diff mbox series

[v1,08/12] mm: replace MAX_ORDER when it is used to indicate max physical contiguity.

Message ID 20220922011252.2266780-9-zi.yan@sent.com (mailing list archive)
State New
Headers show
Series Make MAX_ORDER adjustable as a kernel boot time parameter. | expand

Commit Message

Zi Yan Sept. 22, 2022, 1:12 a.m. UTC 
From: Zi Yan <ziy@nvidia.com>

MAX_ORDER is limited at a memory section size, thus widely used as
a variable to indicate maximum physically contiguous page size. But this
limitation is no longer necessary as kernel only supports sparse memory
model. Add a new variable MAX_PHYS_CONTIG_ORDER to replace such uses of
MAX_ORDER.

Signed-off-by: Zi Yan <ziy@nvidia.com>
---
 Documentation/admin-guide/kernel-parameters.txt |  2 +-
 arch/sparc/mm/tsb.c                             |  4 ++--
 arch/um/kernel/um_arch.c                        |  4 ++--
 include/linux/pageblock-flags.h                 | 12 ++++++++++++
 kernel/dma/pool.c                               |  8 ++++----
 mm/hugetlb.c                                    |  2 +-
 mm/internal.h                                   |  2 +-
 mm/memory.c                                     |  4 ++--
 mm/memory_hotplug.c                             |  6 +++---
 mm/page_isolation.c                             |  2 +-
 mm/page_reporting.c                             |  2 +-
 11 files changed, 30 insertions(+), 18 deletions(-)
diff mbox series

Patch

diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index e1094851c328..5f633844daac 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -3907,7 +3907,7 @@ 
 			[KNL] Minimal page reporting order
 			Format: <integer>
 			Adjust the minimal page reporting order. The page
-			reporting is disabled when it exceeds MAX_ORDER.
+			reporting is disabled when it exceeds MAX_PHYS_CONTIG_ORDER.
 
 	panic=		[KNL] Kernel behaviour on panic: delay <timeout>
 			timeout > 0: seconds before rebooting
diff --git a/arch/sparc/mm/tsb.c b/arch/sparc/mm/tsb.c
index 912205787161..15c31d050dab 100644
--- a/arch/sparc/mm/tsb.c
+++ b/arch/sparc/mm/tsb.c
@@ -402,8 +402,8 @@  void tsb_grow(struct mm_struct *mm, unsigned long tsb_index, unsigned long rss)
 	unsigned long new_rss_limit;
 	gfp_t gfp_flags;
 
-	if (max_tsb_size > (PAGE_SIZE << MAX_ORDER))
-		max_tsb_size = (PAGE_SIZE << MAX_ORDER);
+	if (max_tsb_size > (PAGE_SIZE << MAX_PHYS_CONTIG_ORDER))
+		max_tsb_size = (PAGE_SIZE << MAX_PHYS_CONTIG_ORDER);
 
 	new_cache_index = 0;
 	for (new_size = 8192; new_size < max_tsb_size; new_size <<= 1UL) {
diff --git a/arch/um/kernel/um_arch.c b/arch/um/kernel/um_arch.c
index e0de60e503b9..52a474f4f1c7 100644
--- a/arch/um/kernel/um_arch.c
+++ b/arch/um/kernel/um_arch.c
@@ -368,10 +368,10 @@  int __init linux_main(int argc, char **argv)
 	max_physmem = TASK_SIZE - uml_physmem - iomem_size - MIN_VMALLOC;
 
