Linux-mm Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Muchun Song <songmuchun@bytedance.com>
To: Andrew Morton <akpm@linux-foundation.org>,
	David Hildenbrand <david@kernel.org>,
	Muchun Song <muchun.song@linux.dev>,
	Oscar Salvador <osalvador@suse.de>,
	Michael Ellerman <mpe@ellerman.id.au>,
	Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Lorenzo Stoakes <ljs@kernel.org>,
	"Liam R . Howlett" <Liam.Howlett@oracle.com>,
	Vlastimil Babka <vbabka@kernel.org>,
	Mike Rapoport <rppt@kernel.org>,
	Suren Baghdasaryan <surenb@google.com>,
	Michal Hocko <mhocko@suse.com>,
	Nicholas Piggin <npiggin@gmail.com>,
	Christophe Leroy <chleroy@kernel.org>,
	Ackerley Tng <ackerleytng@google.com>,
	Frank van der Linden <fvdl@google.com>,
	aneesh.kumar@linux.ibm.com, joao.m.martins@oracle.com,
	linux-mm@kvack.org, linuxppc-dev@lists.ozlabs.org,
	linux-kernel@vger.kernel.org,
	Muchun Song <songmuchun@bytedance.com>
Subject: [PATCH v2 37/69] mm/sparse-vmemmap: Factor out shared vmemmap page allocation
Date: Wed, 13 May 2026 21:05:05 +0800	[thread overview]
Message-ID: <20260513130542.35604-38-songmuchun@bytedance.com> (raw)
In-Reply-To: <20260513130542.35604-1-songmuchun@bytedance.com>

HugeTLB and sparse-vmemmap each have their own helper to allocate the
shared tail page used by vmemmap optimization.

Factor that logic into a common vmemmap_shared_tail_page() helper in
sparse-vmemmap.c.  It allocates the page through
vmemmap_alloc_block_zero(), initializes the tail struct pages, and uses
cmpxchg() to install the per-zone shared page.

This removes duplicate allocation logic while still handling both the
early boot and runtime paths through the same helper.

Signed-off-by: Muchun Song <songmuchun@bytedance.com>
---
 include/linux/mm.h   |  1 +
 mm/hugetlb_vmemmap.c | 28 +-----------------
 mm/sparse-vmemmap.c  | 67 ++++++++++++++++++--------------------------
 3 files changed, 29 insertions(+), 67 deletions(-)

diff --git a/include/linux/mm.h b/include/linux/mm.h
index fef39be8acd2..5281f073230c 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -4866,6 +4866,7 @@ int vmemmap_populate(unsigned long start, unsigned long end, int node,
 void vmemmap_wrprotect_hvo(unsigned long start, unsigned long end, int node,
 			  unsigned long headsize);
 void vmemmap_populate_print_last(void);
+struct page *vmemmap_shared_tail_page(unsigned int order, struct zone *zone);
 #ifdef CONFIG_MEMORY_HOTPLUG
 void vmemmap_free(unsigned long start, unsigned long end,
 		struct vmem_altmap *altmap);
diff --git a/mm/hugetlb_vmemmap.c b/mm/hugetlb_vmemmap.c
index 66362e553870..d24143dd6051 100644
--- a/mm/hugetlb_vmemmap.c
+++ b/mm/hugetlb_vmemmap.c
@@ -499,32 +499,6 @@ static bool vmemmap_should_optimize_folio(const struct hstate *h, struct folio *
 	return vmemmap_should_optimize(h);
 }
 
-static struct page *vmemmap_get_tail(unsigned int order, struct zone *zone)
-{
-	const unsigned int idx = order - OPTIMIZABLE_FOLIO_MIN_ORDER;
-	struct page *tail, *p;
-	int node = zone_to_nid(zone);
-
-	tail = READ_ONCE(zone->vmemmap_tails[idx]);
-	if (likely(tail))
-		return tail;
-
-	tail = alloc_pages_node(node, GFP_KERNEL | __GFP_ZERO, 0);
-	if (!tail)
-		return NULL;
-
-	p = page_to_virt(tail);
-	for (int i = 0; i < PAGE_SIZE / sizeof(struct page); i++)
-		init_compound_tail(p + i, NULL, order, zone);
-
-	if (cmpxchg(&zone->vmemmap_tails[idx], NULL, tail)) {
-		__free_page(tail);
-		tail = READ_ONCE(zone->vmemmap_tails[idx]);
-	}
-
-	return tail;
-}
-
 static int __hugetlb_vmemmap_optimize_folio(const struct hstate *h,
 					    struct folio *folio,
 					    struct list_head *vmemmap_pages,
@@ -541,7 +515,7 @@ static int __hugetlb_vmemmap_optimize_folio(const struct hstate *h,
 		return ret;
 
