[PATCH 00/17] arm64/mm: Enable 128 bit page table entries
Anshuman Khandual
anshuman.khandual at arm.com
Tue Sep 1 00:14:09 PDT 2026
On Wed, Jul 29, 2026 at 05:54:35PM +0530, Anshuman Khandual wrote:
> FEAT_D128 is a new arm architecture feature adding support for VMSAv9-128
> translation system. FEAT_D128 is an optional feature from ARMV9.3 onwards.
> arm64 platforms with FEAT_D128 enabled will now have two different memory
> translation systems i.e VMSAv8-64 and VMSAv9-128.
>
> FEAT_D128 adds 128 bit page table entries, thus supporting larger physical
> and virtual address range while also expanding available room for more MMU
> feature bits for both HW and SW usage.
>
> A new config option ARM64_D128 is being added thus enabling users to chose
> between aforementioned memory translations regimes VMSAv8-64 (aka D64) and
> VMSAv9-128 (aka D128) during the build time. If the ARM64_D128 built kernel
> does not detect FEAT_D128 feature on the hardware, then it gets stuck with
> with a new reason code CPU_STUCK_REASON_NO_D128. So ARM64_D128 built kernel
> cannot fallback to D64 translation regime on unsupported HW.
>
> READ_ONCE() on page table entries get routed via level specific pxdp_get()
> helpers which platforms could then override when required. These accessors
> on arm64 platform help in ensuring page table accesses are performed in an
> atomic manner while reading 128 bit page table entries.
>
> ARM64_[VA|PA]_BITS combinations for respective CONFIG_ARM64_XXK_PAGES are
> now supported both on VMSAv8-64 and VMSAv9-128 memory translation regimes.
> Although FEAT_D128 provided 52 bit VA and PA space along with skip levels
> features are not supported currently.
>
> Basic page table geometry has also been changed with VMSAv9-128, as there
> are fewer entries per level as compared to VMSAv8-64. Please refer to the
> following table for various leaf entry sizes including the cont entries.
>
> ------------------------------------------------
> | | D64 | D128 |
> ------------------------------------------------
> | PAGE_SIZE | PMD | PUD | PMD | PUD |
> -----------------------------|-----------------|
> | 4K | 2M | 1G | 1M | 256M |
> | 16K | 32M | 64G | 16M | 16G |
> | 64K | 512M | 4T | 256M | 1T |
> ------------------------------------------------
>
> --------------------------------------------------------------------
> | | D64 | D128 |
> --------------------------------------------------------------------
> | PAGE_SIZE | CONT_PTE | CONT_PMD | CONT_PTE | CONT_PMD |
> --------------------------|------------|-------------|--------------
> | 4K | 64K | 32M | 64K | 16M |
> | 16K | 2M | 1G | 1M | 256M |
> | 64K | 2M | 16G | 1M | 16G |
> --------------------------------------------------------------------
>
> Similarly arm64 kernel features such as KVM, KASAN and UNMAP_KERNEL_AT_EL0
> are currently not supported as well.
>
> This series applies on v7.2-rc5 but after applying the following patch that
> standardized printing pgtable entries. There are no apparent problems while
> running MM kselftests with and without CONFIG_ARM64_D128.
>
> https://lore.kernel.org/all/20260709044334.1741263-1-anshuman.khandual@arm.com/
>
> Changes in V1:
>
> - Rebased on v7.2-rc5
> - Pulled in new standardized pgtable entry print patch
> - Added standardized pgtable entry print on arm64 platform
> - Added KASAN related pgdp_get() changes missing earlier
> - Renamed pxxval_get/set() as ptval_get/set()
> - Renamed ptval_cmpxchg_relaxed() and ptval_xchg_relaxed()
> - Renamed tlbi_args_t as tlbi_arg_t per Ryan
> - Split final D128 enablement patch into four separate ones
> - Updated all commit messages as required
>
> Changes in RFC V2:
>
> https://lore.kernel.org/all/20260513044547.4128549-1-anshuman.khandual@arm.com/
>
> - Dropped some patches that were merged upstream and rebased on v7.1-rc3
> - Updated commit message for the 5-level fixmap change per David
> - Moved pxdval_t definition inside generic page table header
