0718283ab28bc3907e10b61a6b4be6fefa1cbb2f
> > > > 
> > > > There is an earlier fix already at
> > > > https://lore.kernel.org/linux-serial/20260715153707.4181828-1-hugo@xxxxxxxxxxx/
> > > > 
> > > > It's a bit different, but essentially equivalent. So please stick to
> > > > that older patch.
> > > 
> > > They are not essentially equivalent. If 
> > > CONFIG_SERIAL_8250=y and CONFIG_SERIAL_8250_HUB6=m, and using
> > > IS_ENABLED() gives this linker error:
> > > 
> > > ld: vmlinux.o: in function `uart_match_port':
> > > (.text+0x723aac): undefined reference to `hub6_match_port'
> > > 
> > > Using IS_REACHABLE() fixes this.
> > 
> > Oh, I read too quickly over the dependencies, I thought
> > SERIAL_8250_HUB6's
> > 
> > 	depends on SERIAL_8250 != n && ISA && SERIAL_8250_MANY_PORTS
> > 
> > would prevent that. With this configuration and IS_REACHABLE I wonder if
> > it's an issue that hub6_match_port() returns false in
> > drivers/tty/serial/serial_core.c.
> 
> I forgot to mention that I temporary removed ISA in Kconfig to be able
> to build and test it...

I guess that means, you don't have hardware either to test here.

For me the open question remains if the driver works in the
configuration CONFIG_SERIAL_8250=y (and thus CONFIG_SERIAL_CORE=y) +
CONFIG_SERIAL_8250_HUB6=m. In this case at least commit
3d406299d8829747fe2e8692f4c29fe3dc1d101f +
https://lore.kernel.org/linux-serial/20260715153707.4181828-1-hugo@xxxxxxxxxxx/
introduces a logical change in uart_match_port() that isn't explained in
the commit log.

Best regards
Uwe

Attachment: signature.asc
Description: PGP signature