linux-next: sparc tree build failure

linux-next: sparc tree build failure

[Stephen Rothwell]

Hi Dave,

Today's linux-next build (sparc64 defconfig) failed like this:

arch/sparc/oprofile/init.c: In function 'timer_start':
arch/sparc/oprofile/init.c:46: error: 'HZ' undeclared (first use in this function)

I have no idea why this just cropped up, but I have added the following
patch for today.

From: Stephen Rothwell <sfr@canb.auug.org.au>
Date: Mon, 5 Oct 2009 18:32:33 +1100
Subject: [PATCH] sparc: using HZ needs an include of linux/param.h

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
---
 arch/sparc/oprofile/init.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/arch/sparc/oprofile/init.c b/arch/sparc/oprofile/init.c
index f97cb8b..62f4120 100644
--- a/arch/sparc/oprofile/init.c
+++ b/arch/sparc/oprofile/init.c
@@ -11,7 +11,8 @@
 #include <linux/oprofile.h>
 #include <linux/errno.h>
 #include <linux/init.h>
+#include <linux/param.h>	/* for HZ */
  
 #ifdef CONFIG_SPARC64
 #include <linux/notifier.h>
 #include <linux/rcupdate.h>
-- 
1.6.4.3


-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

Re: linux-next: sparc tree build failure

[David Miller]

From: Stephen Rothwell <sfr@canb.auug.org.au>
Date: Mon, 5 Oct 2009 18:43:57 +1100

> Today's linux-next build (sparc64 defconfig) failed like this:
> 
> arch/sparc/oprofile/init.c: In function 'timer_start':
> arch/sparc/oprofile/init.c:46: error: 'HZ' undeclared (first use in this function)
> 
> I have no idea why this just cropped up, but I have added the following
> patch for today.

I suspect:

commit a99bbaf5ee6bad1aca0c88ea65ec6e5373e86184
Author: Alexey Dobriyan <adobriyan@gmail.com>
Date:   Sun Oct 4 16:11:37 2009 +0400

    headers: remove sched.h from poll.h
    
    Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
    Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

I'll add your fix to sparc-2.6, thanks!

linux-next: sparc tree build failure

[Stephen Rothwell]

Hi Dave,

Today's linux-next build (powerpc allyesconfig) failed like this:

drivers/serial/apbuart.c:32:23: error: asm/oplib.h: No such file or directory
drivers/serial/apbuart.c:586: error: variable 'grlib_apbuart_of_driver' has initializer but incomplete type
drivers/serial/apbuart.c:587: error: unknown field 'match_table' specified in initializer
drivers/serial/apbuart.c:587: warning: excess elements in struct initializer
drivers/serial/apbuart.c:587: warning: (near initialization for 'grlib_apbuart_of_driver')
drivers/serial/apbuart.c:588: error: unknown field 'probe' specified in initializer
drivers/serial/apbuart.c:588: warning: excess elements in struct initializer
drivers/serial/apbuart.c:588: warning: (near initialization for 'grlib_apbuart_of_driver')
drivers/serial/apbuart.c:589: error: unknown field 'driver' specified in initializer
drivers/serial/apbuart.c:589: error: extra brace group at end of initializer
drivers/serial/apbuart.c:589: error: (near initialization for 'grlib_apbuart_of_driver')
drivers/serial/apbuart.c:592: warning: excess elements in struct initializer
drivers/serial/apbuart.c:592: warning: (near initialization for 'grlib_apbuart_of_driver')
drivers/serial/apbuart.c: In function 'grlib_apbuart_configure':
drivers/serial/apbuart.c:613: error: implicit declaration of function 'prom_getchild'
drivers/serial/apbuart.c:613: error: 'prom_root_node' undeclared (first use in this function)
drivers/serial/apbuart.c:613: error: (Each undeclared identifier is reported only once
drivers/serial/apbuart.c:613: error: for each function it appears in.)
drivers/serial/apbuart.c:614: error: implicit declaration of function 'prom_getint'
drivers/serial/apbuart.c: In function 'grlib_apbuart_init':
drivers/serial/apbuart.c:679: error: implicit declaration of function 'of_register_driver'
drivers/serial/apbuart.c:679: error: 'of_platform_bus_type' undeclared (first use in this function)

Caused by commit d4ac42a582e46d7f86f0acb4253a310423c72c4c ("sparc:
Support for GRLIB APBUART serial port").  I added the following patch for
today.

