Re: [linux-next:master] BUILD REGRESSION 0a6f624a86e766a27d23cbb73c23be62231d10ff

[Anshuman Khandual <anshuman.khandual@arm.com>]

On 11/9/20 8:25 AM, Andrew Morton wrote:
> On Mon, 9 Nov 2020 08:20:05 +0530 Anshuman Khandual <anshuman.khandual@arm.com> wrote:
> 
>>
>>
>> On 11/7/20 10:37 AM, Andrew Morton wrote:
>>> On Fri, 06 Nov 2020 08:19:36 +0800 kernel test robot <lkp@intel.com> wrote:
>>>
>>>> tree/branch: https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git  master
>>>> branch HEAD: 0a6f624a86e766a27d23cbb73c23be62231d10ff  Add linux-next specific files for 20201105
>>>>
>>>> Error/Warning reports:
>>>>
>>>> https://lore.kernel.org/linux-mm/202010281624.9m2gZw45-lkp@intel.com
>>>> https://lore.kernel.org/linux-mm/202010290238.M1tDrV8p-lkp@intel.com
>>>> https://lore.kernel.org/linux-mm/202010291054.WEZO3olr-lkp@intel.com
>>>> https://lore.kernel.org/linux-mm/202011020749.5XQ3Hfzc-lkp@intel.com
>>>>
>>>> ...
>>>>
>>>> mm/kasan/init.c:318:9: warning: variable 'pud' set but not used [-Wunused-but-set-variable]
>>>>
>>>
>>> This is because mm/kasan/init.c does
>>>
>>> static void kasan_free_pud(pud_t *pud_start, p4d_t *p4d)
>>> {
>>> 	pud_t *pud;
>>> 	int i;
>>>
>>> 	for (i = 0; i < PTRS_PER_PUD; i++) {
>>> 		pud = pud_start + i;
>>> 		if (!pud_none(*pud))
>>> 			return;
>>> 	}
>>>
>>> 	pud_free(&init_mm, (pud_t *)page_to_virt(p4d_page(*p4d)));
>>> 	p4d_clear(p4d);
>>> }
>>>
>>> but arch/arm/include/asm/pgtable-2level.h does
>>>
>>> #define pud_none(pud)           (0)
>>>
>>> The solution here is for the arm implementation to reference `pud'. 
>>> Typically this is done via the use of an empty static inline C function
>>> rather than a macro.  But really all of these
>>>
>>> #define pud_none(pud)           (0)
>>> #define pud_bad(pud)            (0)
>>> #define pud_present(pud)        (1)
>>> #define pud_clear(pudp)         do { } while (0)
>>> #define set_pud(pud,pudp)       do { } while (0)
>>>
>>> should be thus converted.
>>>
>>> Could someone in arm world please attend to this?
>>
>> + Ard Biesheuvel <ardb@kernel.org>
>> + Linus Walleij <linus.walleij@linaro.org>
>>
>> There were some earlier discussions to solve this in a different way.
>>
>> https://lore.kernel.org/linux-mm/CACRpkdbAXCMTW--BmVs8SQ_u5baaeUob+U57E=4=CrMxWtMO2g@mail.gmail.com/
>>
>> with a subsequent follow up patch.
>>
>> https://lore.kernel.org/linux-arm-kernel/20201106085157.11211-1-linus.walleij@linaro.org/
> 
> Yes, but why?  macros-pretending-to-be-functions just keep on causing
> problems and they're so unnecessary.  Why not just write functions in
> the first place??   Did anyone try implementing that?

Just build tested this. Observed that set_pud()'s arguments here seems
to have been reversed. If this looks okay, will send out.

diff --git a/arch/arm/include/asm/pgtable-2level.h b/arch/arm/include/asm/pgtable-2level.h
index 3502c2f..9d4f5ee 100644
--- a/arch/arm/include/asm/pgtable-2level.h
+++ b/arch/arm/include/asm/pgtable-2level.h
@@ -177,11 +177,28 @@
  * the pud: the pud entry is never bad, always exists, and can't be set or
  * cleared.
  */
-#define pud_none(pud)          (0)
-#define pud_bad(pud)           (0)
-#define pud_present(pud)       (1)
-#define pud_clear(pudp)                do { } while (0)
-#define set_pud(pud,pudp)      do { } while (0)
+static inline int pud_none(pud_t pud)
+{
+       return 0;
+}
+
+static inline int pud_bad(pud_t pud)
+{
+       return 0;
+}
+
+static inline int pud_present(pud_t pud)
+{
+       return 1;
+}
+
+static inline void pud_clear(pud_t *pudp)
+{
+}
+
+static inline void set_pud(pud_t *pudp, pud_t pud)
+{
+}