| Message ID | 1349347191-7406-1-git-send-email-s.hauer@pengutronix.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 10/04/2012 05:39 AM, Sascha Hauer wrote: > devicetrees may have the linux,stdout-path property to specify the > console. This patch adds support to the i.MX serial driver for this. > > Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de> > --- > > I was originally looking for a more generic way to handle this, but > since a struct console has no device associated to it, it's not > possible to match a console with a device in a generic way. So we Could we add a device ptr to struct console? > just call add_preferred_console from the driver and let it go down > to a string matching in the console code. > If anyone has a better idea how to handle this, please let me know. > Otherwise I'm happy to see this patch applied aswell. > > Thanks > Sascha > > drivers/tty/serial/imx.c | 29 +++++++++++++++++++++++++++++ > 1 file changed, 29 insertions(+) > > diff --git a/drivers/tty/serial/imx.c b/drivers/tty/serial/imx.c > index e309e8b..b52c4a7 100644 > --- a/drivers/tty/serial/imx.c > +++ b/drivers/tty/serial/imx.c > @@ -1397,6 +1397,32 @@ static int serial_imx_resume(struct platform_device *dev) > > #ifdef CONFIG_OF > /* > + * Check if this device matches the linux,stdout-path property > + * in the chosen node. return true if yes, false otherwise > + */ > +static int serial_imx_is_stdoutpath(struct platform_device *pdev) Couldn't this function be generic? Just move IS_ENABLED(CONFIG_SERIAL_IMX_CONSOLE) outside this function. Arguably, the presence of the property or not could replace the config option all together. And use a struct device so this can work with devices other than platform devices (i.e. amba). Rob > +{ > + struct device_node *dn; > + const char *name; > + > + if (!IS_ENABLED(CONFIG_SERIAL_IMX_CONSOLE)) > + return 0; > + > + name = of_get_property(of_chosen, "linux,stdout-path", NULL); > + if (name == NULL) > + return 0; > + > + dn = of_find_node_by_path(name); > + if (!dn) > + return 0; > + > + if (dn == pdev->dev.of_node) > + return 1; > + > + return 0; > +} > + > +/* > * This function returns 1 iff pdev isn't a device instatiated by dt, 0 iff it > * could successfully get all information from dt or a negative errno. > */ > @@ -1427,6 +1453,9 @@ static int serial_imx_probe_dt(struct imx_port *sport, > > sport->devdata = of_id->data; > > + if (serial_imx_is_stdoutpath(pdev)) > + add_preferred_console(imx_reg.cons->name, sport->port.line, 0); > + > return 0; > } > #else >
On Thu, Oct 04, 2012 at 07:44:45AM -0500, Rob Herring wrote: > On 10/04/2012 05:39 AM, Sascha Hauer wrote: > > devicetrees may have the linux,stdout-path property to specify the > > console. This patch adds support to the i.MX serial driver for this. > > > > Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de> > > --- > > > > I was originally looking for a more generic way to handle this, but > > since a struct console has no device associated to it, it's not > > possible to match a console with a device in a generic way. So we > > Could we add a device ptr to struct console? I think not. Looking at some drivers they register a console very early when there is no device available (even if there will be a device later when the uart driver probes) > > #ifdef CONFIG_OF > > /* > > + * Check if this device matches the linux,stdout-path property > > + * in the chosen node. return true if yes, false otherwise > > + */ > > +static int serial_imx_is_stdoutpath(struct platform_device *pdev) > > Couldn't this function be generic? Just move > IS_ENABLED(CONFIG_SERIAL_IMX_CONSOLE) outside this function. Arguably, > the presence of the property or not could replace the config option all > together. > > And use a struct device so this can work with devices other than > platform devices (i.e. amba). Ok, makes sense. Sascha
diff --git a/drivers/tty/serial/imx.c b/drivers/tty/serial/imx.c index e309e8b..b52c4a7 100644 --- a/drivers/tty/serial/imx.c +++ b/drivers/tty/serial/imx.c @@ -1397,6 +1397,32 @@ static int serial_imx_resume(struct platform_device *dev) #ifdef CONFIG_OF /* + * Check if this device matches the linux,stdout-path property + * in the chosen node. return true if yes, false otherwise + */ +static int serial_imx_is_stdoutpath(struct platform_device *pdev) +{ + struct device_node *dn; + const char *name; + + if (!IS_ENABLED(CONFIG_SERIAL_IMX_CONSOLE)) + return 0; + + name = of_get_property(of_chosen, "linux,stdout-path", NULL); + if (name == NULL) + return 0; + + dn = of_find_node_by_path(name); + if (!dn) + return 0; + + if (dn == pdev->dev.of_node) + return 1; + + return 0; +} + +/* * This function returns 1 iff pdev isn't a device instatiated by dt, 0 iff it * could successfully get all information from dt or a negative errno. */ @@ -1427,6 +1453,9 @@ static int serial_imx_probe_dt(struct imx_port *sport, sport->devdata = of_id->data; + if (serial_imx_is_stdoutpath(pdev)) + add_preferred_console(imx_reg.cons->name, sport->port.line, 0); + return 0; } #else
devicetrees may have the linux,stdout-path property to specify the console. This patch adds support to the i.MX serial driver for this. Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de> --- I was originally looking for a more generic way to handle this, but since a struct console has no device associated to it, it's not possible to match a console with a device in a generic way. So we just call add_preferred_console from the driver and let it go down to a string matching in the console code. If anyone has a better idea how to handle this, please let me know. Otherwise I'm happy to see this patch applied aswell. Thanks Sascha drivers/tty/serial/imx.c | 29 +++++++++++++++++++++++++++++ 1 file changed, 29 insertions(+)