From: Stephen Rothwell <sfr@canb.auug.org.au>
Date: Thu, 5 Nov 2009 18:40:55 +1100
Subject: [PATCH] sparc: the GRLIB APBUART serial port will only build on sparc

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
---
 drivers/serial/Kconfig |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/drivers/serial/Kconfig b/drivers/serial/Kconfig
index 50943ff..7ec806d 100644
--- a/drivers/serial/Kconfig
+++ b/drivers/serial/Kconfig
@@ -1480,6 +1480,7 @@ config SERIAL_BCM63XX_CONSOLE
 config SERIAL_GRLIB_GAISLER_APBUART
 	tristate "GRLIB APBUART serial support"
 	depends on OF
+	depends on SPARC
 	---help---
 	Add support for the GRLIB APBUART serial port.
 
-- 
1.6.5.2


-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

Re: linux-next: sparc tree build failure

[David Miller]

From: Stephen Rothwell <sfr@canb.auug.org.au>
Date: Thu, 5 Nov 2009 18:50:14 +1100

> Caused by commit d4ac42a582e46d7f86f0acb4253a310423c72c4c ("sparc:
> Support for GRLIB APBUART serial port").  I added the following patch for
> today.

Thanks, the real fix is to simply remove that include, it isn't
actually needed and the CONFIG_OF dependency is sufficient.

Re: linux-next: sparc tree build failure

[David Miller]

From: David Miller <davem@davemloft.net>
Date: Wed, 04 Nov 2009 23:53:15 -0800 (PST)

> From: Stephen Rothwell <sfr@canb.auug.org.au>
> Date: Thu, 5 Nov 2009 18:50:14 +1100
> 
>> Caused by commit d4ac42a582e46d7f86f0acb4253a310423c72c4c ("sparc:
>> Support for GRLIB APBUART serial port").  I added the following patch for
>> today.
> 
> Thanks, the real fix is to simply remove that include, it isn't
> actually needed and the CONFIG_OF dependency is sufficient.

Hmm, it took a tiny bit more than that :-)  Here's what I commited
to fix this, thanks.

apbuart: Kill dependency on deprecated Sparc-only PROM interfaces.

Use the proper modern OF ones instead.

Noticed by Stephen Rothwell.

Signed-off-by: David S. Miller <davem@davemloft.net>
---
 drivers/serial/apbuart.c |   12 ++++++------
 1 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/drivers/serial/apbuart.c b/drivers/serial/apbuart.c
index c7883a3..5f9dec3 100644
--- a/drivers/serial/apbuart.c
+++ b/drivers/serial/apbuart.c
@@ -29,7 +29,6 @@
 #include <linux/io.h>
 #include <linux/serial_core.h>
 #include <asm/irq.h>
-#include <asm/oplib.h>
 
 #include "apbuart.h"
 
@@ -596,10 +595,9 @@ static struct of_platform_driver grlib_apbuart_of_driver = {
 static void grlib_apbuart_configure(void)
 {
 	static int enum_done;
-	struct device_node *np;
+	struct device_node *np, *rp;
 	struct uart_port *port = NULL;
-
-	int node;
+	const u32 *prop;
 	int freq_khz;
 	int v = 0, d = 0;
 	unsigned int addr;
@@ -610,8 +608,10 @@ static void grlib_apbuart_configure(void)
 		return;
 
 	/* Get bus frequency */
-	node = prom_getchild(prom_root_node);
-	freq_khz = prom_getint(node, "clock-frequency");
+	rp = of_find_node_by_name(NULL, "/");
+	rp = of_get_next_child(rp, NULL);
+	prop = of_get_property(rp, "clock-frequency", NULL);
+	freq_khz = *prop;
 
 	line = 0;
 	for_each_matching_node(np, apbuart_match) {
-- 
1.6.5.2

Re: linux-next: sparc tree build failure

[Kristoffer Glembo]

