Re: [PATCH 1/1] tty: serial: handle HAS_IOPORT dependencies

[Niklas Schnelle <niks@kernel.org>]

On Fri, 2024-11-22 at 09:07 -0800, Guenter Roeck wrote:
> On 11/22/24 07:35, Niklas Schnelle wrote:
> > On Fri, 2024-11-22 at 07:18 -0800, Guenter Roeck wrote:
> > > On Fri, Apr 05, 2024 at 05:29:24PM +0200, Niklas Schnelle wrote:
> > > > In a future patch HAS_IOPORT=n will disable inb()/outb() and
> > > > friends at
> > > > compile time. We thus need to add HAS_IOPORT as dependency for
> > > > those
> > > > drivers using them unconditionally. For 8250 based drivers some
> > > > support
> > > > MMIO only use so fence only the parts requiring I/O ports.
> > > > 
> > > > Co-developed-by: Arnd Bergmann <arnd@kernel.org>
> > > > Signed-off-by: Arnd Bergmann <arnd@kernel.org>
> > > > Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
> > > ...
> > > > @@ -422,10 +443,12 @@ static void set_io_from_upio(struct
> > > > uart_port *p)
> > > >   	up->dl_write = default_serial_dl_write;
> > > >   
> > > > +	default:
> > > > +		WARN(1, "Unsupported UART type %x\n", p-
> > > > >iotype);
> > > 
> > > So, according to this patch, the serial uart on microblaze,
> > > nios2,
> > > openrisc, xtensa, and possibly others is not or no longer
> > > supported.
> > > 
> > > WARNING: CPU: 0 PID: 0 at drivers/tty/serial/8250/8250_port.c:470
> > > serial8250_set_defaults+0x1a8/0x22c
> > > Unsupported UART type 0
> > > 
> > > Any special reason ?
> > > 
> > > Guenter
> > 
> > So according to the warning the p->iotype is 0 which is UPIO_PORT.
> > For UPIO_PORT the switch above the WARN would pick io_serial_in()
> > and
> > io_serial_out() as handlers. These use inb() respectively outb() to
> > access the serial so I don't see how they would work with
> > !HAS_IOPORT
> > and it most definitely won't work for s390.
> > 
> > Now for Microblaze Kconfig says to select HAS_IOPORT if PCI so I'd
> > assume that it can use inb()/outb() and maybe the PCI requirement
> > is
> > not correct if this isn't a PCI device and it used to work with
> > inb()/outb()? For nios2, openrisc, and xtensa they don't select
> > HAS_IOPORT so either it really won't work anyway or they should
> > select
> > it. Can you tell us more about the devices involved where you saw
> > this?
> > 
> 
> This is seen when booting the affected architectures with qemu.
> 
> Logs:
> 
> https://kerneltests.org/builders/qemu-microblaze-master/builds/327/steps/qemubuildcommand/logs/stdio
> https://kerneltests.org/builders/qemu-nios2-master/builds/314/steps/qemubuildcommand/logs/stdio
> https://kerneltests.org/builders/qemu-openrisc-master/builds/301
> https://kerneltests.org/builders/qemu-xtensa-master/builds/311/steps/qemubuildcommand/logs/stdio
> 
> Guenter
> 

Am I seeing it right that despite the warning and the code setting
no_serial_in / no_serial_out the console=ttyS0 in the above qemu boots
still worked? Also for example in the nios2 case I see the warning 4
times. So this makes me wonder since the warning is new is it possible
that set_io_from_upio() has been called with an invalid / all
zero port before but it was invisible.

The way I'm reading __serial8250_isa_init_ports() and in particular the
first loop if nr_uarts is e.g. 5 in the nios case but only the first
entry in serial8250_ports[] has the IOMEM 8250 it will still call
serial8250_setup_port() on the 4 other unintalized/zero elements which
would explain the iotype being 0. And as far as I can see nr_uarts is
just set to the value of CONFIG_SERIAL_8250_RUNTIME_UARTS in
8250_platform.c.

I may be totally off though this console_init() stuff has me a little
confused and it's been a long day.

Regards,
Niklas