Re: [PATCH V4 0/4] arm64/mm: Convert pgtable READ_ONCE() as pxdp_get()

From: Anshuman Khandual

Date: Fri Nov 13 2026 - 00:03:59 EST


On Thu, Nov 05, 2026 at 10:03:59AM +0530, Anshuman Khandual wrote:
> Route READ_ONCE() accesses on pgtable entries via level specific generic MM
> pxdp_get() helpers which platforms could then override when required. These
> accessors on arm64, will help in ensuring page table accesses are performed
> in an atomic manner while reading 128 bit page table entries later. But for
> now they just default to READ_ONCE(), thus preserving existing behaviour on
> current D64 translation a well.
>
> This is similar to generic MM where all atomic accesses have already been
> converted from READ_ONCE() into pgtable level specific accessors.
>
> These are preparatory patches for D128 enablement, although standing their
> own for D64 translation as well. These patches are split from earlier D128
> V1 series. There are no functional changes here, and besides mm kselftests
> do not show any regression.
>
> https://lore.kernel.org/linux-mm/20260729122452.3797443-1-anshuman.khandual@xxxxxxx/
>
> This series applies on v7.4-rc1 and have been hosted below for reference.
>
> git@xxxxxxxxxxxxxxxxxx:linux-arm/linux-anshuman.git (arm64_pxdp_get_v4)
>
> Changes in V4:
>
> - Dropped all pxdp_get() platform overrides per Catalin
> - Moved pmdp_test_and_clear_young() and [pud|p4d]_offset[_phys]()
> helpers inside arch/arm64/mm/mmu.c to avoid build problems
> - Updated all commit messages as required

This series has run into some KASAN fault issues during boot with 5 level
pgtables but also unlike Sashio suggestions these would require more code
churn to fix. Please find the details below.

https://lore.kernel.org/all/ohyhflzuoxdemccemutoosmj5hsfwylrlt6tvnyehulerk7gf3@ada6wu4zrzkp/

NOTE: somehow Sashiko had dropped LAKML from the thread although it shows
up in MM list.

Given the above scenario, I would suggest falling back on the V3 series,
that defined and override pxdp_get() helpers for arm64 platform. Although
similar to generic MM default fallback, they prevented code movement into
arch/arm64/mm/mmu.c file causing these KASAN issues in the first place.

https://lore.kernel.org/all/20260924062214.1635078-1-anshuman.khandual@xxxxxxx/

V3 series apples cleanly, builds, boots and also clears MM selftests on
v7.4-rc2 as well.

>
> Changes in V3:
>
> https://lore.kernel.org/all/20260924062214.1635078-1-anshuman.khandual@xxxxxxx/
>
> - Added overrides for pxdp_get() and replaced remaining READ_ONCE()
> - Updated all commit messages as required
> - Dropped previous tags
>
> Changes in V2:
>
> https://lore.kernel.org/all/20260922061638.3967467-1-anshuman.khandual@xxxxxxx/
>
> - Rebased on v7.3-rc4 along with all latest dependencies
> - Updated commit messages per Ryan
> - Added tags from Ryan
>
> Changes in V1:
>
> https://lore.kernel.org/all/20260903060657.2717457-1-anshuman.khandual@xxxxxxx/
>
> Cc: Catalin Marinas <catalin.marinas@xxxxxxx>
> Cc: Will Deacon <will@xxxxxxxxxx>
> Cc: Ryan Roberts <ryan.roberts@xxxxxxx>
> Cc: Mark Rutland <mark.rutland@xxxxxxx>
> Cc: Lorenzo Stoakes <ljs@xxxxxxxxxx>
> Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> Cc: David Hildenbrand <david@xxxxxxxxxx>
> Cc: Mike Rapoport <rppt@xxxxxxxxxx>
> Cc: Linu Cherian <linu.cherian@xxxxxxx>
> Cc: linux-arm-kernel@xxxxxxxxxxxxxxxxxxx
> Cc: linux-kernel@xxxxxxxxxxxxxxx
> Cc: linux-mm@xxxxxxxxx
>
> Anshuman Khandual (4):
> arm64/mm: Use pmdp_get() for PMD accesses
> arm64/mm: Use pudp_get() for PUD accesses
> arm64/mm: Use p4dp_get() for P4D accesses
> arm64/mm: Use pgdp_get() for PGD accesses
>
> arch/arm64/include/asm/pgtable.h | 38 +++----------
> arch/arm64/mm/fault.c | 8 +--
> arch/arm64/mm/fixmap.c | 6 +-
> arch/arm64/mm/hugetlbpage.c | 10 ++--
> arch/arm64/mm/kasan_init.c | 18 +++---
> arch/arm64/mm/mmu.c | 97 ++++++++++++++++++++++----------
> arch/arm64/mm/pageattr.c | 8 +--
> arch/arm64/mm/trans_pgd.c | 14 ++---
> 8 files changed, 108 insertions(+), 91 deletions(-)
>
> --
> 2.43.0
>