David Miller wrote:
> 
> Hmm, it took a tiny bit more than that :-)  Here's what I commited
> to fix this, thanks.
> 
> apbuart: Kill dependency on deprecated Sparc-only PROM interfaces.
> 
> Use the proper modern OF ones instead.
> 
> Noticed by Stephen Rothwell.
> 
> Signed-off-by: David S. Miller <davem@davemloft.net>

Thanks guys. Sorry for the obsolete code :-)

Best regards,
Kristoffer

linux-next: sparc tree build failure

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1963 bytes --]

Hi Dave,

Today's linux-next build (powerpc allyesconfig) failed like this:

drivers/serial/apbuart.c:585: error: variable 'grlib_apbuart_of_driver' has initializer but incomplete type
drivers/serial/apbuart.c:586: error: unknown field 'match_table' specified in initializer
drivers/serial/apbuart.c:586: warning: excess elements in struct initializer
drivers/serial/apbuart.c:586: warning: (near initialization for 'grlib_apbuart_of_driver')
drivers/serial/apbuart.c:587: error: unknown field 'probe' specified in initializer
drivers/serial/apbuart.c:587: warning: excess elements in struct initializer
drivers/serial/apbuart.c:587: warning: (near initialization for 'grlib_apbuart_of_driver')
drivers/serial/apbuart.c:588: error: unknown field 'driver' specified in initializer
drivers/serial/apbuart.c:588: error: extra brace group at end of initializer
drivers/serial/apbuart.c:588: error: (near initialization for 'grlib_apbuart_of_driver')
drivers/serial/apbuart.c:591: warning: excess elements in struct initializer
drivers/serial/apbuart.c:591: warning: (near initialization for 'grlib_apbuart_of_driver')
drivers/serial/apbuart.c: In function 'grlib_apbuart_init':
drivers/serial/apbuart.c:679: error: implicit declaration of function 'of_register_driver'
drivers/serial/apbuart.c:679: error: 'of_platform_bus_type' undeclared (first use in this function)

Caused by commit d4ac42a582e46d7f86f0acb4253a310423c72c4c ("sparc:
Support for GRLIB APBUART serial port") (yes again :-)).  I have applied
my patch to make it only build on SPARC again ...

(Including linux/of_platform.h - where struct of_platform_driver,
of_register_driver() and of_platform_bus_type are declared - would be a
good start ...  and the code should probably be using
of_register_platform_driver() instead of of_register_driver() in any
case.)

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: linux-next: sparc tree build failure

[David Miller]

From: Stephen Rothwell <sfr@canb.auug.org.au>
Date: Fri, 6 Nov 2009 18:02:48 +1100

> Caused by commit d4ac42a582e46d7f86f0acb4253a310423c72c4c ("sparc:
> Support for GRLIB APBUART serial port") (yes again :-)).  I have applied
> my patch to make it only build on SPARC again ...
> 
> (Including linux/of_platform.h - where struct of_platform_driver,
> of_register_driver() and of_platform_bus_type are declared - would be a
> good start ...  and the code should probably be using
> of_register_platform_driver() instead of of_register_driver() in any
> case.)

Sorry, sparc is getting those headers indirectly.

I'll fix this up.

Re: linux-next: sparc tree build failure

[David Miller]

From: David Miller <davem@davemloft.net>
Date: Fri, 06 Nov 2009 00:19:30 -0800 (PST)

> From: Stephen Rothwell <sfr@canb.auug.org.au>
> Date: Fri, 6 Nov 2009 18:02:48 +1100
> 
>> Caused by commit d4ac42a582e46d7f86f0acb4253a310423c72c4c ("sparc:
>> Support for GRLIB APBUART serial port") (yes again :-)).  I have applied
>> my patch to make it only build on SPARC again ...
>> 
>> (Including linux/of_platform.h - where struct of_platform_driver,
>> of_register_driver() and of_platform_bus_type are declared - would be a
>> good start ...  and the code should probably be using
>> of_register_platform_driver() instead of of_register_driver() in any
>> case.)
> 
> Sorry, sparc is getting those headers indirectly.
> 
> I'll fix this up.

Stephen, can you let me know if this takes care of everything?