 	nid = folio_nid(folio);
-	vmemmap_tail = vmemmap_get_tail(h->order, folio_zone(folio));
+	vmemmap_tail = vmemmap_shared_tail_page(h->order, folio_zone(folio));
 	if (!vmemmap_tail)
 		return -ENOMEM;
 
diff --git a/mm/sparse-vmemmap.c b/mm/sparse-vmemmap.c
index dde4486195ad..53a341fcde74 100644
--- a/mm/sparse-vmemmap.c
+++ b/mm/sparse-vmemmap.c
@@ -34,27 +34,13 @@
 
 #include "internal.h"
 
-/*
- * Allocate a block of memory to be used to back the virtual memory map
- * or to back the page tables that are used to create the mapping.
- * Uses the main allocators if they are available, else bootmem.
- */
-
-static void * __ref __earlyonly_bootmem_alloc(int node,
-				unsigned long size,
-				unsigned long align,
-				unsigned long goal)
-{
-	return memmap_alloc(size, align, goal, node, false);
-}
-
-void * __meminit vmemmap_alloc_block(unsigned long size, int node)
+void __ref *vmemmap_alloc_block(unsigned long size, int node)
 {
 	/* If the main allocator is up use that, fallback to bootmem. */
 	if (slab_is_available()) {
 		gfp_t gfp_mask = GFP_KERNEL|__GFP_RETRY_MAYFAIL|__GFP_NOWARN;
 		int order = get_order(size);
-		static bool warned __meminitdata;
+		static bool warned;
 		struct page *page;
 
 		page = alloc_pages_node(node, gfp_mask, order);
@@ -68,8 +54,7 @@ void * __meminit vmemmap_alloc_block(unsigned long size, int node)
 		}
 		return NULL;
 	} else
-		return __earlyonly_bootmem_alloc(node, size, size,
-				__pa(MAX_DMA_ADDRESS));
+		return memmap_alloc(size, size, __pa(MAX_DMA_ADDRESS), node, false);
 }
 
 static void * __meminit altmap_alloc_block_buf(unsigned long size,
@@ -138,8 +123,6 @@ void __meminit vmemmap_verify(pte_t *pte, int node,
 			start, end - 1);
 }
 
-static __meminit struct page *vmemmap_get_tail(unsigned int order, struct zone *zone);
-
 static pte_t * __meminit vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node,
 					      struct vmem_altmap *altmap,
 					      unsigned long ptpfn)
@@ -158,7 +141,7 @@ static pte_t * __meminit vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, in
 
 			if (WARN_ON_ONCE(!zone))
 				return NULL;
-			page = vmemmap_get_tail(section_order(ms), zone);
+			page = vmemmap_shared_tail_page(section_order(ms), zone);
 			if (!page)
 				return NULL;
 			ptpfn = page_to_pfn(page);
@@ -190,7 +173,7 @@ static pte_t * __meminit vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, in
 	return pte;
 }
 
-static void * __meminit vmemmap_alloc_block_zero(unsigned long size, int node)
+static void *vmemmap_alloc_block_zero(unsigned long size, int node)
 {
 	void *p = vmemmap_alloc_block(size, node);
 
@@ -329,32 +312,36 @@ void vmemmap_wrprotect_hvo(unsigned long addr, unsigned long end,
 	}
 }
 
-static __meminit struct page *vmemmap_get_tail(unsigned int order, struct zone *zone)
+struct page __ref *vmemmap_shared_tail_page(unsigned int order, struct zone *zone)
 {
-	struct page *p, *tail;
-	unsigned int idx;
-	int node = zone_to_nid(zone);
+	void *addr;
+	struct page *page;
+	const unsigned int idx = order - OPTIMIZABLE_FOLIO_MIN_ORDER;
 
-	if (WARN_ON_ONCE(order < OPTIMIZABLE_FOLIO_MIN_ORDER))
-		return NULL;
-	if (WARN_ON_ONCE(order > MAX_FOLIO_ORDER))
+	if (WARN_ON_ONCE(idx >= ARRAY_SIZE(zone->vmemmap_tails)))
 		return NULL;
 
-	idx = order - OPTIMIZABLE_FOLIO_MIN_ORDER;
-	tail = zone->vmemmap_tails[idx];
-	if (tail)
-		return tail;
+	page = READ_ONCE(zone->vmemmap_tails[idx]);
+	if (likely(page))
+		return page;
 
