On Mon, Jan 2, 2012 at 1:25 AM, Grant Likely [off-list ref] wrote:

On Wed, Dec 28, 2011 at 09:32:14PM +1100, Benjamin Herrenschmidt wrote:

quoted

On Wed, 2011-12-28 at 19:49 +1100, Stephen Rothwell wrote:

quoted

Hi ,

After merging the final tree, today's linux-next build (powerpc
allyesconfig) failed like this:

kernel/built-in.o: In function `irq_dispose_mapping':
(.opd+0x159f0): multiple definition of `irq_dispose_mapping'
arch/powerpc/kernel/built-in.o:(.opd+0x960): first defined here
kernel/built-in.o: In function `irq_create_of_mapping':
(.opd+0x15a20): multiple definition of `irq_create_of_mapping'
arch/powerpc/kernel/built-in.o:(.opd+0x9a8): first defined here
kernel/built-in.o: In function `.irq_create_of_mapping':
(.text+0x147030): multiple definition of `.irq_create_of_mapping'
arch/powerpc/kernel/built-in.o:(.text+0x9de0): first defined here
kernel/built-in.o: In function `.irq_dispose_mapping':
(.text+0x146f4c): multiple definition of `.irq_dispose_mapping'
arch/powerpc/kernel/built-in.o:(.text+0x9684): first defined here

I am not sure what caused this. And have just left it broken.

Grant, is your irq remapper misbehaving ?

Hmmmm, that's odd.  I've not touched it.  I'll investigate.

It looks like CONFIG_IRQ_DOMAIN is getting selected by TWL4030_CORE.
Drivers must not select that config symbol.  It looks like commit
da28adbd (mfd: twl-core: Add initial DT support for twl4030/twl6030)
is the culprit.

The following patch should solve the problem:

diff --git a/drivers/mfd/Kconfig b/drivers/mfd/Kconfig
index 13c468e..e43a570 100644
--- a/drivers/mfd/Kconfig
+++ b/drivers/mfd/Kconfig

@@ -200,8 +200,7 @@ config MENELAUS

 config TWL4030_CORE
 	bool "Texas Instruments TWL4030/TWL5030/TWL6030/TPS659x0 Support"
-	depends on I2C=y && GENERIC_HARDIRQS
-	select IRQ_DOMAIN
+	depends on I2C=y && GENERIC_HARDIRQS && IRQ_DOMAIN
 	help
 	  Say yes here if you have TWL4030 / TWL6030 family chip on your board.
 	  This core driver provides register access and IRQ handling