[PATCH] x86_64: mm fix x86 cpuid keys used in alternative_smp fix

[PATCH] x86_64: mm fix x86 cpuid keys used in alternative_smp fix

[Paul Jackson]

From: Paul Jackson <pj@sgi.com>

The x86_64 crosstool build was failing due to:

  kernel/built-in.o(.smp_altinstructions+0x58): \
    include/asm/spinlock.h:54: undefined reference to `X86_FEATURE_UP'

This reference to X86_FEATURE_UP was added to the definition of
the alternative_smp() macro in alternative.h by the patch:
  x86_64-mm-fix-x86-cpuid-keys-used-in-alternative_smp

The alternative_smp() macro is used in spinlock.h.

The definition of X86_FEATURE_UP is in asm-x86_64/cpufeature.h.
If this is included in alternative.h, then the build succeeds.

Signed-off-by: Paul Jackson <pj@sgi.com>

---

 include/asm-x86_64/alternative.h |    1 +
 1 files changed, 1 insertion(+)

--- 2.6.18-rc4-mm2.orig/include/asm-x86_64/alternative.h	2006-08-24 03:37:22.000000000 -0700
+++ 2.6.18-rc4-mm2/include/asm-x86_64/alternative.h	2006-08-24 03:38:09.000000000 -0700
@@ -4,6 +4,7 @@
 #ifdef __KERNEL__
 
 #include <linux/types.h>
+#include <asm/cpufeature.h>
 
 struct alt_instr {
 	u8 *instr; 		/* original instruction */

-- 
                  I won't rest till it's the best ...
                  Programmer, Linux Scalability
                  Paul Jackson <pj@sgi.com> 1.925.600.0401

Re: [PATCH] x86_64: mm fix x86 cpuid keys used in alternative_smp fix

[Andi Kleen]

On Thursday 24 August 2006 12:58, Paul Jackson wrote:
> From: Paul Jackson <pj@sgi.com>
> 
> The x86_64 crosstool build was failing due to:
> 
>   kernel/built-in.o(.smp_altinstructions+0x58): \
>     include/asm/spinlock.h:54: undefined reference to `X86_FEATURE_UP'
> 
> This reference to X86_FEATURE_UP was added to the definition of
> the alternative_smp() macro in alternative.h by the patch:
>   x86_64-mm-fix-x86-cpuid-keys-used-in-alternative_smp
> 
> The alternative_smp() macro is used in spinlock.h.
> 
> The definition of X86_FEATURE_UP is in asm-x86_64/cpufeature.h.
> If this is included in alternative.h, then the build succeeds.

Ok it seems to depend on the .config, but nobody before submitted 
the buggy .config

I folded the patch in

-Andi