[PATCH 1/2] serial: 8250: fix building mediatek with SERIAL_8250=m

[PATCH 1/2] serial: 8250: fix building mediatek with SERIAL_8250=m

[Arnd Bergmann]

The Mediatek 8250 driver has a 'bool' Kconfig symbol, but that
breaks when SERIAL_8250 is a loadable module:

drivers/tty/built-in.o: In function `mtk8250_set_termios':
8250_ingenic.c:(.text+0x1bee8): undefined reference to `serial8250_do_set_termios'
8250_ingenic.c:(.text+0x1bf10): undefined reference to `uart_get_baud_rate'
8250_ingenic.c:(.text+0x1c09c): undefined reference to `uart_get_divisor'
drivers/tty/built-in.o: In function `mtk8250_do_pm':
8250_ingenic.c:(.text+0x1c0d0): undefined reference to `serial8250_do_pm'
drivers/tty/built-in.o: In function `mtk8250_probe':
8250_ingenic.c:(.text+0x1c2e4): undefined reference to `serial8250_register_8250_port'
serial/8250/8250_mtk.c:287:242: error: data definition has no type or storage class [-Werror]
serial/8250/8250_mtk.c:287:122: error: 'mtk8250_platform_driver_init' defined but not used [-Werror=unused-function]

This changes the symbol to a 'tristate', so the dependency on
SERIAL_8250 also works when that is set to 'm'.
To actually build the driver, we also need to include <linux/module.h>.

Note that the driver uses builtin_platform_driver() and implements
no .remove() callback, so unloading the module is not possible.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
 drivers/tty/serial/8250/8250_mtk.c | 1 +
 drivers/tty/serial/8250/Kconfig    | 2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/tty/serial/8250/8250_mtk.c b/drivers/tty/serial/8250/8250_mtk.c
index 9038843cadc7..7e9242a5ddfe 100644
--- a/drivers/tty/serial/8250/8250_mtk.c
+++ b/drivers/tty/serial/8250/8250_mtk.c
@@ -19,6 +19,7 @@
 #include <linux/init.h>
 #include <linux/of_irq.h>
 #include <linux/of_platform.h>
+#include <linux/module.h>
 #include <linux/platform_device.h>
 #include <linux/pm_runtime.h>
 #include <linux/serial_8250.h>
diff --git a/drivers/tty/serial/8250/Kconfig b/drivers/tty/serial/8250/Kconfig
index 67ad6b0d595b..6ae0fae4f796 100644
--- a/drivers/tty/serial/8250/Kconfig
+++ b/drivers/tty/serial/8250/Kconfig
@@ -370,7 +370,7 @@ config SERIAL_8250_LPC18XX
 	  serial port, say Y to this option. If unsure, say Y.
 
 config SERIAL_8250_MT6577
-	bool "Mediatek serial port support"
+	tristate "Mediatek serial port support"
 	depends on SERIAL_8250 && ARCH_MEDIATEK
 	help
 	  If you have a Mediatek based board and want to use the
-- 
2.7.0

[PATCH 2/2] serial: 8250: fix building Ingenic with SERIAL_8250=m

[Arnd Bergmann]

The Ingenic 8250 driver has a 'bool' Kconfig symbol, but that
breaks when SERIAL_8250 is a loadable module:

drivers/tty/built-in.o: In function `ingenic_uart_probe':
8250_ingenic.c:(.text+0x1c1a0): undefined reference to `serial8250_register_8250_port'

This changes the symbol to a 'tristate', plus a dependency on
SERIAL_8250, which makes it work again. Unlike the other
soc-specific backends, this one has no dependency on an
architecture or a platform. I'm adding a dependency on
MIPS || COMPILE_TEST as well here, to avoid showing the driver
on architectures that are not interested in it.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
 drivers/tty/serial/8250/Kconfig | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/tty/serial/8250/Kconfig b/drivers/tty/serial/8250/Kconfig
index 6ae0fae4f796..13c63b14179f 100644
--- a/drivers/tty/serial/8250/Kconfig
+++ b/drivers/tty/serial/8250/Kconfig
@@ -384,8 +384,8 @@ config SERIAL_8250_UNIPHIER
 	  serial ports, say Y to this option. If unsure, say N.
 
 config SERIAL_8250_INGENIC
-	bool "Support for Ingenic SoC serial ports"
-	depends on OF_FLATTREE
+	tristate "Support for Ingenic SoC serial ports"
+	depends on SERIAL_8250 && OF_FLATTREE && (MIPS || COMPILE_TEST)
 	select LIBFDT
 	help
 	  If you have a system using an Ingenic SoC and wish to make use of
-- 
2.7.0

