Re: [PATCH] x86/irqflags: mark native_restore_fl extern inline

[Juergen Gross]

On 28/08/18 07:13, Greg Kroah-Hartman wrote:
> On Mon, Aug 27, 2018 at 02:40:09PM -0700, Nick Desaulniers wrote:
>> Fixes commit 208cbb325589 ("x86/irqflags: Provide a declaration for
>> native_save_fl")
>>
>> This should have been marked extern inline in order to pick up the out
>> of line definition in arch/x86/kernel/irqflags.S.
>>
>> Cc: stable@xxxxxxxxxxxxxxx # 4.18, 4.14, 4.9, 4.4
>> Reported-by: Ben Hutchings <ben.hutchings@xxxxxxxxxxxxxxx>
>> Signed-off-by: Nick Desaulniers <ndesaulniers@xxxxxxxxxx>
>> ---
>>  arch/x86/include/asm/irqflags.h | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/arch/x86/include/asm/irqflags.h b/arch/x86/include/asm/irqflags.h
>> index c14f2a74b2be..15450a675031 100644
>> --- a/arch/x86/include/asm/irqflags.h
>> +++ b/arch/x86/include/asm/irqflags.h
>> @@ -33,7 +33,8 @@ extern inline unsigned long native_save_fl(void)
>>  	return flags;
>>  }
>>  
>> -static inline void native_restore_fl(unsigned long flags)
>> +extern inline void native_restore_fl(unsigned long flags);
>> +extern inline void native_restore_fl(unsigned long flags)
> 
> This looks odd to me, but my coffee hasn't kicked in yet this morning.
> Why do you need both lines here?  Shouldn't the actual function be
> sufficient?  If not, a comment explaining this would be nice.

Look just some lines further up:

/* Declaration required for gcc < 4.9 to prevent
-Werror=missing-prototypes */
extern inline unsigned long native_save_fl(void);
extern inline unsigned long native_save_fl(void)


Juergen