From: Jinjiang Tu <tujinjiang@huawei.com>
To: "David Hildenbrand (Arm)" <david@kernel.org>,
<akpm@linux-foundation.org>, <ljs@kernel.org>,
<catalin.marinas@arm.com>, <will@kernel.org>,
<mark.rutland@arm.com>, <ryan.roberts@arm.com>,
<peterx@redhat.com>, <dave.jiang@intel.com>,
<linux-arm-kernel@lists.infradead.org>, <linux-mm@kvack.org>
Cc: <wangkefeng.wang@huawei.com>, <sunnanyong@huawei.com>
Subject: Re: [PATCH 1/3] mm: define generic pudp_invalidate() implementation
Date: Fri, 11 Sep 2026 10:28:49 +0800 [thread overview]
Message-ID: <5cb48654-8d77-4525-9854-5823934ae0e8@huawei.com> (raw)
In-Reply-To: <6fb6b2b8-daae-4707-aa98-c2c447782a44@kernel.org>
在 2026/9/9 18:24, David Hildenbrand (Arm) 写道:
> On 9/9/26 11:00, Jinjiang Tu wrote:
>> x86 and riscv have identical pudp_invalidate() implementations. Move it to
>> mm/pgtable-generic.c with the __HAVE_ARCH_PUDP_INVALIDATE override guard.
>> PowerPC keeps its own implementation. In the later patch, arm64 will use
>> the generic implementation too.
>>
>> No functional change intended.
>>
>> Signed-off-by: Jinjiang Tu <tujinjiang@huawei.com>
>> ---
>> arch/powerpc/include/asm/book3s/64/pgtable.h | 1 +
>> arch/riscv/include/asm/pgtable.h | 3 ---
>> arch/riscv/mm/pgtable.c | 9 ---------
>> arch/x86/include/asm/pgtable.h | 3 ---
>> arch/x86/mm/pgtable.c | 12 ------------
>> include/linux/pgtable.h | 7 +++++++
>> mm/pgtable-generic.c | 14 ++++++++++++++
>> 7 files changed, 22 insertions(+), 27 deletions(-)
>>
>> diff --git a/arch/powerpc/include/asm/book3s/64/pgtable.h b/arch/powerpc/include/asm/book3s/64/pgtable.h
>> index f4db7d7fbd5c..55d33c5537f1 100644
>> --- a/arch/powerpc/include/asm/book3s/64/pgtable.h
>> +++ b/arch/powerpc/include/asm/book3s/64/pgtable.h
>> @@ -1419,6 +1419,7 @@ static inline pgtable_t pgtable_trans_huge_withdraw(struct mm_struct *mm,
>> #define __HAVE_ARCH_PMDP_INVALIDATE
>> extern pmd_t pmdp_invalidate(struct vm_area_struct *vma, unsigned long address,
>> pmd_t *pmdp);
>> +#define __HAVE_ARCH_PUDP_INVALIDATE
>> extern pud_t pudp_invalidate(struct vm_area_struct *vma, unsigned long address,
>> pud_t *pudp);
>>
>> diff --git a/arch/riscv/include/asm/pgtable.h b/arch/riscv/include/asm/pgtable.h
>> index 40b1ed4f3ea8..327fa378032b 100644
>> --- a/arch/riscv/include/asm/pgtable.h
>> +++ b/arch/riscv/include/asm/pgtable.h
>> @@ -1159,9 +1159,6 @@ static inline pud_t pud_mkinvalid(pud_t pud)
>> return __pud(pud_val(pud) & ~(_PAGE_PRESENT | _PAGE_PROT_NONE));
>> }
>>
>> -extern pud_t pudp_invalidate(struct vm_area_struct *vma, unsigned long address,
>> - pud_t *pudp);
>> -
>> static inline pud_t pud_modify(pud_t pud, pgprot_t newprot)
>> {
>> return pte_pud(pte_modify(pud_pte(pud), newprot));
>> diff --git a/arch/riscv/mm/pgtable.c b/arch/riscv/mm/pgtable.c
>> index 9c4427d0b187..156e385e7db4 100644
>> --- a/arch/riscv/mm/pgtable.c
>> +++ b/arch/riscv/mm/pgtable.c
>> @@ -154,15 +154,6 @@ pmd_t pmdp_collapse_flush(struct vm_area_struct *vma,
>> return pmd;
>> }