[PATCH 2/2] serial: 8250: fix building Ingenic with SERIAL_8250=m

[Paul Gortmaker]

[[PATCH 2/2] serial: 8250: fix building Ingenic with SERIAL_8250=m] On 08/02/2016 (Mon 13:51) Arnd Bergmann wrote:

> The Ingenic 8250 driver has a 'bool' Kconfig symbol, but that
> breaks when SERIAL_8250 is a loadable module:
> 
> drivers/tty/built-in.o: In function `ingenic_uart_probe':
> 8250_ingenic.c:(.text+0x1c1a0): undefined reference to `serial8250_register_8250_port'
> 
> This changes the symbol to a 'tristate', plus a dependency on

Note that Greg just queued a patch yesterday (but sent in 2015) that
removed what was the dead modular code.

It could also depend on SERIAL_8250=y to resolve the build issue you
found, if nobody cares about having modular SOC console UART drivers.

Either way, I don't have issue with it being tristate vs bool, but if
you do go with tristate, then my patch will have to be dropped/reverted.

Paul.
--

> SERIAL_8250, which makes it work again. Unlike the other
> soc-specific backends, this one has no dependency on an
> architecture or a platform. I'm adding a dependency on
> MIPS || COMPILE_TEST as well here, to avoid showing the driver
> on architectures that are not interested in it.
> 
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
>  drivers/tty/serial/8250/Kconfig | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/tty/serial/8250/Kconfig b/drivers/tty/serial/8250/Kconfig
> index 6ae0fae4f796..13c63b14179f 100644
> --- a/drivers/tty/serial/8250/Kconfig
> +++ b/drivers/tty/serial/8250/Kconfig
> @@ -384,8 +384,8 @@ config SERIAL_8250_UNIPHIER
>  	  serial ports, say Y to this option. If unsure, say N.
>  
>  config SERIAL_8250_INGENIC
> -	bool "Support for Ingenic SoC serial ports"
> -	depends on OF_FLATTREE
> +	tristate "Support for Ingenic SoC serial ports"
> +	depends on SERIAL_8250 && OF_FLATTREE && (MIPS || COMPILE_TEST)
>  	select LIBFDT
>  	help
>  	  If you have a system using an Ingenic SoC and wish to make use of
> -- 
> 2.7.0
> 

[PATCH 1/2] serial: 8250: fix building mediatek with SERIAL_8250=m

[Paul Gortmaker]

[[PATCH 1/2] serial: 8250: fix building mediatek with SERIAL_8250=m] On 08/02/2016 (Mon 13:51) Arnd Bergmann wrote:

> The Mediatek 8250 driver has a 'bool' Kconfig symbol, but that
> breaks when SERIAL_8250 is a loadable module:
> 
> drivers/tty/built-in.o: In function `mtk8250_set_termios':
> 8250_ingenic.c:(.text+0x1bee8): undefined reference to `serial8250_do_set_termios'
> 8250_ingenic.c:(.text+0x1bf10): undefined reference to `uart_get_baud_rate'
> 8250_ingenic.c:(.text+0x1c09c): undefined reference to `uart_get_divisor'
> drivers/tty/built-in.o: In function `mtk8250_do_pm':
> 8250_ingenic.c:(.text+0x1c0d0): undefined reference to `serial8250_do_pm'
> drivers/tty/built-in.o: In function `mtk8250_probe':
> 8250_ingenic.c:(.text+0x1c2e4): undefined reference to `serial8250_register_8250_port'
> serial/8250/8250_mtk.c:287:242: error: data definition has no type or storage class [-Werror]
> serial/8250/8250_mtk.c:287:122: error: 'mtk8250_platform_driver_init' defined but not used [-Werror=unused-function]

Probably should drop the ingenic lines above ; assuming they are make -j N
related and that mtk doesn't depend on or build ingenic.

> 
> This changes the symbol to a 'tristate', so the dependency on
> SERIAL_8250 also works when that is set to 'm'.
> To actually build the driver, we also need to include <linux/module.h>.
> 
> Note that the driver uses builtin_platform_driver() and implements
> no .remove() callback, so unloading the module is not possible.

But this is due to d72d391c126e0ffd3047c06c4bef4d795853d5d5.  So again,
I'm not sure what this achives vs. just making it depend on having
SERIAL_8250=y as a trivial fix for the one build issue.

If you really see a use case for tristate, then the above should be
reverted before adding your conversion to tristate I think.

