[PATCH mm-new resend 0/5] mm: accelerate gigantic folio allocation

[Kefeng Wang <wangkefeng.wang@huawei.com>]

Optimize pfn_range_valid_contig() and replace_free_hugepage_folios() in
alloc_contig_frozen_pages() to speed up gigantic folio allocation. The
allocation time for 120*1G folios drops from 3.605s to 0.431s.

resend:
- no function changes,
  - split out the page_is_unmovable() into a separate patch, per Andrew 
  - mark two global values with __ro_after_init in patch5

---
This is part of [patch v2 0/8] mm: hugetlb: allocate frozen gigantic folio[1],
and as suggested by David[2], reuses some logic from the has_unmovable_pages().

[1] https://lore.kernel.org/linux-mm/20250918132000.1951232-1-wangkefeng.wang@huawei.com/
[2] https://lore.kernel.org/linux-mm/17bd2977-6a04-402f-ae84-306167cd16d6@redhat.com/

Kefeng Wang (5):
  mm: page_isolation: introduce page_is_unmovable()
  mm: page_alloc: optimize pfn_range_valid_contig()
  mm: hugetlb: optimize replace_free_hugepage_folios()
  mm: hugetlb_cma: optimize hugetlb_cma_alloc_frozen_folio()
  mm: hugetlb_cma: mark hugetlb_cma{_only} as __ro_after_init

 include/linux/page-isolation.h |   2 +
 mm/hugetlb.c                   |  54 +++++++---
 mm/hugetlb_cma.c               |   9 +-
 mm/page_alloc.c                |  25 ++---
 mm/page_isolation.c            | 187 +++++++++++++++++----------------
 5 files changed, 157 insertions(+), 120 deletions(-)

-- 
2.27.0