> - Dropped _once from pgprot_read/write() per Mike
> - Updated ARM64_CONT_[PTE|PMD]_SHIFT both for 16K and 64K base pages
> - Renamed __PRIpte as __PRIpxx per David
> - Added abstraction for tlbi_op
> - Adopted TLBIP implementation to recent TLB flush changes
> - Moved back all helpers back from arch/arm64/mm/mmu.c into the header
> - Moved arm64 pgtable header READ_ONCE() replacements later in the series
> - Renamed all ptdesc_ instances as pxxval_ instead
> - Updated all commit messages as required and suggested
> - Restored print format in __print_bad_page_map_pgtable() per Usama
>
> Changes in RFC V1:
>
> https://lore.kernel.org/linux-arm-kernel/20260224051153.3150613-2-anshuman.khandual@arm.com/
>
> Cc: Catalin Marinas <catalin.marinas at arm.com>
> Cc: Will Deacon <will at kernel.org>
> Cc: Ryan Roberts <ryan.roberts at arm.com>
> Cc: Mark Rutland <mark.rutland at arm.com>
> Cc: Lorenzo Stoakes <ljs at kernel.org>
> Cc: Andrew Morton <akpm at linux-foundation.org>
> Cc: David Hildenbrand <david at kernel.org>
> Cc: Mike Rapoport <rppt at kernel.org>
> Cc: Linu Cherian <linu.cherian at arm.com>
> Cc: linux-arm-kernel at lists.infradead.org
> Cc: linux-kernel at vger.kernel.org
> Cc: linux-mm at kvack.org
>
> Anshuman Khandual (15):
> mm: Add read-write accessors for vm_page_prot
> arm64/mm: Convert READ_ONCE() as pmdp_get() while accessing PMD
> arm64/mm: Convert READ_ONCE() as pudp_get() while accessing PUD
> arm64/mm: Convert READ_ONCE() as p4dp_get() while accessing P4D
> arm64/mm: Convert READ_ONCE() as pgdp_get() while accessing PGD
> arm64/mm: Route all pgtable reads via ptval_get()
> arm64/mm: Route all pgtable writes via ptval_set()
> arm64/mm: Route all pgtable atomics to central helpers
> arm64/mm: Override read-write accessors for vm_page_prot
> arm64/mm: Standardize printing for pgtable entries
Page table entries print standardization patch has been taken out
from this series here and posted separately which uses a recently
added generic MM helper ptval_to_str().
https://lore.kernel.org/all/20260901065454.1906343-1-anshuman.khandual@arm.com/
Similarly the following READ_ONCE() ---> pxdp_get() conversion
patches too could also be split out from this series here and
reviewed independently.
arm64/mm: Convert READ_ONCE() as pmdp_get() while accessing PMD
arm64/mm: Convert READ_ONCE() as pudp_get() while accessing PUD
arm64/mm: Convert READ_ONCE() as p4dp_get() while accessing P4D
arm64/mm: Convert READ_ONCE() as pgdp_get() while accessing PGD
Please do let me know if there are any concerns.
- Anshuman
> arm64/mm: Enable fixmap with 5 level page table
> arm64/mm: Enable pgtable geometry for FEAT_D128
> arm64/mm: Enable pgtable descriptor for FEAT_D128
> arm64/mm: Enable 128 bit translation
> arm64/mm: Make ARM64_D128 selectable
>
> Linu Cherian (2):
> arm64/mm: Add an abstraction level for tlbi_op
> arm64/mm: Enable TLBIP instruction based TLB flush
>
> arch/arm64/Kconfig | 51 ++++++-
> arch/arm64/Makefile | 4 +
> arch/arm64/include/asm/assembler.h | 4 +-
> arch/arm64/include/asm/el2_setup.h | 9 ++
> arch/arm64/include/asm/pgtable-hwdef.h | 137 ++++++++++++++++++
> arch/arm64/include/asm/pgtable-prot.h | 18 ++-
> arch/arm64/include/asm/pgtable-types.h | 4 +
> arch/arm64/include/asm/pgtable.h | 188 +++++++++++++++++++------
> arch/arm64/include/asm/smp.h | 1 +
> arch/arm64/include/asm/tlbflush.h | 138 ++++++++++++------
> arch/arm64/kernel/head.S | 12 ++
> arch/arm64/mm/fault.c | 31 ++--
> arch/arm64/mm/fixmap.c | 24 +++-
> arch/arm64/mm/hugetlbpage.c | 10 +-
> arch/arm64/mm/kasan_init.c | 18 +--
> arch/arm64/mm/mmu.c | 81 ++++++-----
> arch/arm64/mm/pageattr.c | 8 +-
> arch/arm64/mm/proc.S | 25 +++-
> arch/arm64/mm/trans_pgd.c | 14 +-
> include/linux/mm_types.h | 40 ++++++
> include/linux/pgtable.h | 14 ++
> mm/huge_memory.c | 4 +-
> mm/memory.c | 42 +-----
> mm/migrate.c | 2 +-
> mm/mmap.c | 2 +-
> 25 files changed, 668 insertions(+), 213 deletions(-)
>
> --
> 2.43.0
>
More information about the linux-arm-kernel
mailing list