Paul.
--

> 
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
>  drivers/tty/serial/8250/8250_mtk.c | 1 +
>  drivers/tty/serial/8250/Kconfig    | 2 +-
>  2 files changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/tty/serial/8250/8250_mtk.c b/drivers/tty/serial/8250/8250_mtk.c
> index 9038843cadc7..7e9242a5ddfe 100644
> --- a/drivers/tty/serial/8250/8250_mtk.c
> +++ b/drivers/tty/serial/8250/8250_mtk.c
> @@ -19,6 +19,7 @@
>  #include <linux/init.h>
>  #include <linux/of_irq.h>
>  #include <linux/of_platform.h>
> +#include <linux/module.h>
>  #include <linux/platform_device.h>
>  #include <linux/pm_runtime.h>
>  #include <linux/serial_8250.h>
> diff --git a/drivers/tty/serial/8250/Kconfig b/drivers/tty/serial/8250/Kconfig
> index 67ad6b0d595b..6ae0fae4f796 100644
> --- a/drivers/tty/serial/8250/Kconfig
> +++ b/drivers/tty/serial/8250/Kconfig
> @@ -370,7 +370,7 @@ config SERIAL_8250_LPC18XX
>  	  serial port, say Y to this option. If unsure, say Y.
>  
>  config SERIAL_8250_MT6577
> -	bool "Mediatek serial port support"
> +	tristate "Mediatek serial port support"
>  	depends on SERIAL_8250 && ARCH_MEDIATEK
>  	help
>  	  If you have a Mediatek based board and want to use the
> -- 
> 2.7.0
> 

[PATCH 2/2] serial: 8250: fix building Ingenic with SERIAL_8250=m

[Arnd Bergmann]

On Monday 08 February 2016 11:45:44 Paul Gortmaker wrote:
> [[PATCH 2/2] serial: 8250: fix building Ingenic with SERIAL_8250=m] On 08/02/2016 (Mon 13:51) Arnd Bergmann wrote:
> 
> > The Ingenic 8250 driver has a 'bool' Kconfig symbol, but that
> > breaks when SERIAL_8250 is a loadable module:
> > 
> > drivers/tty/built-in.o: In function `ingenic_uart_probe':
> > 8250_ingenic.c:(.text+0x1c1a0): undefined reference to `serial8250_register_8250_port'
> > 
> > This changes the symbol to a 'tristate', plus a dependency on
> 
> Note that Greg just queued a patch yesterday (but sent in 2015) that
> removed what was the dead modular code.
> 
> It could also depend on SERIAL_8250=y to resolve the build issue you
> found, if nobody cares about having modular SOC console UART drivers.

All other UART drivers can be modules, no need for this to be an
exception I think. Having it depend on SERIAL_8250=y would make the
driver disabled on 'allmodconfig', which is a bit counterproductive for
the purpose of allmodconfig. I don't think anyone actually cares about
the driver being runtime loaded on real hardware.

> Either way, I don't have issue with it being tristate vs bool, but if
> you do go with tristate, then my patch will have to be dropped/reverted.

Fine with me too.

I also now saw in the randconfig builds that the driver is missing an #include
<linux/module.h>, which I guess is going to be there after reverting your
patch.

I'll send a new series.

	Arnd

[PATCH 1/2] serial: 8250: fix building mediatek with SERIAL_8250=m

[Matthias Brugger]

On 08/02/16 13:51, Arnd Bergmann wrote:
> The Mediatek 8250 driver has a 'bool' Kconfig symbol, but that
> breaks when SERIAL_8250 is a loadable module:
>
> drivers/tty/built-in.o: In function `mtk8250_set_termios':
> 8250_ingenic.c:(.text+0x1bee8): undefined reference to `serial8250_do_set_termios'
> 8250_ingenic.c:(.text+0x1bf10): undefined reference to `uart_get_baud_rate'
> 8250_ingenic.c:(.text+0x1c09c): undefined reference to `uart_get_divisor'
> drivers/tty/built-in.o: In function `mtk8250_do_pm':
> 8250_ingenic.c:(.text+0x1c0d0): undefined reference to `serial8250_do_pm'
> drivers/tty/built-in.o: In function `mtk8250_probe':
> 8250_ingenic.c:(.text+0x1c2e4): undefined reference to `serial8250_register_8250_port'
> serial/8250/8250_mtk.c:287:242: error: data definition has no type or storage class [-Werror]
> serial/8250/8250_mtk.c:287:122: error: 'mtk8250_platform_driver_init' defined but not used [-Werror=unused-function]
>
> This changes the symbol to a 'tristate', so the dependency on
> SERIAL_8250 also works when that is set to 'm'.
> To actually build the driver, we also need to include <linux/module.h>.
>
> Note that the driver uses builtin_platform_driver() and implements
> no .remove() callback, so unloading the module is not possible.
>
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---

I tried this with:
commit 2178cbc68f3602dc0b5949b9be2c8383ad3d93ef

$ grep 8250 .config
CONFIG_SERIAL_8250=m
# CONFIG_SERIAL_8250_DEPRECATED_OPTIONS is not set
CONFIG_SERIAL_8250_PCI=m
CONFIG_SERIAL_8250_NR_UARTS=4
CONFIG_SERIAL_8250_RUNTIME_UARTS=4
# CONFIG_SERIAL_8250_EXTENDED is not set
# CONFIG_SERIAL_8250_DW is not set
# CONFIG_SERIAL_8250_RT288X is not set
CONFIG_SERIAL_8250_MT6577=y
# CONFIG_SERIAL_8250_INGENIC is not set
# CONFIG_SERIAL_8250_MID is not set

$ make ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu- drivers/tty/

compiles just fine.

