A few issues have been found with device naming for the serial core
controller port device. These issues currently mostly affect the output
for /sys/bus/serial-base/devices, but need to be also fixed to avoid
port addressing issues later on.
Changes since v3:
- Drop unnecessary else on the return path in serial_base_device_init()
as noted by Andy
- Add Andy's Reviewed-by
- Update first patch description for port_id instead of port port_id
for the first patch
Changes since v2:
- Fix my email script as it had started to drop linux-serial as noted by
Greg
- Explain why we're changing ctrl_id as requested by Greg
Changes since v1:
- Port id cannot be negative as noted by Jiri
- Controller id cannot be negative as noted by Andy
- Port name is missing the controller instance as noted by Andy
Tony Lindgren (3):
serial: core: Controller id cannot be negative
serial: core: Fix serial core port id to not use port->line
serial: core: Fix serial core controller port name to show controller
id
drivers/tty/serial/8250/8250_core.c | 2 ++
drivers/tty/serial/serial_base_bus.c | 34 ++++++++++++++++++----------
include/linux/serial_core.h | 3 ++-
3 files changed, 26 insertions(+), 13 deletions(-)
--
2.41.0
The serial core port id should be serial core controller specific port
instance, which is not always the port->line index.
For example, 8250 driver maps a number of legacy ports, and when a
hardware specific device driver takes over, we typically have one
driver instance for each port. Let's instead add port->port_id to
keep track serial ports mapped to each serial core controller instance.
Currently this is only a cosmetic issue for the serial core port device
names. The issue can be noticed looking at /sys/bus/serial-base/devices
for example though. Let's fix the issue to avoid port addressing issues
later on.
Fixes: 84a9582fd203 ("serial: core: Start managing serial controllers to enable runtime PM")
Andy Shevchenko <[email protected]>
Signed-off-by: Tony Lindgren <[email protected]>
---
drivers/tty/serial/8250/8250_core.c | 2 ++
drivers/tty/serial/serial_base_bus.c | 2 +-
include/linux/serial_core.h | 1 +
3 files changed, 4 insertions(+), 1 deletion(-)
diff --git a/drivers/tty/serial/8250/8250_core.c b/drivers/tty/serial/8250/8250_core.c
--- a/drivers/tty/serial/8250/8250_core.c
+++ b/drivers/tty/serial/8250/8250_core.c
@@ -497,6 +497,7 @@ static struct uart_8250_port *serial8250_setup_port(int index)
up = &serial8250_ports[index];
up->port.line = index;
+ up->port.port_id = index;
serial8250_init_port(up);
if (!base_ops)
@@ -1040,6 +1041,7 @@ int serial8250_register_8250_port(const struct uart_8250_port *up)
uart_remove_one_port(&serial8250_reg, &uart->port);
uart->port.ctrl_id = up->port.ctrl_id;
+ uart->port.port_id = up->port.port_id;
uart->port.iobase = up->port.iobase;
uart->port.membase = up->port.membase;
uart->port.irq = up->port.irq;
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
@@ -136,7 +136,7 @@ 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->line);
+ port->port_id);
if (err)
goto err_put_device;
diff --git a/include/linux/serial_core.h b/include/linux/serial_core.h
--- a/include/linux/serial_core.h
+++ b/include/linux/serial_core.h
@@ -460,6 +460,7 @@ struct uart_port {
int (*iso7816_config)(struct uart_port *,
struct serial_iso7816 *iso7816);
unsigned int ctrl_id; /* optional serial core controller id */
+ unsigned int port_id; /* optional serial core port id */
unsigned int irq; /* irq number */
unsigned long irqflags; /* irq flags */
unsigned int uartclk; /* base uart clock */
--
2.41.0
The controller id cannot be negative. Let's fix the ctrl_id in preparation
for adding port_id to fix the device name.
Fixes: 84a9582fd203 ("serial: core: Start managing serial controllers to enable runtime PM")
Reported-by: Andy Shevchenko <[email protected]>
Andy Shevchenko <[email protected]>
Signed-off-by: Tony Lindgren <[email protected]>
---
include/linux/serial_core.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/include/linux/serial_core.h b/include/linux/serial_core.h
--- a/include/linux/serial_core.h
+++ b/include/linux/serial_core.h
@@ -459,7 +459,7 @@ struct uart_port {
struct serial_rs485 *rs485);
int (*iso7816_config)(struct uart_port *,
struct serial_iso7816 *iso7816);
- int ctrl_id; /* optional serial core controller id */
+ unsigned int ctrl_id; /* optional serial core controller id */
unsigned int irq; /* irq number */
unsigned long irqflags; /* irq flags */
unsigned int uartclk; /* base uart clock */
--
2.41.0
We are missing the serial core controller id for the serial core port
name. Let's fix the issue for sane /sys/bus/serial-core/devices, and to
avoid issues addressing serial ports later on.