 	/*
-	 * Zones have to begin on a 1 << MAX_ORDER page boundary,
+	 * Zones have to begin on a 1 << MAX_PHYS_CONTIG_ORDER page boundary,
 	 * so this makes sure that's true for highmem
 	 */
-	max_physmem &= ~((1 << (PAGE_SHIFT + MAX_ORDER)) - 1);
+	max_physmem &= ~((1 << (PAGE_SHIFT + MAX_PHYS_CONTIG_ORDER)) - 1);
 	if (physmem_size + iomem_size > max_physmem) {
 		highmem = physmem_size + iomem_size - max_physmem;
 		physmem_size -= highmem;
diff --git a/include/linux/pageblock-flags.h b/include/linux/pageblock-flags.h
index 95589b24fff9..5469ba6f97c1 100644
--- a/include/linux/pageblock-flags.h
+++ b/include/linux/pageblock-flags.h
@@ -61,6 +61,18 @@  extern unsigned int pageblock_order;
 #define pageblock_start_pfn(pfn)	ALIGN_DOWN((pfn), pageblock_nr_pages)
 #define pageblock_end_pfn(pfn)		ALIGN((pfn) + 1, pageblock_nr_pages)
 
+/*
+ * memory section is only defined in sparsemem and in flatmem, pages are always
+ * physically contiguous, but we use MAX_ORDER since all users assume so.
+ */
+#ifdef CONFIG_FLATMEM
+#define MAX_PHYS_CONTIG_ORDER	MAX_ORDER
+#else /* SPARSEMEM */
+#define MAX_PHYS_CONTIG_ORDER	(min(PFN_SECTION_SHIFT, MAX_ORDER))
+#endif /* CONFIG_FLATMEM */
+
+#define MAX_PHYS_CONTIG_NR_PAGES	(1UL << MAX_PHYS_CONTIG_ORDER)
+
 /* Forward declaration */
 struct page;
 
diff --git a/kernel/dma/pool.c b/kernel/dma/pool.c
index 1acec2e22827..11af1063eb86 100644
--- a/kernel/dma/pool.c
+++ b/kernel/dma/pool.c
@@ -84,8 +84,8 @@  static int atomic_pool_expand(struct gen_pool *pool, size_t pool_size,
 	void *addr;
 	int ret = -ENOMEM;
 
-	/* Cannot allocate larger than MAX_ORDER */
-	order = min(get_order(pool_size), MAX_ORDER);
+	/* Cannot allocate larger than MAX_PHYS_CONTIG_ORDER */
+	order = min(get_order(pool_size), MAX_PHYS_CONTIG_ORDER);
 
 	do {
 		pool_size = 1 << (PAGE_SHIFT + order);
@@ -190,11 +190,11 @@  static int __init dma_atomic_pool_init(void)
 
 	/*
 	 * If coherent_pool was not used on the command line, default the pool
-	 * sizes to 128KB per 1GB of memory, min 128KB, max MAX_ORDER.
+	 * sizes to 128KB per 1GB of memory, min 128KB, max MAX_PHYS_CONTIG_ORDER.
 	 */
 	if (!atomic_pool_size) {
 		unsigned long pages = totalram_pages() / (SZ_1G / SZ_128K);
-		pages = min_t(unsigned long, pages, MAX_ORDER_NR_PAGES);
+		pages = min_t(unsigned long, pages, MAX_PHYS_CONTIG_NR_PAGES);
 		atomic_pool_size = max_t(size_t, pages << PAGE_SHIFT, SZ_128K);
 	}
 	INIT_WORK(&atomic_pool_work, atomic_pool_work_fn);
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 84a4ea87f394..e6c829a581d6 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -1922,7 +1922,7 @@  pgoff_t hugetlb_basepage_index(struct page *page)
 	pgoff_t index = page_index(page_head);
 	unsigned long compound_idx;
 
-	if (compound_order(page_head) > MAX_ORDER)
+	if (compound_order(page_head) > MAX_PHYS_CONTIG_ORDER)
 		compound_idx = page_to_pfn(page) - page_to_pfn(page_head);
 	else
 		compound_idx = page - page_head;
diff --git a/mm/internal.h b/mm/internal.h
index d688c0320cda..1b1abfc2196e 100644
--- a/mm/internal.h
+++ b/mm/internal.h
@@ -304,7 +304,7 @@  static inline bool page_is_buddy(struct page *page, struct page *buddy,
  * satisfies the following equation:
  *     P = B & ~(1 << O)
  *
- * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
+ * Assumption: *_mem_map is contiguous at least up to MAX_PHYS_CONTIG_ORDER
  */
 static inline unsigned long
 __find_buddy_pfn(unsigned long page_pfn, unsigned int order)
diff --git a/mm/memory.c b/mm/memory.c
index 118e5f023597..06f5f2de3a2c 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -5717,7 +5717,7 @@  void clear_huge_page(struct page *page,
 	unsigned long addr = addr_hint &
 		~(((unsigned long)pages_per_huge_page << PAGE_SHIFT) - 1);
 