I also noticed that the of_driver was not being unregistered on
module unload.

apbuart: Fix build and missing driver unregister.

linux/of_platform.h needs to be included

OF driver needs to be unregistered

Signed-off-by: David S. Miller <davem@davemloft.net>

Re: linux-next: sparc tree build failure

[David Miller]

From: David Miller <davem@davemloft.net>
Date: Fri, 06 Nov 2009 00:26:51 -0800 (PST)

> apbuart: Fix build and missing driver unregister.
> 
> linux/of_platform.h needs to be included
> 
> OF driver needs to be unregistered
> 
> Signed-off-by: David S. Miller <davem@davemloft.net>

Sorry left out the patch :-)

Here it is:

diff --git a/drivers/serial/apbuart.c b/drivers/serial/apbuart.c
index 5f9dec3..a1e9503 100644
--- a/drivers/serial/apbuart.c
+++ b/drivers/serial/apbuart.c
@@ -25,6 +25,7 @@
 #include <linux/device.h>
 #include <linux/of.h>
 #include <linux/of_device.h>
+#include <linux/of_platform.h>
 #include <linux/platform_device.h>
 #include <linux/io.h>
 #include <linux/serial_core.h>
@@ -676,8 +677,7 @@ static int __init grlib_apbuart_init(void)
 		return ret;
 	}
 
-	ret = of_register_driver(&grlib_apbuart_of_driver, &of_platform_bus_type);
-
+	ret = of_register_platform_driver(&grlib_apbuart_of_driver);
 	if (ret) {
 		printk(KERN_ERR
 		       "%s: of_register_platform_driver failed (%i)\n",
@@ -698,7 +698,7 @@ static void __exit grlib_apbuart_exit(void)
 				     &grlib_apbuart_ports[i]);
 
 	uart_unregister_driver(&grlib_apbuart_driver);
-
+	of_unregister_platform_driver(&grlib_apbuart_of_driver);
 }
 
 module_init(grlib_apbuart_init);
-- 
1.6.5.2

Re: linux-next: sparc tree build failure

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 378 bytes --]

Hi Dave,

On Fri, 06 Nov 2009 00:26:51 -0800 (PST) David Miller <davem@davemloft.net> wrote:
>
> Stephen, can you let me know if this takes care of everything?

Sorry, sleep intervened :-)

That patch fixes the build of the .o (I didn't do the link).

Thanks.
-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: linux-next: sparc tree build failure

[Kristoffer Glembo]

I noticed just now that this patch broke the driver on LEON SPARC. It
does not find the root node by using of_find_node_by_name(NULL, "/").

I'm not sure whether this is because we build our OF tree wrong or if
it is an incorrect way of doing things. But if I instead change it to
of_find_node_by_path("/") it works fine.

Signed-off-by: Kristoffer Glembo <kristoffer@gaisler.com>

---
  drivers/serial/apbuart.c |    2 +-
  1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/serial/apbuart.c b/drivers/serial/apbuart.c
index a1e9503..fe91319 100644
--- a/drivers/serial/apbuart.c
+++ b/drivers/serial/apbuart.c
@@ -609,7 +609,7 @@ static void grlib_apbuart_configure(void)
                 return;

         /* Get bus frequency */
-       rp = of_find_node_by_name(NULL, "/");
+       rp = of_find_node_by_path("/");
         rp = of_get_next_child(rp, NULL);
         prop = of_get_property(rp, "clock-frequency", NULL);
         freq_khz = *prop;
--
1.5.2.2