>>
>> -pud_t pudp_invalidate(struct vm_area_struct *vma, unsigned long address,
>> - pud_t *pudp)
>> -{
>> - VM_WARN_ON_ONCE(!pud_present(*pudp));
>> - pud_t old = pudp_establish(vma, address, pudp, pud_mkinvalid(*pudp));
>> -
>> - flush_pud_tlb_range(vma, address, address + HPAGE_PUD_SIZE);
>> - return old;
>> -}
>> #endif /* CONFIG_TRANSPARENT_HUGEPAGE */
>>
>> pte_t pte_mkwrite(pte_t pte, struct vm_area_struct *vma)
>> diff --git a/arch/x86/include/asm/pgtable.h b/arch/x86/include/asm/pgtable.h
>> index d5f4917c1edc..97051c02385e 100644
>> --- a/arch/x86/include/asm/pgtable.h
>> +++ b/arch/x86/include/asm/pgtable.h
>> @@ -1387,9 +1387,6 @@ static inline pud_t pudp_establish(struct vm_area_struct *vma,
>> extern pmd_t pmdp_invalidate_ad(struct vm_area_struct *vma,
>> unsigned long address, pmd_t *pmdp);
>>
>> -pud_t pudp_invalidate(struct vm_area_struct *vma, unsigned long address,
>> - pud_t *pudp);
>> -
>> /*
>> * Page table pages are page-aligned. The lower half of the top
>> * level is used for userspace and the top half for the kernel.
>> diff --git a/arch/x86/mm/pgtable.c b/arch/x86/mm/pgtable.c
>> index cb03f5a2b243..0945908631dc 100644
>> --- a/arch/x86/mm/pgtable.c
>> +++ b/arch/x86/mm/pgtable.c
>> @@ -519,18 +519,6 @@ pmd_t pmdp_invalidate_ad(struct vm_area_struct *vma, unsigned long address,
>> }
>> #endif
>>
>> -#if defined(CONFIG_TRANSPARENT_HUGEPAGE) && \
>> - defined(CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD)
>> -pud_t pudp_invalidate(struct vm_area_struct *vma, unsigned long address,
>> - pud_t *pudp)
>> -{
>> - VM_WARN_ON_ONCE(!pud_present(*pudp));
>> - pud_t old = pudp_establish(vma, address, pudp, pud_mkinvalid(*pudp));
>> - flush_pud_tlb_range(vma, address, address + HPAGE_PUD_SIZE);
>> - return old;
>> -}
>> -#endif
>> -
>> /**
>> * reserve_top_address - Reserve a hole in the top of the kernel address space
>> * @reserve: Size of hole to reserve
>> diff --git a/include/linux/pgtable.h b/include/linux/pgtable.h
>> index 8c093c119e5a..32173005b087 100644
>> --- a/include/linux/pgtable.h
>> +++ b/include/linux/pgtable.h
>> @@ -1318,6 +1318,13 @@ extern pmd_t pmdp_invalidate_ad(struct vm_area_struct *vma,
>> unsigned long address, pmd_t *pmdp);
>> #endif
>>
>> +#ifndef __HAVE_ARCH_PUDP_INVALIDATE
>> +#ifdef CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
>> +extern pud_t pudp_invalidate(struct vm_area_struct *vma, unsigned long address,
>> + pud_t *pudp);
> No "extern" please.
Will update it.
>
> Staring at this file, I do wonder why we are not smarter about combining blocks. E.g., here simply:'
Do you mean merging all #ifdef CONFIG_TRANSPARENT_HUGEPAGE blocks? There are many of them scattered
across include/linux/pgtable.h and arch/arm64/include/asm/pgtable.h. They are intentionally kept
separate to group the PTE, PMD, and PUD level variants of the same operation together. For example,
ptep_set_access_flags, pmdp_set_access_flags, and pudp_set_access_flags are placed adjacently.
