[PATCH] tty/serial: make early console depend on SERIAL_CORE=y

[PATCH] tty/serial: make early console depend on SERIAL_CORE=y

[Paul Gortmaker]

On an powerpc allmodconfig build, the following is seen:

paul@builder:~/git/linux-head$ make O=../ppc-build -j30 > /dev/null
drivers/built-in.o: In function `.setup_earlycon':
(.init.text+0x5b00): undefined reference to `.uart_parse_earlycon'
make[1]: *** [vmlinux] Error 1

Since uart_parse_earlycon lives in serial_core.c we need to ensure
it is built in if early console is to be used.

Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Jiri Slaby <jslaby@suse.com>
Cc: linux-serial@vger.kernel.org
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---

[A google shows this cropped up in April of this year; I guess it wasn't
 completely fixed?
 http://lists.infradead.org/pipermail/linux-arm-kernel/2015-April/336954.html
 FWIW, I was testing on tty/tty-testing pulled today, from Greg  ]

 drivers/tty/serial/Kconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/tty/serial/Kconfig b/drivers/tty/serial/Kconfig
index 1aec4404062d..a646047e4051 100644
--- a/drivers/tty/serial/Kconfig
+++ b/drivers/tty/serial/Kconfig
@@ -9,6 +9,7 @@ menu "Serial drivers"
 
 config SERIAL_EARLYCON
 	bool
+	depends on SERIAL_CORE=y
 	help
 	  Support for early consoles with the earlycon parameter. This enables
 	  the console before standard serial driver is probed. The console is
-- 
2.6.1

Re: [PATCH] tty/serial: make early console depend on SERIAL_CORE=y

[Arnd Bergmann]

On Sunday 18 October 2015 18:06:48 Paul Gortmaker wrote:
> On an powerpc allmodconfig build, the following is seen:
> 
> paul@builder:~/git/linux-head$ make O=../ppc-build -j30 > /dev/null
> drivers/built-in.o: In function `.setup_earlycon':
> (.init.text+0x5b00): undefined reference to `.uart_parse_earlycon'
> make[1]: *** [vmlinux] Error 1
> 
> Since uart_parse_earlycon lives in serial_core.c we need to ensure
> it is built in if early console is to be used.

I saw the same thing and came up with a different fix

> Cc: Arnd Bergmann <arnd@arndb.de>
> Cc: Peter Hurley <peter@hurleysoftware.com>
> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> Cc: Jiri Slaby <jslaby@suse.com>
> Cc: linux-serial@vger.kernel.org
> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
> ---
> 
> [A google shows this cropped up in April of this year; I guess it wasn't
>  completely fixed?
>  http://lists.infradead.org/pipermail/linux-arm-kernel/2015-April/336954.html
>  FWIW, I was testing on tty/tty-testing pulled today, from Greg  ]

The new problem was clearly caused by 1d59b382f1c4 ("serial: fsl_lpuart:
add earlycon support"), which selects SERIAL_EARLYCON but not
SERIAL_CORE_CONSOLE.

However, the way I read Peter's patch from the URL you provided,
it seems that it is supposed to work without my patch, so I'm not
sure.

	Arnd

>From 8b8704525b47ba3287de2d7ff89a53e3e335f9d8 Mon Sep 17 00:00:00 2001
From: Arnd Bergmann <arnd@arndb.de>
Date: Mon, 19 Oct 2015 13:02:08 +0200
Subject: [PATCH] serial: fsl-lpuart: select SERIAL_CORE_CONSOLE

drivers/built-in.o: In function `setup_earlycon':
:(.init.text+0x2c4c): undefined reference to `uart_parse_earlycon'

Signed-off-by: Arnd Bergmann <arnd@arndb.de>

diff --git a/drivers/tty/serial/Kconfig b/drivers/tty/serial/Kconfig
index 1aec4404062d..f1a4f9eccc44 100644
--- a/drivers/tty/serial/Kconfig
+++ b/drivers/tty/serial/Kconfig
@@ -1540,6 +1540,7 @@ config SERIAL_FSL_LPUART
 	depends on HAS_DMA
 	select SERIAL_CORE
 	select SERIAL_EARLYCON
+	select SERIAL_CORE_CONSOLE
 	help
 	  Support for the on-chip lpuart on some Freescale SOCs.
 