David Miller wrote:
> 
> Hmm, it took a tiny bit more than that :-)  Here's what I commited
> to fix this, thanks.
> 
> apbuart: Kill dependency on deprecated Sparc-only PROM interfaces.
> 
> Use the proper modern OF ones instead.
> 
> Noticed by Stephen Rothwell.
> 
> Signed-off-by: David S. Miller <davem@davemloft.net>
> ---
>  drivers/serial/apbuart.c |   12 ++++++------
>  1 files changed, 6 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/serial/apbuart.c b/drivers/serial/apbuart.c
> index c7883a3..5f9dec3 100644
> --- a/drivers/serial/apbuart.c
> +++ b/drivers/serial/apbuart.c
> @@ -29,7 +29,6 @@
>  #include <linux/io.h>
>  #include <linux/serial_core.h>
>  #include <asm/irq.h>
> -#include <asm/oplib.h>
>  
>  #include "apbuart.h"
>  
> @@ -596,10 +595,9 @@ static struct of_platform_driver grlib_apbuart_of_driver = {
>  static void grlib_apbuart_configure(void)
>  {
>  	static int enum_done;
> -	struct device_node *np;
> +	struct device_node *np, *rp;
>  	struct uart_port *port = NULL;
> -
> -	int node;
> +	const u32 *prop;
>  	int freq_khz;
>  	int v = 0, d = 0;
>  	unsigned int addr;
> @@ -610,8 +608,10 @@ static void grlib_apbuart_configure(void)
>  		return;
>  
>  	/* Get bus frequency */
> -	node = prom_getchild(prom_root_node);
> -	freq_khz = prom_getint(node, "clock-frequency");
> +	rp = of_find_node_by_name(NULL, "/");
> +	rp = of_get_next_child(rp, NULL);
> +	prop = of_get_property(rp, "clock-frequency", NULL);
> +	freq_khz = *prop;
>  
>  	line = 0;
>  	for_each_matching_node(np, apbuart_match) {

Re: linux-next: sparc tree build failure

[David Miller]

Your patch is correct but is whitespace damaged by your email
client.

Please take extra care to avoid this, because I spent the time to
prepare your patch for application and then it failed, and therefore I
wasted my time on your patch.

linux-next: sparc tree build failure

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 577 bytes --]

Hi Dave,

Today's linux-next build (sparc32 defconfig) failed like this:

arch/sparc/lib/atomic32.c: In function '__atomic_add_return':
arch/sparc/lib/atomic32.c:34: error: implicit declaration of function '__raw_spin_lock_irqsave'
arch/sparc/lib/atomic32.c:38: error: implicit declaration of function '__raw_spin_unlock_irqrestore'

Caused by commit 4df286e52917c95c415400367cfd523dfbb0f93a ("sparc: Make
atomic locks raw") which I have reverted for today.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: linux-next: sparc tree build failure

[David Miller]

From: Stephen Rothwell <sfr@canb.auug.org.au>
Date: Thu, 26 Nov 2009 20:16:07 +1100

> Today's linux-next build (sparc32 defconfig) failed like this:
> 
> arch/sparc/lib/atomic32.c: In function '__atomic_add_return':
> arch/sparc/lib/atomic32.c:34: error: implicit declaration of function '__raw_spin_lock_irqsave'
> arch/sparc/lib/atomic32.c:38: error: implicit declaration of function '__raw_spin_unlock_irqrestore'
> 
> Caused by commit 4df286e52917c95c415400367cfd523dfbb0f93a ("sparc: Make
> atomic locks raw") which I have reverted for today.

Ho hum, that's the second iteration and it broke the build first time
too.

__raw_spin_lock_irqsave() definitions don't exist anywhere in the
tree, I wonder what this was even build tested against?

I'm reverting, and I'll be hard pressed to add new versions without
solid proof that it doesn't break the build a third time :-)

Re: linux-next: sparc tree build failure

[Thomas Gleixner]

On Thu, 26 Nov 2009, David Miller wrote:
> From: Stephen Rothwell <sfr@canb.auug.org.au>
> Date: Thu, 26 Nov 2009 20:16:07 +1100
> 
> > Today's linux-next build (sparc32 defconfig) failed like this:
> > 
> > arch/sparc/lib/atomic32.c: In function '__atomic_add_return':
> > arch/sparc/lib/atomic32.c:34: error: implicit declaration of function '__raw_spin_lock_irqsave'
> > arch/sparc/lib/atomic32.c:38: error: implicit declaration of function '__raw_spin_unlock_irqrestore'
> > 
> > Caused by commit 4df286e52917c95c415400367cfd523dfbb0f93a ("sparc: Make
> > atomic locks raw") which I have reverted for today.
> 
> Ho hum, that's the second iteration and it broke the build first time
> too.
> 
> __raw_spin_lock_irqsave() definitions don't exist anywhere in the
> tree, I wonder what this was even build tested against?
> 
> I'm reverting, and I'll be hard pressed to add new versions without
> solid proof that it doesn't break the build a third time :-)

/me blushes. Dunno what went wrong this time. Sorry.

    tglx