Re: [PATCH] serial: Revert serial: core: Fix serial core port id to not use port->line

[Tony Lindgren <tony@atomide.com>]

* Matthew Howell <matthew.howell@sealevel.com> [230831 14:58]:
> On Tue, 29 Aug 2023, Tony Lindgren wrote:
> > If it is a port_id conflict I'm not sure why commit 3d9e6f556e23 is not
> > working for your as it has commit a4a79e03bab5 ("serial: core: Revert
> > port_id use"). Care to check that again, or maybe try with v6.5 with just
> > the commit below reverted?
> > 
> > 04c7f60ca477 ("serial: core: Fix serial core port id, including multiport devices")
> 
> Just tried that, but no difference. Same error.

OK thanks to testing it. So it's starting to look like the issue is
somehow related to the serial8250_setup_port() change in commit d962de6ae51f
("serial: core: Fix serial core port id to not use port->line").

The experimental patch below should confirm if the issue is related to the
port_id usage or serial8250_setup_port(). Care to give this a try against
v6.5 without other patches?

> > Dmesg output might help also to figure out if this happens on the first
> > port or the second port.
> 
> The full error in dmesg is:
> [Aug30 15:48] exar_serial 0000:04:00.0: Couldn't register serial port 0, irq 24, type 2, error -22
> 
> This is on a 2-port adapter. I don't see any indication in dmesg that it 
> attempted to register the other port.

OK. Are there other 8520 related uarts probing before that?

Regards,

Tony

8< --------------------
diff --git a/drivers/tty/serial/serial_base_bus.c b/drivers/tty/serial/serial_base_bus.c
--- a/drivers/tty/serial/serial_base_bus.c
+++ b/drivers/tty/serial/serial_base_bus.c
@@ -169,7 +169,7 @@ struct serial_port_device *serial_base_port_add(struct uart_port *port,
 	err = serial_base_device_init(port, &port_dev->dev,
 				      &ctrl_dev->dev, &serial_port_type,
 				      serial_base_port_release,
-				      port->ctrl_id, port->port_id);
+				      port->ctrl_id, port->line);
 	if (err)
 		goto err_put_device;