Re: [PATCH] tty/serial: make early console depend on SERIAL_CORE=y

[Paul Gortmaker]

[Re: [PATCH] tty/serial: make early console depend on SERIAL_CORE=y] On 19/10/2015 (Mon 14:56) Arnd Bergmann wrote:

> On Sunday 18 October 2015 18:06:48 Paul Gortmaker wrote:
> > On an powerpc allmodconfig build, the following is seen:
> > 
> > paul@builder:~/git/linux-head$ make O=../ppc-build -j30 > /dev/null
> > drivers/built-in.o: In function `.setup_earlycon':
> > (.init.text+0x5b00): undefined reference to `.uart_parse_earlycon'
> > make[1]: *** [vmlinux] Error 1
> > 
> > Since uart_parse_earlycon lives in serial_core.c we need to ensure
> > it is built in if early console is to be used.
> 
> I saw the same thing and came up with a different fix
> 
> > Cc: Arnd Bergmann <arnd@arndb.de>
> > Cc: Peter Hurley <peter@hurleysoftware.com>
> > Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> > Cc: Jiri Slaby <jslaby@suse.com>
> > Cc: linux-serial@vger.kernel.org
> > Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
> > ---
> > 
> > [A google shows this cropped up in April of this year; I guess it wasn't
> >  completely fixed?
> >  http://lists.infradead.org/pipermail/linux-arm-kernel/2015-April/336954.html
> >  FWIW, I was testing on tty/tty-testing pulled today, from Greg  ]
> 
> The new problem was clearly caused by 1d59b382f1c4 ("serial: fsl_lpuart:
> add earlycon support"), which selects SERIAL_EARLYCON but not
> SERIAL_CORE_CONSOLE.
> 
> However, the way I read Peter's patch from the URL you provided,
> it seems that it is supposed to work without my patch, so I'm not
> sure.
> 
> 	Arnd
> 
> From 8b8704525b47ba3287de2d7ff89a53e3e335f9d8 Mon Sep 17 00:00:00 2001
> From: Arnd Bergmann <arnd@arndb.de>
> Date: Mon, 19 Oct 2015 13:02:08 +0200
> Subject: [PATCH] serial: fsl-lpuart: select SERIAL_CORE_CONSOLE
> 
> drivers/built-in.o: In function `setup_earlycon':
> :(.init.text+0x2c4c): undefined reference to `uart_parse_earlycon'
> 
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> 
> diff --git a/drivers/tty/serial/Kconfig b/drivers/tty/serial/Kconfig
> index 1aec4404062d..f1a4f9eccc44 100644
> --- a/drivers/tty/serial/Kconfig
> +++ b/drivers/tty/serial/Kconfig
> @@ -1540,6 +1540,7 @@ config SERIAL_FSL_LPUART
>  	depends on HAS_DMA
>  	select SERIAL_CORE
>  	select SERIAL_EARLYCON
> +	select SERIAL_CORE_CONSOLE

So the next uart that is not FSL that selects EARLYCON will fail in the
same way.  Isn't it better to have EARLYCON select it or depend on it;
i.e. go to the root of the problem?

Paul.
--

>  	help
>  	  Support for the on-chip lpuart on some Freescale SOCs.
>  
> 

Re: [PATCH] tty/serial: make early console depend on SERIAL_CORE=y

[Peter Hurley]

On 10/19/2015 08:56 AM, Arnd Bergmann wrote:
> On Sunday 18 October 2015 18:06:48 Paul Gortmaker wrote:
[...]
>> [A google shows this cropped up in April of this year; I guess it wasn't
>>  completely fixed?
>>  http://lists.infradead.org/pipermail/linux-arm-kernel/2015-April/336954.html
>>  FWIW, I was testing on tty/tty-testing pulled today, from Greg  ]
> 
> The new problem was clearly caused by 1d59b382f1c4 ("serial: fsl_lpuart:
> add earlycon support"), which selects SERIAL_EARLYCON but not
> SERIAL_CORE_CONSOLE.
> 
> However, the way I read Peter's patch from the URL you provided,
> it seems that it is supposed to work without my patch, so I'm not
> sure.

Looks like my patch never got applied; I'll resend it.

Regards,
Peter Hurley