Fixes: 84a9582fd203 ("serial: core: Start managing serial controllers to enable runtime PM")
Reported-by: Andy Shevchenko <[email protected]>
Andy Shevchenko <[email protected]>
Signed-off-by: Tony Lindgren <[email protected]>
---
drivers/tty/serial/serial_base_bus.c | 34 ++++++++++++++++++----------
1 file changed, 22 insertions(+), 12 deletions(-)
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
@@ -19,6 +19,14 @@
static bool serial_base_initialized;
+static const struct device_type serial_ctrl_type = {
+ .name = "ctrl",
+};
+
+static const struct device_type serial_port_type = {
+ .name = "port",
+};
+
static int serial_base_match(struct device *dev, struct device_driver *drv)
{
int len = strlen(drv->name);
@@ -48,7 +56,8 @@ static int serial_base_device_init(struct uart_port *port,
struct device *parent_dev,
const struct device_type *type,
void (*release)(struct device *dev),
- int id)
+ unsigned int ctrl_id,
+ unsigned int port_id)
{
device_initialize(dev);
dev->type = type;
@@ -61,12 +70,17 @@ static int serial_base_device_init(struct uart_port *port,
return -EPROBE_DEFER;
}
- return dev_set_name(dev, "%s.%s.%d", type->name, dev_name(port->dev), id);
-}
+ if (type == &serial_ctrl_type)
+ return dev_set_name(dev, "%s.%s.%d", type->name,
+ dev_name(port->dev), ctrl_id);
-static const struct device_type serial_ctrl_type = {
- .name = "ctrl",
-};
+ if (type == &serial_port_type)
+ return dev_set_name(dev, "%s.%s.%d.%d", type->name,
+ dev_name(port->dev), ctrl_id,
+ port_id);
+
+ return -EINVAL;
+}
static void serial_base_ctrl_release(struct device *dev)
{
@@ -96,7 +110,7 @@ struct serial_ctrl_device *serial_base_ctrl_add(struct uart_port *port,
err = serial_base_device_init(port, &ctrl_dev->dev,
parent, &serial_ctrl_type,
serial_base_ctrl_release,
- port->ctrl_id);
+ port->ctrl_id, 0);
if (err)
goto err_put_device;
@@ -112,10 +126,6 @@ struct serial_ctrl_device *serial_base_ctrl_add(struct uart_port *port,
return ERR_PTR(err);
}
-static const struct device_type serial_port_type = {
- .name = "port",
-};
-
static void serial_base_port_release(struct device *dev)
{
struct serial_port_device *port_dev = to_serial_base_port_device(dev);
@@ -136,7 +146,7 @@ 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->port_id);
+ port->ctrl_id, port->port_id);
if (err)
goto err_put_device;
--
2.41.0
On Mon, Jul 24, 2023 at 08:07:05AM +0300, Tony Lindgren wrote:
> We are missing the serial core controller id for the serial core port
> name. Let's fix the issue for sane /sys/bus/serial-core/devices, and to
> avoid issues addressing serial ports later on.
...
> - return dev_set_name(dev, "%s.%s.%d", type->name, dev_name(port->dev), id);
> -}
> + if (type == &serial_ctrl_type)
...
> +}
While at it, have you tried to use --patience when forming patches? Does it
make any (better) difference?
--
With Best Regards,
Andy Shevchenko
On Mon, Jul 24, 2023 at 08:07:03AM +0300, Tony Lindgren wrote:
> The controller id cannot be negative. Let's fix the ctrl_id in preparation
> for adding port_id to fix the device name.
>
> Fixes: 84a9582fd203 ("serial: core: Start managing serial controllers to enable runtime PM")
> Reported-by: Andy Shevchenko <[email protected]>
> Andy Shevchenko <[email protected]>
???
Missing Reviewed-by: I suppose?
--
With Best Regards,
Andy Shevchenko
On Mon, Jul 24, 2023 at 11:52:39AM +0300, Andy Shevchenko wrote:
> On Mon, Jul 24, 2023 at 08:07:03AM +0300, Tony Lindgren wrote:
> > The controller id cannot be negative. Let's fix the ctrl_id in preparation
> > for adding port_id to fix the device name.
> >
> > Fixes: 84a9582fd203 ("serial: core: Start managing serial controllers to enable runtime PM")
> > Reported-by: Andy Shevchenko <[email protected]>
>
> > Andy Shevchenko <[email protected]>
>
> ???
> Missing Reviewed-by: I suppose?
Seems like it wen in all patches... :-(
--
With Best Regards,
Andy Shevchenko
On 24. 07. 23, 10:55, Andy Shevchenko wrote:
> On Mon, Jul 24, 2023 at 08:07:05AM +0300, Tony Lindgren wrote:
>> We are missing the serial core controller id for the serial core port
>> name. Let's fix the issue for sane /sys/bus/serial-core/devices, and to
>> avoid issues addressing serial ports later on.
>
> ...
>
>> - return dev_set_name(dev, "%s.%s.%d", type->name, dev_name(port->dev), id);
>> -}
>> + if (type == &serial_ctrl_type)
>
> ...
>
>> +}
>
> While at it, have you tried to use --patience when forming patches? Does it
> make any (better) difference?
Or moving "struct device_type"s to one place in a separate patch might
do the desired trick too ;).
--
js
suse labs
* Andy Shevchenko <[email protected]> [230724 08:56]:
> On Mon, Jul 24, 2023 at 08:07:05AM +0300, Tony Lindgren wrote:
> > We are missing the serial core controller id for the serial core port
> > name. Let's fix the issue for sane /sys/bus/serial-core/devices, and to
> > avoid issues addressing serial ports later on.
>
> ...
>
> > - return dev_set_name(dev, "%s.%s.%d", type->name, dev_name(port->dev), id);
> > -}
> > + if (type == &serial_ctrl_type)
>
> ...
>
> > +}
>
> While at it, have you tried to use --patience when forming patches? Does it
> make any (better) difference?
I'll check if it makes it more readable. Seems I need to also enable the
--patience option for myself for sending patches to avoid messing up things
like Reviewed-by tags :)
Tony