Build failure: bisected: v3.1-rc1 with config ARCH_OMAP && !ARCH_OMAP4 fails with linker error

Commit Message

Paul Walmsley Aug. 10, 2011, 6:57 a.m. UTC

On Tue, 9 Aug 2011, Paul Walmsley wrote:

> On Tue, 9 Aug 2011, Tony Lindgren wrote:
> 
> > Hmm, there are also these when CONFIG_ARCH_OMAP4 is not selected:
> > 
> > arch/arm/mach-omap2/built-in.o: In function `_enable_module':
> > arch/arm/mach-omap2/omap_hwmod.c:701: undefined reference to `omap4_cminst_module_enable'
> > arch/arm/mach-omap2/built-in.o: In function `_disable_module':
> > arch/arm/mach-omap2/omap_hwmod.c:726: undefined reference to `omap4_cminst_module_disable'
> > arch/arm/mach-omap2/built-in.o: In function `_wait_target_disable':
> > arch/arm/mach-omap2/omap_hwmod.c:1179: undefined reference to `omap4_cminst_wait_module_idle'
> > distcc[27594] ERROR: compile (null) on localhost failed
> > make: *** [.tmp_vmlinux1] Error 1
> > 
> > Care to take a look?
> 
> Weird, those don't show up on my n800 config (below) with CodeSourcery 
> 2010.09-50.  Looks like something isn't removing the dead code.  will post 
> a patch for this.
> 
> If you post which compiler you're using, I'll add it to my build tests 
> too...

Hmmm, playing around with this further, this is probably not a
compiler problem.  Looks like what triggers this is the MULTI_OMAP2
preprocessor trickery in arch/arm/plat-omap/include/plat/cpu.h, which
would cause something like cpu_is_omap24xx() to be resolved at runtime
rather than by the preprocessor.

This should fix the immediate issue.  Will include in a 3.1-rc fixes branch.


- Paul


From: Paul Walmsley <[e-mail]>
Date: Tue, 9 Aug 2011 22:27:01 -0600
Subject: [PATCH] OMAP: hwmod: fix build break on non-OMAP4 multi-OMAP2 builds

Builds for multi-OMAP2 (e.g., OMAP2420 with OMAP2430) with
CONFIG_ARCH_OMAP4=n fail with the following errors:

arch/arm/mach-omap2/built-in.o: In function `_enable_module':
arch/arm/mach-omap2/omap_hwmod.c:701: undefined reference to `omap4_cminst_module_enable'
arch/arm/mach-omap2/built-in.o: In function `_disable_module':
arch/arm/mach-omap2/omap_hwmod.c:726: undefined reference to `omap4_cminst_module_disable'
arch/arm/mach-omap2/built-in.o: In function `_wait_target_disable':
arch/arm/mach-omap2/omap_hwmod.c:1179: undefined reference to `omap4_cminst_wait_module_idle'

This is probably due to the preprocessor directives in
arch/arm/plat-omap/include/plat/cpu.h that convert some cpu_is_omap*()
expressions from preprocessor directives into something that is only
resolvable during runtime, if multiple OMAP2 build targets are
selected.

Thanks to Tony Lindgren <[e-mail]> for reporting.

Signed-off-by: Paul Walmsley <[e-mail]>
Cc: Tony Lindgren <[e-mail]>
---
 arch/arm/mach-omap2/cminst44xx.h |   19 ++++++++++++++++++-
 1 files changed, 18 insertions(+), 1 deletions(-)

Patch

diff --git a/arch/arm/mach-omap2/cminst44xx.h b/arch/arm/mach-omap2/cminst44xx.h
index f2ea645..df9be67 100644
--- a/arch/arm/mach-omap2/cminst44xx.h
+++ b/arch/arm/mach-omap2/cminst44xx.h
@@ -18,13 +18,30 @@  extern void omap4_cminst_clkdm_force_sleep(u8 part, s16 inst, u16 cdoffs);
 extern void omap4_cminst_clkdm_force_wakeup(u8 part, s16 inst, u16 cdoffs);
 
 extern int omap4_cminst_wait_module_ready(u8 part, u16 inst, s16 cdoffs, u16 clkctrl_offs);
-extern int omap4_cminst_wait_module_idle(u8 part, u16 inst, s16 cdoffs, u16 clkctrl_offs);
+
+# ifdef CONFIG_ARCH_OMAP4
+extern int omap4_cminst_wait_module_idle(u8 part, u16 inst, s16 cdoffs,
+					 u16 clkctrl_offs);
 
 extern void omap4_cminst_module_enable(u8 mode, u8 part, u16 inst, s16 cdoffs,
 				       u16 clkctrl_offs);
 extern void omap4_cminst_module_disable(u8 part, u16 inst, s16 cdoffs,
 					u16 clkctrl_offs);
 
+# else
+
+static inline int omap4_cminst_wait_module_idle(u8 part, u16 inst, s16 cdoffs, u16 clkctrl_offs)
+{
+	return 0;
+}
+
+static inline void omap4_cminst_module_enable(u8 mode, u8 part, u16 inst, s16 cdoffs,
+				u16 clkctrl_offs) {}
+static inline void omap4_cminst_module_disable(u8 part, u16 inst, s16 cdoffs,
+				 u16 clkctrl_offs) {}
+
+# endif
+
 /*
  * In an ideal world, we would not export these low-level functions,
  * but this will probably take some time to fix properly