Re: [PATCH] mm,x86: fix SMP x86 32bit build for native_pud_clear()

[Dave Jiang]

On 02/27/2017 12:29 PM, Boris Ostrovsky wrote:
>> On 02/15/2017 12:31 PM, Dave Jiang wrote:
>>> The fix introduced by e4decc90 to fix the UP case for 32bit x86, however
>>> that broke the SMP case that was working previously. Add ifdef so the dummy
>>> function only show up for 32bit UP case only.
>>>
>>> Fix: e4decc90 mm,x86: native_pud_clear missing on i386 build
>>>
>>> Reported-by: Alexander Kapshuk <alexander.kapshuk@xxxxxxxxx>
>>> Signed-off-by: Dave Jiang <dave.jiang@xxxxxxxxx>
>>> ---
>>>  arch/x86/include/asm/pgtable-3level.h |    2 ++
>>>  1 file changed, 2 insertions(+)
>>>
>>> diff --git a/arch/x86/include/asm/pgtable-3level.h b/arch/x86/include/asm/pgtable-3level.h
>>> index 50d35e3..8f50fb3 100644
>>> --- a/arch/x86/include/asm/pgtable-3level.h
>>> +++ b/arch/x86/include/asm/pgtable-3level.h
>>> @@ -121,9 +121,11 @@ static inline void native_pmd_clear(pmd_t *pmd)
>>>  	*(tmp + 1) = 0;
>>>  }
>>>  
>>> +#ifndef CONFIG_SMP
>>>  static inline void native_pud_clear(pud_t *pudp)
>>>  {
>>>  }
>>> +#endif
>>>  
>>>  static inline void pud_clear(pud_t *pudp)
>>>  {
>>>
> 
>>
>> This breaks one of the Fedora configurations as of 
>> e5d56efc97f8240d0b5d66c03949382b6d7e5570
>>
>> In file included from ./include/linux/mm.h:68:0,
>>                  from ./include/linux/highmem.h:7,
>>                  from ./include/linux/bio.h:21,
>>                  from ./include/linux/writeback.h:205,
>>                  from ./include/linux/memcontrol.h:30,
>>                  from ./include/linux/swap.h:8,
>>                  from ./include/linux/suspend.h:4,
>>                  from arch/x86/kernel/asm-offsets.c:12:
>> ./arch/x86/include/asm/pgtable.h: In function 'native_local_pudp_get_and_clear':
>> ./arch/x86/include/asm/pgtable.h:888:2: error: implicit declaration of function 'native_pud_clear';did you mean 'native_pmd_clear'? [-Werror=implicit-function-declaration]
>>   native_pud_clear(pudp);
>>   ^~~~~~~~~~~~~~~~
>>
>> Kernel configuration attached. I'm probably just going to revert
>> this part unless someone sends me a better fix.
> 
> 
> This breakage happens when CONFIG_HIGHMEM64G (i.e. CONFIG_X86_PAE) and CONFIG_PARAVIRT.

Thanks for the additional breakage configs report. I think I need to
extend that ifdef to additional configs. I'll send out a fix as soon as
I test the various different combination of configs.

> 
> -boris
>