[External] Re: [RFC PATCH v2 00/21] riscv: Introduce 64K base page

Xu Lu luxu.kernel at bytedance.com
Thu Dec 5 18:41:28 PST 2024


Hi Zi Yan,

On Fri, Dec 6, 2024 at 10:00 AM Zi Yan <ziy at nvidia.com> wrote:
>
> On 5 Dec 2024, at 5:37, Xu Lu wrote:
>
> > This patch series attempts to break through the limitation of MMU and
> > supports larger base page on RISC-V, which only supports 4K page size
> > now. The key idea is to always manage and allocate memory at a
> > granularity of 64K and use SVNAPOT to accelerate address translation.
> > This is the second version and the detailed introduction can be found
> > in [1].
> >
> > Changes from v1:
> > - Rebase on v6.12.
> >
> > - Adjust the page table entry shift to reduce page table memory usage.
> >     For example, in SV39, the traditional va behaves as:
> >
> >     ----------------------------------------------
> >     | pgd index | pmd index | pte index | offset |
> >     ----------------------------------------------
> >     | 38     30 | 29     21 | 20     12 | 11   0 |
> >     ----------------------------------------------
> >
> >     When we choose 64K as basic software page, va now behaves as:
> >
> >     ----------------------------------------------
> >     | pgd index | pmd index | pte index | offset |
> >     ----------------------------------------------
> >     | 38     34 | 33     25 | 24     16 | 15   0 |
> >     ----------------------------------------------
> >
> > - Fix some bugs in v1.
> >
> > Thanks in advance for comments.
> >
> > [1] https://lwn.net/Articles/952722/
>
> This looks very interesting. Can you cc me and linux-mm at kvack.org
> in the future? Thanks.

Of course. Hope this patch can be of any help.

>
> Have you thought about doing it for ARM64 4KB as well? ARM64’s contig PTE
> should have similar effect of RISC-V’s SVNAPOT, right?

I have not thought about it yet. ARM64 has native 64K MMU. The kernel
can directly configure the page size as 64K and MMU will do
translation at corresponding granularity. So I doubt if there is a
need to implement 64K Page Size based on CONT PTE. If you want to use
CONT PTE for acceleration instead of 64K MMU, maybe you can have a try
on THP_CONTPTE[1] which has been merged~

[1] https://lwn.net/Articles/935887/

Best regards,

Xu Lu

>
> >
> > Xu Lu (21):
> >   riscv: mm: Distinguish hardware base page and software base page
> >   riscv: mm: Configure satp with hw page pfn
> >   riscv: mm: Reimplement page table entry structures
> >   riscv: mm: Reimplement page table entry constructor function
> >   riscv: mm: Reimplement conversion functions between page table entry
> >   riscv: mm: Avoid pte constructor during pte conversion
> >   riscv: mm: Reimplement page table entry get function
> >   riscv: mm: Reimplement page table entry atomic get function
> >   riscv: mm: Replace READ_ONCE with atomic pte get function
> >   riscv: mm: Reimplement PTE A/D bit check function
> >   riscv: mm: Reimplement mk_huge_pte function
> >   riscv: mm: Reimplement tlb flush function
> >   riscv: mm: Adjust PGDIR/P4D/PUD/PMD_SHIFT
> >   riscv: mm: Only apply svnapot region bigger than software page
> >   riscv: mm: Adjust FIX_BTMAPS_SLOTS for variable PAGE_SIZE
> >   riscv: mm: Adjust FIX_FDT_SIZE for variable PMD_SIZE
> >   riscv: mm: Apply Svnapot for base page mapping if possible
> >   riscv: Kconfig: Introduce 64K page size
> >   riscv: Kconfig: Adjust mmap rnd bits for 64K Page
> >   riscv: mm: Adjust address space layout and init page table for 64K
> >     Page
> >   riscv: mm: Update EXEC_PAGESIZE for 64K Page
> >
> >  arch/riscv/Kconfig                    |  34 +-
> >  arch/riscv/include/asm/fixmap.h       |   3 +-
> >  arch/riscv/include/asm/hugetlb.h      |   5 +
> >  arch/riscv/include/asm/page.h         |  56 ++-
> >  arch/riscv/include/asm/pgtable-32.h   |  12 +-
> >  arch/riscv/include/asm/pgtable-64.h   | 128 ++++--
> >  arch/riscv/include/asm/pgtable-bits.h |   3 +-
> >  arch/riscv/include/asm/pgtable.h      | 564 +++++++++++++++++++++++---
> >  arch/riscv/include/asm/tlbflush.h     |  26 +-
> >  arch/riscv/include/uapi/asm/param.h   |  24 ++
> >  arch/riscv/kernel/head.S              |   4 +-
> >  arch/riscv/kernel/hibernate.c         |  21 +-
> >  arch/riscv/mm/context.c               |   7 +-
> >  arch/riscv/mm/fault.c                 |  15 +-
> >  arch/riscv/mm/hugetlbpage.c           |  30 +-
> >  arch/riscv/mm/init.c                  |  45 +-
> >  arch/riscv/mm/kasan_init.c            |   7 +-
> >  arch/riscv/mm/pgtable.c               | 111 ++++-
> >  arch/riscv/mm/tlbflush.c              |  31 +-
> >  arch/s390/include/asm/hugetlb.h       |   2 +-
> >  include/asm-generic/hugetlb.h         |   5 +-
> >  include/linux/pgtable.h               |  21 +
> >  kernel/events/core.c                  |   6 +-
> >  mm/debug_vm_pgtable.c                 |   6 +-
> >  mm/gup.c                              |  10 +-
> >  mm/hmm.c                              |   2 +-
> >  mm/hugetlb.c                          |   4 +-
> >  mm/mapping_dirty_helpers.c            |   2 +-
> >  mm/memory.c                           |   4 +-
> >  mm/mprotect.c                         |   2 +-
> >  mm/ptdump.c                           |   8 +-
> >  mm/sparse-vmemmap.c                   |   2 +-
> >  mm/vmscan.c                           |   2 +-
> >  33 files changed, 1029 insertions(+), 173 deletions(-)
> >  create mode 100644 arch/riscv/include/uapi/asm/param.h
> >
> > --
> > 2.20.1
>
>
> Best Regards,
> Yan, Zi



More information about the linux-riscv mailing list