Re: [PATCH] mm: Move set_pxd_safe() helpers from generic to platform

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

On 9/20/24 12:09, David Hildenbrand wrote:
> On 20.09.24 07:30, Anshuman Khandual wrote:
>> set_pxd_safe() helpers that serve a specific purpose for both x86 and riscv
>> platforms, do not need to be in the common memory code. Otherwise they just
>> unnecessarily make the common API more complicated. This moves the helpers
>> from common code to platform instead.
>>
>> Cc: Paul Walmsley <paul.walmsley@sifive.com>
>> Cc: Palmer Dabbelt <palmer@dabbelt.com>
>> Cc: Thomas Gleixner <tglx@linutronix.de>
>> Cc: Dave Hansen <dave.hansen@linux.intel.com>
>> Cc: David Hildenbrand <david@redhat.com>
>> Cc: Ryan Roberts <ryan.roberts@arm.com>
>> Cc: Andrew Morton <akpm@linux-foundation.org>
>> Cc: x86@kernel.org
>> Cc: linux-mm@kvack.org
>> Cc: linux-riscv@lists.infradead.org
>> Cc: linux-kernel@vger.kernel.org
>> Suggested-by: David Hildenbrand <david@redhat.com>
>> Signed-off-by: Anshuman Khandual <anshuman.khandual@arm.com>
>> ---
>>   arch/riscv/include/asm/pgtable.h | 19 ++++++++++++++++
>>   arch/x86/include/asm/pgtable.h   | 37 +++++++++++++++++++++++++++++++
>>   include/linux/pgtable.h          | 38 --------------------------------
>>   3 files changed, 56 insertions(+), 38 deletions(-)
>>
>> diff --git a/arch/riscv/include/asm/pgtable.h b/arch/riscv/include/asm/pgtable.h
>> index 089f3c9f56a3..39ca652c5ebe 100644
>> --- a/arch/riscv/include/asm/pgtable.h
>> +++ b/arch/riscv/include/asm/pgtable.h
>> @@ -957,6 +957,25 @@ void misc_mem_init(void);
>>   extern unsigned long empty_zero_page[PAGE_SIZE / sizeof(unsigned long)];
>>   #define ZERO_PAGE(vaddr) (virt_to_page(empty_zero_page))
>>   +/*
>> + * Use set_p*_safe(), and elide TLB flushing, when confident that *no*
>> + * TLB flush will be required as a result of the "set". For example, use
>> + * in scenarios where it is known ahead of time that the routine is
>> + * setting non-present entries, or re-setting an existing entry to the
>> + * same value. Otherwise, use the typical "set" helpers and flush the
>> + * TLB.
>> + */
>> +#define set_p4d_safe(p4dp, p4d) \
>> +({ \
>> +    WARN_ON_ONCE(p4d_present(*p4dp) && !p4d_same(*p4dp, p4d)); \
>> +    set_p4d(p4dp, p4d); \
>> +})
>> +
>> +#define set_pgd_safe(pgdp, pgd) \
>> +({ \
>> +    WARN_ON_ONCE(pgd_present(*pgdp) && !pgd_same(*pgdp, pgd)); \
>> +    set_pgd(pgdp, pgd); \
>> +})
>>   #endif /* !__ASSEMBLY__ */
>>     #endif /* _ASM_RISCV_PGTABLE_H */
>> diff --git a/arch/x86/include/asm/pgtable.h b/arch/x86/include/asm/pgtable.h
>> index e39311a89bf4..fefb52bb6b4d 100644
>> --- a/arch/x86/include/asm/pgtable.h
>> +++ b/arch/x86/include/asm/pgtable.h
>> @@ -1701,6 +1701,43 @@ bool arch_is_platform_page(u64 paddr);
>>   #define arch_is_platform_page arch_is_platform_page
>>   #endif
>>   +/*
>> + * Use set_p*_safe(), and elide TLB flushing, when confident that *no*
>> + * TLB flush will be required as a result of the "set". For example, use
>> + * in scenarios where it is known ahead of time that the routine is
>> + * setting non-present entries, or re-setting an existing entry to the
>> + * same value. Otherwise, use the typical "set" helpers and flush the
>> + * TLB.
>> + */
>> +#define set_pte_safe(ptep, pte) \
>> +({ \
>> +    WARN_ON_ONCE(pte_present(*ptep) && !pte_same(*ptep, pte)); \
>> +    set_pte(ptep, pte); \
>> +})
>> +
>> +#define set_pmd_safe(pmdp, pmd) \
>> +({ \
>> +    WARN_ON_ONCE(pmd_present(*pmdp) && !pmd_same(*pmdp, pmd)); \
>> +    set_pmd(pmdp, pmd); \
>> +})
>> +
>> +#define set_pud_safe(pudp, pud) \
>> +({ \
>> +    WARN_ON_ONCE(pud_present(*pudp) && !pud_same(*pudp, pud)); \
>> +    set_pud(pudp, pud); \
>> +})
>> +
>> +#define set_p4d_safe(p4dp, p4d) \
>> +({ \
>> +    WARN_ON_ONCE(p4d_present(*p4dp) && !p4d_same(*p4dp, p4d)); \
>> +    set_p4d(p4dp, p4d); \
>> +})
>> +
>> +#define set_pgd_safe(pgdp, pgd) \
>> +({ \
>> +    WARN_ON_ONCE(pgd_present(*pgdp) && !pgd_same(*pgdp, pgd)); \
>> +    set_pgd(pgdp, pgd); \
>> +})
>>   #endif    /* __ASSEMBLY__ */
> 
> I'm wondering if we can completely get rid of these, for example via:
> 
> diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
> index d8dbeac8b206..bc71c25930bb 100644
> --- a/arch/x86/mm/init_64.c
> +++ b/arch/x86/mm/init_64.c
> @@ -79,10 +79,8 @@ DEFINE_POPULATE(pmd_populate_kernel, pmd, pte, init)
>  static inline void set_##type1##_init(type1##_t *arg1,         \
>                         type2##_t arg2, bool init)              \
>  {                                                              \
> -       if (init)                                               \
> -               set_##type1##_safe(arg1, arg2);                 \
> -       else                                                    \
> -               set_##type1(arg1, arg2);                        \
> +       WARN_ON_ONCE(init && ##type1##_present(*arg1) && !##type1##_same(*arg1, arg2)); \
> +       set_##type1(arg1, arg2);                                \
>  }
>  
> 
> We might be able to handle the pgd_populate etc part similarly, possibly getting
> rid of the pgd_populate_safe etc as well.
> 
> Assuming I don't miss anything important :)

Sounds feasible but will just leave that upto the x86 platform folks to
change later on, after this patch which just moves these helpers inside
the platform code.

> 
> Ideally, we get rid of the macros here and just use inline functions ...
> 

Sure, makes sense. Will change these as inline functions.

-#define set_pte_safe(ptep, pte) \
-({ \
-       WARN_ON_ONCE(pte_present(*ptep) && !pte_same(*ptep, pte)); \
-       set_pte(ptep, pte); \
-})
+static inline void set_pte_safe(pte_t *ptep, pte_t pte)
+{
+       WARN_ON_ONCE(pte_present(*ptep) && !pte_same(*ptep, pte));
+       set_pte(ptep, pte);
+}