On 24/07/2023 14:46, Usama Arif wrote:
If the region is for gigantic hugepages and if HVO is enabled, then those struct pages which will be freed later by HVO don't need to be prepared and initialized. This can save significant time when a large number of hugepages are allocated at boot time. For a 1G hugepage, this series avoid initialization and preparation of 262144 - 64 = 262080 struct pages per hugepage. When tested on a 512G system (which can allocate max 500 1G hugepages), the kexec-boot time with HVO and DEFERRED_STRUCT_PAGE_INIT enabled without this patchseries to running init is 3.9 seconds. With this patch it is 1.2 seconds. This represents an approximately 70% reduction in boot time and will significantly reduce server downtime when using a large number of gigantic pages.
There were a few errors reported by kernel-bot if different config options were changed (CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP disabled, CONFIG_CMA enabled). I will fix these in the next revision if the general approach in the patches looks good to the community to start review.
Thanks, Usama
Thanks, Usama Usama Arif (4): mm/hugetlb: Skip prep of tail pages when HVO is enabled mm/memblock: Add hugepage_size member to struct memblock_region mm/hugetlb_vmemmap: Use nid of the head page to reallocate it mm/memblock: Skip initialization of struct pages freed later by HVO arch/arm64/mm/kasan_init.c | 2 +- arch/powerpc/platforms/pasemi/iommu.c | 2 +- arch/powerpc/platforms/pseries/setup.c | 4 +- arch/powerpc/sysdev/dart_iommu.c | 2 +- include/linux/memblock.h | 8 +- mm/cma.c | 4 +- mm/hugetlb.c | 36 +++++--- mm/hugetlb_vmemmap.c | 6 +- mm/hugetlb_vmemmap.h | 4 + mm/memblock.c | 87 +++++++++++++------- mm/mm_init.c | 2 +- mm/sparse-vmemmap.c | 2 +- tools/testing/memblock/tests/alloc_nid_api.c | 2 +- 13 files changed, 106 insertions(+), 55 deletions(-)