-	if (unlikely(pages_per_huge_page > MAX_ORDER_NR_PAGES)) {
+	if (unlikely(pages_per_huge_page > MAX_PHYS_CONTIG_NR_PAGES)) {
 		clear_gigantic_page(page, addr, pages_per_huge_page);
 		return;
 	}
@@ -5769,7 +5769,7 @@  void copy_user_huge_page(struct page *dst, struct page *src,
 		.vma = vma,
 	};
 
-	if (unlikely(pages_per_huge_page > MAX_ORDER_NR_PAGES)) {
+	if (unlikely(pages_per_huge_page > MAX_PHYS_CONTIG_NR_PAGES)) {
 		copy_user_gigantic_page(dst, src, addr, vma,
 					pages_per_huge_page);
 		return;
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index 2985d5302c57..994a648c0f5c 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -596,16 +596,16 @@  static void online_pages_range(unsigned long start_pfn, unsigned long nr_pages)
 	unsigned long pfn;
 
 	/*
-	 * Online the pages in MAX_ORDER aligned chunks. The callback might
+	 * Online the pages in MAX_PHYS_CONTIG_ORDER aligned chunks. The callback might
 	 * decide to not expose all pages to the buddy (e.g., expose them
 	 * later). We account all pages as being online and belonging to this
 	 * zone ("present").
 	 * When using memmap_on_memory, the range might not be aligned to
-	 * MAX_ORDER_NR_PAGES - 1, but pageblock aligned. __ffs() will detect
+	 * MAX_PHYS_CONTIG_NR_PAGES - 1, but pageblock aligned. __ffs() will detect
 	 * this and the first chunk to online will be pageblock_nr_pages.
 	 */
 	for (pfn = start_pfn; pfn < end_pfn;) {
-		int order = min_t(unsigned long, MAX_ORDER, __ffs(pfn));
+		int order = min_t(unsigned long, MAX_PHYS_CONTIG_ORDER, __ffs(pfn));
 
 		(*online_page_callback)(pfn_to_page(pfn), order);
 		pfn += (1UL << order);
diff --git a/mm/page_isolation.c b/mm/page_isolation.c
index 0c2b7d4f7054..2c183827d365 100644
--- a/mm/page_isolation.c
+++ b/mm/page_isolation.c
@@ -226,7 +226,7 @@  static void unset_migratetype_isolate(struct page *page, int migratetype)
 	 */
 	if (PageBuddy(page)) {
 		order = buddy_order(page);
-		if (order >= pageblock_order && order <= MAX_ORDER) {
+		if (order >= pageblock_order && order <= MAX_PHYS_CONTIG_ORDER) {
 			buddy = find_buddy_page_pfn(page, page_to_pfn(page),
 						    order, NULL);
 			if (buddy && !is_migrate_isolate_page(buddy)) {
diff --git a/mm/page_reporting.c b/mm/page_reporting.c
index e3d6f4af34e0..001438f3dbeb 100644
--- a/mm/page_reporting.c
+++ b/mm/page_reporting.c
@@ -248,7 +248,7 @@  page_reporting_process_zone(struct page_reporting_dev_info *prdev,
 		return err;
 
 	/* Process each free list starting from lowest order/mt */
-	for (order = page_reporting_order; order <= MAX_ORDER; order++) {
+	for (order = page_reporting_order; order <= MAX_PHYS_CONTIG_ORDER; order++) {
 		for (mt = 0; mt < MIGRATE_TYPES; mt++) {
 			/* We do not pull pages from the isolate free list */
 			if (is_migrate_isolate(mt))