| Message ID | 20230724190759.483013-1-aneesh.kumar@linux.ibm.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | Add support for DAX vmemmap optimization for ppc64 | expand |
On Tue, 25 Jul 2023 00:37:46 +0530 "Aneesh Kumar K.V" <aneesh.kumar@linux.ibm.com> wrote: > This patch series implements changes required to support DAX vmemmap > optimization for ppc64. Do we have any measurements to help us understand the magnitude of this optimization? And any documentation which helps users understand whether and why they should enable this feature?
On 7/26/23 12:59 AM, Andrew Morton wrote: > On Tue, 25 Jul 2023 00:37:46 +0530 "Aneesh Kumar K.V" <aneesh.kumar@linux.ibm.com> wrote: > >> This patch series implements changes required to support DAX vmemmap >> optimization for ppc64. > > Do we have any measurements to help us understand the magnitude > of this optimization? > > And any documentation which helps users understand whether and > why they should enable this feature? That is memory space optimization due to kernel reusing the tail page struct pages. The details of the optimization is documented in patch 11. We document there the impact with both 4k and 64K page size. -aneesh
On Wed, 26 Jul 2023 10:59:32 +0530 Aneesh Kumar K V <aneesh.kumar@linux.ibm.com> wrote: > On 7/26/23 12:59 AM, Andrew Morton wrote: > > On Tue, 25 Jul 2023 00:37:46 +0530 "Aneesh Kumar K.V" <aneesh.kumar@linux.ibm.com> wrote: > > > >> This patch series implements changes required to support DAX vmemmap > >> optimization for ppc64. > > > > Do we have any measurements to help us understand the magnitude > > of this optimization? > > > > And any documentation which helps users understand whether and > > why they should enable this feature? > > That is memory space optimization due to kernel reusing the tail page struct pages. The details > of the optimization is documented in patch 11. We document there the impact with both 4k and > 64K page size. I suppose that with sufficient arithmetic one could use Documentation/powerpc/vmemmap_dedup.rst to figure out the bottom-line savings. I was more expecting a straightforward statement in the [0/N] overview to help people understand why they're reading this patchset at all. Like "saves 5% of total memory on my XXX machine".
Andrew Morton <akpm@linux-foundation.org> writes: > On Wed, 26 Jul 2023 10:59:32 +0530 Aneesh Kumar K V <aneesh.kumar@linux.ibm.com> wrote: > >> On 7/26/23 12:59 AM, Andrew Morton wrote: >> > On Tue, 25 Jul 2023 00:37:46 +0530 "Aneesh Kumar K.V" <aneesh.kumar@linux.ibm.com> wrote: >> > >> >> This patch series implements changes required to support DAX vmemmap >> >> optimization for ppc64. >> > >> > Do we have any measurements to help us understand the magnitude >> > of this optimization? >> > >> > And any documentation which helps users understand whether and >> > why they should enable this feature? >> >> That is memory space optimization due to kernel reusing the tail page struct pages. The details >> of the optimization is documented in patch 11. We document there the impact with both 4k and >> 64K page size. > > I suppose that with sufficient arithmetic one could use > Documentation/powerpc/vmemmap_dedup.rst to figure out the bottom-line > savings. > > I was more expecting a straightforward statement in the [0/N] overview > to help people understand why they're reading this patchset at all. > Like "saves 5% of total memory on my XXX machine". This is specific to devdax usage and also depends on devdax alignment. The actual saving details are also documented in mm/vmemmap_dedup.rst. The saving will be based on the devdax device memory size and aligment. With 64K page size for 16384 pages added (1G) we save 14 pages With 4K page size for 262144 pages added (1G) we save 4094 pages With 4K page size for 512 pages added (2M) we save 6 pages -aneesh
This patch series implements changes required to support DAX vmemmap optimization for ppc64. The vmemmap optimization is only enabled with radix MMU translation and 1GB PUD mapping with 64K page size. The patch series also split hugetlb vmemmap optimization as a separate Kconfig variable so that architectures can enable DAX vmemmap optimization without enabling hugetlb vmemmap optimization. This should enable architectures like arm64 to enable DAX vmemmap optimization while they can't enable hugetlb vmemmap optimization. More details of the same are in patch "mm/vmemmap optimization: Split hugetlb and devdax vmemmap optimization" Changes from v5: * rebase to mm-unstable branch Changes from v4: * Address review feedback * Add the Reviewed-by: Changes from v3: * Rebase to latest linus tree * Build fix with SPARSEMEM_VMEMMP disabled * Add hash_pud_same outisde THP Kconfig Changes from v2: * Rebase to latest linus tree * Address review feedback Changes from V1: * Fix make htmldocs warning * Fix vmemmap allocation bugs with different alignment values. * Correctly check for section validity to before we free vmemmap area Aneesh Kumar K.V (13): mm/hugepage pud: Allow arch-specific helper function to check huge page pud support mm: Change pudp_huge_get_and_clear_full take vm_area_struct as arg mm/vmemmap: Improve vmemmap_can_optimize and allow architectures to override mm/vmemmap: Allow architectures to override how vmemmap optimization works mm: Add pud_same similar to __HAVE_ARCH_P4D_SAME mm/huge pud: Use transparent huge pud helpers only with CONFIG_TRANSPARENT_HUGEPAGE mm/vmemmap optimization: Split hugetlb and devdax vmemmap optimization powerpc/mm/trace: Convert trace event to trace event class powerpc/book3s64/mm: Enable transparent pud hugepage powerpc/book3s64/vmemmap: Switch radix to use a different vmemmap handling function powerpc/book3s64/radix: Add support for vmemmap optimization for radix powerpc/book3s64/radix: Remove mmu_vmemmap_psize powerpc/book3s64/radix: Add debug message to give more details of vmemmap allocation Documentation/mm/vmemmap_dedup.rst | 1 + Documentation/powerpc/index.rst | 1 + Documentation/powerpc/vmemmap_dedup.rst | 101 ++++ arch/loongarch/Kconfig | 2 +- arch/powerpc/Kconfig | 1 + arch/powerpc/include/asm/book3s/64/hash.h | 9 + arch/powerpc/include/asm/book3s/64/pgtable.h | 155 ++++- arch/powerpc/include/asm/book3s/64/radix.h | 47 ++ .../include/asm/book3s/64/tlbflush-radix.h | 2 + arch/powerpc/include/asm/book3s/64/tlbflush.h | 8 + arch/powerpc/include/asm/pgtable.h | 6 + arch/powerpc/mm/book3s64/hash_pgtable.c | 2 +- arch/powerpc/mm/book3s64/pgtable.c | 78 +++ arch/powerpc/mm/book3s64/radix_pgtable.c | 572 ++++++++++++++++-- arch/powerpc/mm/book3s64/radix_tlb.c | 7 + arch/powerpc/mm/init_64.c | 37 +- arch/powerpc/platforms/Kconfig.cputype | 1 + arch/riscv/Kconfig | 2 +- arch/s390/Kconfig | 2 +- arch/x86/Kconfig | 3 +- drivers/nvdimm/pfn_devs.c | 2 +- fs/Kconfig | 2 +- include/linux/mm.h | 29 +- include/linux/pgtable.h | 12 +- include/trace/events/thp.h | 33 +- mm/Kconfig | 5 +- mm/debug_vm_pgtable.c | 2 +- mm/huge_memory.c | 2 +- mm/mm_init.c | 2 +- mm/mremap.c | 2 +- mm/sparse-vmemmap.c | 3 + 31 files changed, 1049 insertions(+), 82 deletions(-) create mode 100644 Documentation/powerpc/vmemmap_dedup.rst