[PATCH tip] Fix build failure x86_default_fixup_cpu_id() with !CONFIG_SMP

[PATCH tip] Fix build failure x86_default_fixup_cpu_id() with !CONFIG_SMP

[Kamalesh Babulal]

Hi,

tip (a4a7e7e29333) build fails, when compiled with !CONFIG_SMP. 
I have only build tested the patch.

x86: Fix x86_default_fixup_cpu_id() build failure with !CONFIG_SMP

arch/x86/kernel/cpu/common.c: In function 'x86_default_fixup_cpu_id':
arch/x86/kernel/cpu/common.c:1149: error: 'struct cpuinfo_x86' has no member named 'phys_proc_id'
make[3]: *** [arch/x86/kernel/cpu/common.o] Error 1

the code was introduced by commit 64be4c1c24. This patch
introduced #ifdef to guard !SMP case.

Signed-off-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
--
 arch/x86/kernel/cpu/common.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index ad4da45..1154528 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -1146,7 +1146,9 @@ static void dbg_restore_debug_regs(void)
  */
 void __cpuinit x86_default_fixup_cpu_id(struct cpuinfo_x86 *c, int node)
 {
+#ifdef CONFIG_SMP
 	pr_err("NUMA core number %d differs from configured core number %d\n", node, c->phys_proc_id);
+#endif
 }
 
 /*

Re: [PATCH tip] Fix build failure x86_default_fixup_cpu_id() with !CONFIG_SMP

[Srivatsa S. Bhat]

Hi Kamalesh,

On 12/06/2011 04:17 PM, Kamalesh Babulal wrote:

> Hi,
> 
> tip (a4a7e7e29333) build fails, when compiled with !CONFIG_SMP. 
> I have only build tested the patch.
> 
> x86: Fix x86_default_fixup_cpu_id() build failure with !CONFIG_SMP
> 
> arch/x86/kernel/cpu/common.c: In function 'x86_default_fixup_cpu_id':
> arch/x86/kernel/cpu/common.c:1149: error: 'struct cpuinfo_x86' has no member named 'phys_proc_id'
> make[3]: *** [arch/x86/kernel/cpu/common.o] Error 1
> 
> the code was introduced by commit 64be4c1c24. This patch
> introduced #ifdef to guard !SMP case.
> 
> Signed-off-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
> ----


There was a similar build-fix patch by Steffen Persvold at:
https://lkml.org/lkml/2011/12/6/28

That would solve your case too, right?

 

Regards,
Srivatsa S. Bhat

Re: [PATCH tip] Fix build failure x86_default_fixup_cpu_id() with !CONFIG_SMP

[Steffen Persvold]

On 12/6/2011 12:37, Srivatsa S. Bhat wrote:
> Hi Kamalesh,
>
> On 12/06/2011 04:17 PM, Kamalesh Babulal wrote:
>
>> Hi,
>>
>> tip (a4a7e7e29333) build fails, when compiled with !CONFIG_SMP.
>> I have only build tested the patch.
>>
>> x86: Fix x86_default_fixup_cpu_id() build failure with !CONFIG_SMP
>>
>> arch/x86/kernel/cpu/common.c: In function 'x86_default_fixup_cpu_id':
>> arch/x86/kernel/cpu/common.c:1149: error: 'struct cpuinfo_x86' has no member named 'phys_proc_id'
>> make[3]: *** [arch/x86/kernel/cpu/common.o] Error 1
>>
>> the code was introduced by commit 64be4c1c24. This patch
>> introduced #ifdef to guard !SMP case.
>>
>> Signed-off-by: Kamalesh Babulal<kamalesh@linux.vnet.ibm.com>
>> ----
>
>
> There was a similar build-fix patch by Steffen Persvold at:
> https://lkml.org/lkml/2011/12/6/28
>
> That would solve your case too, right?
>

I would hope so (I tested with !CONFIG_SMP also), albeit I used :

#ifdef CONFIG_NUMA

and not

#ifdef CONFIG_SMP


but I don't think you can compile with CONFIG_NUMA without having 
CONFIG_SMP  (if I read the Kconfig correctly..).

I used CONFIG_NUMA because the code-path doesn't really make sense on 
non-numa nodes.

Cheers,
-- 
Steffen Persvold, Chief Architect NumaChip
Numascale AS - www.numascale.com
Tel: +47 92 49 25 54 Skype: spersvold

Re: [PATCH tip] Fix build failure x86_default_fixup_cpu_id() with !CONFIG_SMP

[Kamalesh Babulal]

* Steffen Persvold <sp@numascale.com> [2011-12-06 12:44:46]:

> On 12/6/2011 12:37, Srivatsa S. Bhat wrote:
> >Hi Kamalesh,
> >
> >On 12/06/2011 04:17 PM, Kamalesh Babulal wrote:
> >
> >>Hi,
> >>
> >>tip (a4a7e7e29333) build fails, when compiled with !CONFIG_SMP.
> >>I have only build tested the patch.
> >>
> >>x86: Fix x86_default_fixup_cpu_id() build failure with !CONFIG_SMP
> >>
> >>arch/x86/kernel/cpu/common.c: In function 'x86_default_fixup_cpu_id':
> >>arch/x86/kernel/cpu/common.c:1149: error: 'struct cpuinfo_x86' has no member named 'phys_proc_id'
> >>make[3]: *** [arch/x86/kernel/cpu/common.o] Error 1
> >>
> >>the code was introduced by commit 64be4c1c24. This patch
> >>introduced #ifdef to guard !SMP case.
> >>
> >>Signed-off-by: Kamalesh Babulal<kamalesh@linux.vnet.ibm.com>
> >>----
> >
> >
> >There was a similar build-fix patch by Steffen Persvold at:
> >https://lkml.org/lkml/2011/12/6/28
> >
> >That would solve your case too, right?
> >

Thanks for pointing out the patch.

> 
> I would hope so (I tested with !CONFIG_SMP also), albeit I used :
> 
> #ifdef CONFIG_NUMA
> 
> and not
> 
> #ifdef CONFIG_SMP
> 
> 
> but I don't think you can compile with CONFIG_NUMA without having
> CONFIG_SMP  (if I read the Kconfig correctly..).
> 
> I used CONFIG_NUMA because the code-path doesn't really make sense
> on non-numa nodes.

agree, cpuinfo_x86->phys_proc_id is protected within CONFIG_SMP but the
right way is to fix using CONFIG_NUMA as explained by you in 
https://lkml.org/lkml/2011/12/6/28.

Regards,
Kamalesh.