>
> diff --git a/include/linux/pgtable.h b/include/linux/pgtable.h
> index e3c8ab96941c5..8c9fe3bf580d4 100644
> --- a/include/linux/pgtable.h
> +++ b/include/linux/pgtable.h
> @@ -1229,9 +1229,9 @@ static inline void pmdp_set_wrprotect(struct mm_struct *mm,
> }
> #endif /* CONFIG_TRANSPARENT_HUGEPAGE */
> #endif
> -#ifndef __HAVE_ARCH_PUDP_SET_WRPROTECT
> -#ifdef CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
> #ifdef CONFIG_TRANSPARENT_HUGEPAGE
> +#ifdef CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
> +#ifndef __HAVE_ARCH_PUDP_SET_WRPROTECT
> static inline void pudp_set_wrprotect(struct mm_struct *mm,
> unsigned long address, pud_t *pudp)
> {
> @@ -1245,9 +1245,13 @@ static inline void pudp_set_wrprotect(struct mm_struct *mm,
> {
> BUILD_BUG();
> }
> -#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
> -#endif /* CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD */
> #endif
> +#ifndef __HAVE_ARCH_PUDP_INVALIDATE
> +pud_t pudp_invalidate(struct vm_area_struct *vma, unsigned long address,
> + pud_t *pudp);
> +#endif
> +#endif /* CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD */
> +#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
>
>> +#endif
>> +#endif
>> +
>> #ifndef __HAVE_ARCH_PTE_SAME
>> static inline int pte_same(pte_t pte_a, pte_t pte_b)
>> {
>> diff --git a/mm/pgtable-generic.c b/mm/pgtable-generic.c
>> index b91b1a98029c..ac1175d9a59a 100644
>> --- a/mm/pgtable-generic.c
>> +++ b/mm/pgtable-generic.c
>> @@ -217,6 +217,20 @@ pmd_t pmdp_invalidate_ad(struct vm_area_struct *vma, unsigned long address,
>> }
>> #endif
>>
>> +#ifndef __HAVE_ARCH_PUDP_INVALIDATE
>> +#ifdef CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
> I'd revert these and similarly see if they can be combined somehow (likely combining with
> pudp_huge_clear_flush is harder as we are quite inconsistent with it, gluing it to
> __HAVE_ARCH_PMDP_HUGE_CLEAR_FLUSH).
>
> Which raises the question whether we could also piggy-back on __HAVE_ARCH_PMDP_SET_WRPROTECT
> (I suspect no without a closer look.)
>
next prev parent reply other threads:[~2026-09-11 2:29 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-09-09 9:00 [PATCH 0/3] arm64: Add support for PUD THP Jinjiang Tu
2026-09-09 9:00 ` [PATCH 1/3] mm: define generic pudp_invalidate() implementation Jinjiang Tu
2026-09-09 10:24 ` David Hildenbrand (Arm)
2026-09-11 2:28 ` Jinjiang Tu [this message]
2026-09-09 9:00 ` [PATCH 2/3] arm64: simplify pmd_special and pmd_mkspecial Jinjiang Tu
2026-09-09 9:00 ` [PATCH 3/3] arm64: Add support for PUD THP Jinjiang Tu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5cb48654-8d77-4525-9854-5823934ae0e8@huawei.com \
--to=tujinjiang@huawei.com \
--cc=akpm@linux-foundation.org \
--cc=catalin.marinas@arm.com \
--cc=dave.jiang@intel.com \
--cc=david@kernel.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-mm@kvack.org \
--cc=ljs@kernel.org \
--cc=mark.rutland@arm.com \
--cc=peterx@redhat.com \
--cc=ryan.roberts@arm.com \
--cc=sunnanyong@huawei.com \
--cc=wangkefeng.wang@huawei.com \
--cc=will@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox