On Sat, Nov 02, 2013 at 03:30:25AM +0300, Sergei Shtylyov wrote:

On 11/01/2013 03:43 AM, Simon Horman wrote:

quoted

quoted

quoted

Do not build the phy fixup unless CONFIG_PHYLIB is enabled.
Other than not being useful it is also not possible to compile

quoted

quoted

   s/compile/link/.

quoted

quoted

quoted

the code under this condition as phy_register_fixup_for_id()
is not defined.

quoted

quoted

   Not only this function is absent...

quoted

quoted

quoted

This problem was introduced by 48c8b96f21817aad
("ARM: shmobile: Lager: add Micrel KSZ8041 PHY fixup")

quoted

quoted

quoted

Cc: Sergei Shtylyov <redacted>
Signed-off-by: Simon Horman <redacted>
---
 arch/arm/mach-shmobile/board-lager.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

quoted

quoted

quoted

diff --git a/arch/arm/mach-shmobile/board-lager.c b/arch/arm/mach-shmobile/board-lager.c
index 78a31b6..d1a8ddd 100644
--- a/arch/arm/mach-shmobile/board-lager.c
+++ b/arch/arm/mach-shmobile/board-lager.c

@@ -245,7 +245,9 @@ static void __init lager_init(void)
 {
 	lager_add_standard_devices();

-	phy_register_fixup_for_id("r8a7790-ether-ff:01", lager_ksz8041_fixup);
+	if (IS_ENABLED(CONFIG_PHYLIB))
+		phy_register_fixup_for_id("r8a7790-ether-ff:01",
+					  lager_ksz8041_fixup);

quoted

quoted

   Perhaps you would consider enclosing the fixup function itself
into #ifdef as it causes link errors as well?  Doesn't seem
necessary as gcc probably drops it anyway but for completeness'
sake...

quoted

My understanding of the motivation for IS_ENABLED() is to
take advantage of gcc optimising away unused code and making such
problems disappear.

quoted

With this patch in place do you still see any warnings?

   No (but those were errors :-).
   I'd like the changelog to be adjusted at least...

Sure. Would you like me to list in the changelog the compile problems that
gcc reports without this patch? Or do you have something else in mind?