[PATCH v2 05/17] mm/sparse-vmemmap: support section-based vmemmap accounting

[Muchun Song]

section_nr_vmemmap_pages() can account ordinary sections and DAX sections,
but section-based vmemmap optimization keeps its compound order in struct
mem_section and retains a different number of vmemmap pages.

Teach section_nr_vmemmap_pages() to recognize section-based optimized
sections and calculate their vmemmap page count from the section order
and the HVO retained page count.

Signed-off-by: Muchun Song <songmuchun@xxxxxxxxxxxxx>
---
Hi Mike,

I reintroduced VMEMMAP_OPTIMIZATION_PAGES (the intermediate macro), which we
discussed removing in v1, so the section-based HVO path doesn't have to
hard-code its retained vmemmap page count (i.e. 1) in section_nr_vmemmap_pages().
To better unify the codebase in the future, the new macro introduced in this patch
will also be applied to the HugeTLB code (e.g., to calculate
HUGETLB_VMEMMAP_RESERVE_SIZE from this new macro as well).

v2:
- Remove an unnecessary vmemmap_can_optimize() call to simplify the code
  (suggested by Mike Rapoport).
- Rewrite the commit message for better understanding.
---
 include/linux/mmzone.h |  6 ++++--
 mm/sparse-vmemmap.c    | 10 ++++++----
 mm/sparse.h            |  8 ++++++++
 3 files changed, 18 insertions(+), 6 deletions(-)

diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index 2a32101d55e6..025ea69f57d8 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -107,8 +107,10 @@
 	 is_power_of_2(sizeof(struct page)) ? \
 	 MAX_FOLIO_NR_PAGES * sizeof(struct page) : 0)
 
-/* The number of struct pages covered by the retained vmemmap pages with HVO enabled. */
-#define VMEMMAP_OPTIMIZATION_NR_STRUCT_PAGES	(PAGE_SIZE / sizeof(struct page))
+/* The number of retained vmemmap pages with HVO enabled. */
+#define VMEMMAP_OPTIMIZATION_PAGES		1
+#define VMEMMAP_OPTIMIZATION_NR_STRUCT_PAGES	\
+	(VMEMMAP_OPTIMIZATION_PAGES * PAGE_SIZE / sizeof(struct page))
 #define VMEMMAP_OPTIMIZATION_MIN_ORDER		(ilog2(VMEMMAP_OPTIMIZATION_NR_STRUCT_PAGES) + 1)
 
 #define __VMEMMAP_OPTIMIZATION_NR_ORDERS	\
diff --git a/mm/sparse-vmemmap.c b/mm/sparse-vmemmap.c
index 107215cf8488..b7abc5494bb9 100644
--- a/mm/sparse-vmemmap.c
+++ b/mm/sparse-vmemmap.c
@@ -649,24 +649,26 @@ void offline_mem_sections(unsigned long start_pfn, unsigned long end_pfn)
 static int __meminit section_nr_vmemmap_pages(unsigned long pfn, unsigned long nr_pages,
 		struct vmem_altmap *altmap, struct dev_pagemap *pgmap)
 {
-	const unsigned int order = pgmap ? pgmap->vmemmap_shift : 0;
+	const struct mem_section *ms = __pfn_to_section(pfn);
+	const int order = pgmap ? pgmap->vmemmap_shift : section_order(ms);
+	const int vmemmap_pages = pgmap ? VMEMMAP_RESERVE_NR : VMEMMAP_OPTIMIZATION_PAGES;
 	const unsigned long pages_per_compound = 1UL << order;
 
 	VM_WARN_ON_ONCE(!IS_ALIGNED(pfn | nr_pages, PAGES_PER_SUBSECTION));
 	VM_WARN_ON_ONCE(nr_pages > PAGES_PER_SECTION);
 
-	if (!vmemmap_can_optimize(altmap, pgmap))
+	if (!vmemmap_can_optimize(altmap, pgmap) && !section_vmemmap_optimizable(ms))
 		return DIV_ROUND_UP(nr_pages * sizeof(struct page), PAGE_SIZE);
 
 	if (order < PFN_SECTION_SHIFT) {
 		VM_WARN_ON_ONCE(!IS_ALIGNED(pfn | nr_pages, pages_per_compound));
-		return VMEMMAP_RESERVE_NR * nr_pages / pages_per_compound;
+		return vmemmap_pages * nr_pages / pages_per_compound;
 	}
 
 	VM_WARN_ON_ONCE(!IS_ALIGNED(pfn | nr_pages, PAGES_PER_SECTION));
 
 	if (IS_ALIGNED(pfn, pages_per_compound))
-		return VMEMMAP_RESERVE_NR;
+		return vmemmap_pages;
 
 	return 0;
 }
diff --git a/mm/sparse.h b/mm/sparse.h
index 030248030dc7..1e9fcea10181 100644
--- a/mm/sparse.h
+++ b/mm/sparse.h
@@ -74,6 +74,14 @@ static inline bool pfn_vmemmap_optimizable(unsigned long pfn)
 	return (pfn & (nr_pages - 1)) >= VMEMMAP_OPTIMIZATION_NR_STRUCT_PAGES;
 }
 
+static inline bool section_vmemmap_optimizable(const struct mem_section *section)
+{
+	if (!is_power_of_2(sizeof(struct page)))
+		return false;
+
+	return section_order(section) >= VMEMMAP_OPTIMIZATION_MIN_ORDER;
+}
+
 /*
  * mm/sparse-vmemmap.c
  */
-- 
2.54.0