[PATCH] arm64/mm: Set only the PTE_DIRTY bit while preserving the HW dirty state

[PATCH] arm64/mm: Set only the PTE_DIRTY bit while preserving the HW dirty state

[Anshuman Khandual]

pte_mkdirty() creates dirty states both in SW and HW bits, which is really
not required, either in pte_wrprotect() or pte_modify() for preserving the
HW dirty state. Because pte_mkdirty() sets PTE_DIRTY and clears PTE_RDONLY
as pte_write() always evaluates to be true - otherwise pte_hw_dirty() will
not test out in the first place. Clearing PTE_RDONLY again is not required
here because the pte is already in pte_hw_dirty() but might soon loose its
dirty state thus requiring preservation in SW dirty bit i.e PTE_DIRTY.

Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Anshuman Khandual <anshuman.khandual@arm.com>
---
This applies on v6.5-rc1

 arch/arm64/include/asm/pgtable.h | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
index 0bd18de9fd97..171d6d7f8087 100644
--- a/arch/arm64/include/asm/pgtable.h
+++ b/arch/arm64/include/asm/pgtable.h
@@ -212,7 +212,7 @@ static inline pte_t pte_wrprotect(pte_t pte)
 	 * clear), set the PTE_DIRTY bit.
 	 */
 	if (pte_hw_dirty(pte))
-		pte = pte_mkdirty(pte);
+		pte = set_pte_bit(pte, __pgprot(PTE_DIRTY));
 
 	pte = clear_pte_bit(pte, __pgprot(PTE_WRITE));
 	pte = set_pte_bit(pte, __pgprot(PTE_RDONLY));
@@ -823,7 +823,8 @@ static inline pte_t pte_modify(pte_t pte, pgprot_t newprot)
 			      PTE_ATTRINDX_MASK;
 	/* preserve the hardware dirty information */
 	if (pte_hw_dirty(pte))
-		pte = pte_mkdirty(pte);
+		pte = set_pte_bit(pte, __pgprot(PTE_DIRTY));
+
 	pte_val(pte) = (pte_val(pte) & ~mask) | (pgprot_val(newprot) & mask);
 	return pte;
 }
-- 
2.30.2

Re: [PATCH] arm64/mm: Set only the PTE_DIRTY bit while preserving the HW dirty state

[David Hildenbrand]

On 13.07.23 09:15, Anshuman Khandual wrote:
> pte_mkdirty() creates dirty states both in SW and HW bits, which is really
> not required, either in pte_wrprotect() or pte_modify() for preserving the
> HW dirty state. Because pte_mkdirty() sets PTE_DIRTY and clears PTE_RDONLY
> as pte_write() always evaluates to be true - otherwise pte_hw_dirty() will
> not test out in the first place. Clearing PTE_RDONLY again is not required
> here because the pte is already in pte_hw_dirty() but might soon loose its
> dirty state thus requiring preservation in SW dirty bit i.e PTE_DIRTY.
> 
> Cc: Catalin Marinas <catalin.marinas@arm.com>
> Cc: Will Deacon <will@kernel.org>
> Cc: Mark Rutland <mark.rutland@arm.com>
> Cc: linux-arm-kernel@lists.infradead.org
> Cc: linux-kernel@vger.kernel.org
> Signed-off-by: Anshuman Khandual <anshuman.khandual@arm.com>
> ---
> This applies on v6.5-rc1
> 
>   arch/arm64/include/asm/pgtable.h | 5 +++--
>   1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
> index 0bd18de9fd97..171d6d7f8087 100644
> --- a/arch/arm64/include/asm/pgtable.h
> +++ b/arch/arm64/include/asm/pgtable.h
> @@ -212,7 +212,7 @@ static inline pte_t pte_wrprotect(pte_t pte)
>   	 * clear), set the PTE_DIRTY bit.
>   	 */
>   	if (pte_hw_dirty(pte))
> -		pte = pte_mkdirty(pte);
> +		pte = set_pte_bit(pte, __pgprot(PTE_DIRTY));
>   
>   	pte = clear_pte_bit(pte, __pgprot(PTE_WRITE));
>   	pte = set_pte_bit(pte, __pgprot(PTE_RDONLY));
> @@ -823,7 +823,8 @@ static inline pte_t pte_modify(pte_t pte, pgprot_t newprot)
>   			      PTE_ATTRINDX_MASK;
>   	/* preserve the hardware dirty information */
>   	if (pte_hw_dirty(pte))
> -		pte = pte_mkdirty(pte);
> +		pte = set_pte_bit(pte, __pgprot(PTE_DIRTY));
> +
>   	pte_val(pte) = (pte_val(pte) & ~mask) | (pgprot_val(newprot) & mask);
>   	return pte;
>   }

Reviewed-by: David Hildenbrand <david@redhat.com>

-- 
Cheers,

David / dhildenb

Re: [PATCH] arm64/mm: Set only the PTE_DIRTY bit while preserving the HW dirty state

[Catalin Marinas]

On Thu, Jul 13, 2023 at 12:45:18PM +0530, Anshuman Khandual wrote:
> pte_mkdirty() creates dirty states both in SW and HW bits, which is really
> not required, either in pte_wrprotect() or pte_modify() for preserving the
> HW dirty state. Because pte_mkdirty() sets PTE_DIRTY and clears PTE_RDONLY
> as pte_write() always evaluates to be true - otherwise pte_hw_dirty() will
> not test out in the first place. Clearing PTE_RDONLY again is not required
> here because the pte is already in pte_hw_dirty() but might soon loose its
> dirty state thus requiring preservation in SW dirty bit i.e PTE_DIRTY.
> 
> Cc: Catalin Marinas <catalin.marinas@arm.com>
> Cc: Will Deacon <will@kernel.org>
> Cc: Mark Rutland <mark.rutland@arm.com>
> Cc: linux-arm-kernel@lists.infradead.org
> Cc: linux-kernel@vger.kernel.org
> Signed-off-by: Anshuman Khandual <anshuman.khandual@arm.com>

Acked-by: Catalin Marinas <catalin.marinas@arm.com>

Re: [PATCH] arm64/mm: Set only the PTE_DIRTY bit while preserving the HW dirty state

[Will Deacon]

On Thu, 13 Jul 2023 12:45:18 +0530, Anshuman Khandual wrote:
> pte_mkdirty() creates dirty states both in SW and HW bits, which is really
> not required, either in pte_wrprotect() or pte_modify() for preserving the
> HW dirty state. Because pte_mkdirty() sets PTE_DIRTY and clears PTE_RDONLY
> as pte_write() always evaluates to be true - otherwise pte_hw_dirty() will
> not test out in the first place. Clearing PTE_RDONLY again is not required
> here because the pte is already in pte_hw_dirty() but might soon loose its
> dirty state thus requiring preservation in SW dirty bit i.e PTE_DIRTY.
> 
> [...]

Applied to arm64 (for-next/mm), thanks!

[1/1] arm64/mm: Set only the PTE_DIRTY bit while preserving the HW dirty state
      https://git.kernel.org/arm64/c/6477c3886ae1

Cheers,
-- 
Will

https://fixes.arm64.dev
https://next.arm64.dev
https://will.arm64.dev