Do I missing something?

Regards,
Matthias

[PATCH 1/2] serial: 8250: fix building mediatek with SERIAL_8250=m

[Arnd Bergmann]

On Wednesday 10 February 2016 19:06:27 Matthias Brugger wrote:
> > ---
> 
> I tried this with:
> commit 2178cbc68f3602dc0b5949b9be2c8383ad3d93ef
> 
> $ grep 8250 .config
> CONFIG_SERIAL_8250=m
> # CONFIG_SERIAL_8250_DEPRECATED_OPTIONS is not set
> CONFIG_SERIAL_8250_PCI=m
> CONFIG_SERIAL_8250_NR_UARTS=4
> CONFIG_SERIAL_8250_RUNTIME_UARTS=4
> # CONFIG_SERIAL_8250_EXTENDED is not set
> # CONFIG_SERIAL_8250_DW is not set
> # CONFIG_SERIAL_8250_RT288X is not set
> CONFIG_SERIAL_8250_MT6577=y
> # CONFIG_SERIAL_8250_INGENIC is not set
> # CONFIG_SERIAL_8250_MID is not set
> 
> $ make ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu- drivers/tty/
> 
> compiles just fine.
> 
> Do I missing something?
> 

I was simplifying the situation in the patch description: When CONFIG_SERIAL_8250=m,
Kbuild does not actually descend into drivers/tty/serial/8250/ while building
built-in code, so none of the built-in subdrivers actually get built.

After spotting the problem, I was building with this patch that I considered
submitting:

diff --git a/drivers/tty/serial/Makefile b/drivers/tty/serial/Makefile
index ceba33c4ebb4..28121885e78e 100644
--- a/drivers/tty/serial/Makefile
+++ b/drivers/tty/serial/Makefile
@@ -18,7 +18,7 @@ obj-$(CONFIG_SERIAL_SUNSU) += sunsu.o
 obj-$(CONFIG_SERIAL_SUNSAB) += sunsab.o
 
 # Now bring in any enabled 8250/16450/16550 type drivers.
-obj-$(CONFIG_SERIAL_8250) += 8250/
+obj-y += 8250/
 
 obj-$(CONFIG_SERIAL_AMBA_PL010) += amba-pl010.o
 obj-$(CONFIG_SERIAL_AMBA_PL011) += amba-pl011.o

However, after the other problems are fixed, it's not actually needed
any more, so I did not submit it.

	Arnd

[PATCH 1/2] serial: 8250: fix building mediatek with SERIAL_8250=m

[Arnd Bergmann]

On Monday 08 February 2016 11:55:35 Paul Gortmaker wrote:
> > drivers/tty/built-in.o: In function `mtk8250_set_termios':
> > 8250_ingenic.c:(.text+0x1bee8): undefined reference to `serial8250_do_set_termios'
> > 8250_ingenic.c:(.text+0x1bf10): undefined reference to `uart_get_baud_rate'
> > 8250_ingenic.c:(.text+0x1c09c): undefined reference to `uart_get_divisor'
> > drivers/tty/built-in.o: In function `mtk8250_do_pm':
> > 8250_ingenic.c:(.text+0x1c0d0): undefined reference to `serial8250_do_pm'
> > drivers/tty/built-in.o: In function `mtk8250_probe':
> > 8250_ingenic.c:(.text+0x1c2e4): undefined reference to `serial8250_register_8250_port'
> > serial/8250/8250_mtk.c:287:242: error: data definition has no type or storage class [-Werror]
> > serial/8250/8250_mtk.c:287:122: error: 'mtk8250_platform_driver_init' defined but not used [-Werror=unused-function]
> 
> Probably should drop the ingenic lines above ; assuming they are make -j N
> related and that mtk doesn't depend on or build ingenic.

I looked at these again and noticed that this is just being misreported
by the linker. I don't know exactly what happens here but I see it a lot:
the function names ("mtk8250_set_termios") are all correct, but the file
names ("8250_ingenic.c") are wrong. I normally edit out the file names
manually but forgot that here. I'll fix it up when resending.

	Arnd