-	p = vmemmap_alloc_block_zero(PAGE_SIZE, node);
-	if (!p)
+	addr = vmemmap_alloc_block_zero(PAGE_SIZE, zone_to_nid(zone));
+	if (!addr)
 		return NULL;
-	for (int i = 0; i < PAGE_SIZE / sizeof(struct page); i++)
-		init_compound_tail(p + i, NULL, order, zone);
 
-	tail = virt_to_page(p);
-	zone->vmemmap_tails[idx] = tail;
+	for (int i = 0; i < PAGE_SIZE / sizeof(struct page); i++)
+		init_compound_tail((struct page *)addr + i, NULL, order, zone);
+
+	page = virt_to_page(addr);
+	if (cmpxchg(&zone->vmemmap_tails[idx], NULL, page) != NULL) {
+		if (slab_is_available())
+			__free_page(page);
+		else
+			memblock_free(page_to_virt(page), PAGE_SIZE);
+		page = READ_ONCE(zone->vmemmap_tails[idx]);
+	}
 
-	return tail;
+	return page;
 }
 
 void __weak __meminit vmemmap_set_pmd(pmd_t *pmd, void *p, int node,
-- 
2.54.0



  parent reply	other threads:[~2026-05-13 13:11 UTC|newest]

Thread overview: 72+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-05-13 13:04 [PATCH v2 00/69] mm: Generalize HVO for HugeTLB and device DAX Muchun Song
2026-05-13 13:04 ` [PATCH v2 01/69] mm/hugetlb: Fix boot panic with CONFIG_DEBUG_VM and HVO bootmem pages Muchun Song
2026-05-13 13:04 ` [PATCH v2 02/69] mm/hugetlb_vmemmap: Fix __hugetlb_vmemmap_optimize_folios() Muchun Song
2026-05-13 13:04 ` [PATCH v2 03/69] powerpc/mm: Fix wrong addr_pfn tracking in compound vmemmap population Muchun Song
2026-05-13 13:04 ` [PATCH v2 04/69] mm/hugetlb: Initialize gigantic bootmem hugepage struct pages earlier Muchun Song
2026-05-13 13:04 ` [PATCH v2 05/69] mm/mm_init: Simplify deferred_free_pages() migratetype init Muchun Song
2026-05-13 13:04 ` [PATCH v2 06/69] mm/sparse: Panic on memmap and usemap allocation failure Muchun Song
2026-05-13 13:04 ` [PATCH v2 07/69] mm/sparse: Move subsection_map_init() into sparse_init() Muchun Song
2026-05-13 13:04 ` [PATCH v2 08/69] mm/mm_init: Defer sparse_init() until after zone initialization Muchun Song
2026-05-13 13:04 ` [PATCH v2 09/69] mm/mm_init: Defer hugetlb reservation " Muchun Song
2026-05-13 13:04 ` [PATCH v2 10/69] mm/mm_init: Remove set_pageblock_order() call from sparse_init() Muchun Song
2026-05-13 13:04 ` [PATCH v2 11/69] mm/sparse: Move sparse_vmemmap_init_nid_late() into sparse_init_nid() Muchun Song
2026-05-13 13:04 ` [PATCH v2 12/69] mm/hugetlb_cma: Validate hugetlb CMA range by zone at reserve time Muchun Song
2026-05-13 13:04 ` [PATCH v2 13/69] mm/hugetlb: Refactor early boot gigantic hugepage allocation Muchun Song
2026-05-13 13:04 ` [PATCH v2 14/69] mm/hugetlb: Free cross-zone bootmem gigantic pages after allocation Muchun Song
2026-05-13 13:04 ` [PATCH v2 15/69] mm/hugetlb_vmemmap: Move bootmem HVO setup to early init Muchun Song
2026-05-13 13:04 ` [PATCH v2 16/69] mm/hugetlb: Remove obsolete bootmem cross-zone checks Muchun Song
2026-05-13 13:04 ` [PATCH v2 17/69] mm/sparse-vmemmap: Remove sparse_vmemmap_init_nid_late() Muchun Song
2026-05-13 13:04 ` [PATCH v2 18/69] mm/hugetlb: Remove unused bootmem cma field Muchun Song
2026-05-13 13:04 ` [PATCH v2 19/69] mm/mm_init: Make __init_page_from_nid() static Muchun Song
2026-05-13 13:04 ` [PATCH v2 20/69] mm/sparse-vmemmap: Drop VMEMMAP_POPULATE_PAGEREF Muchun Song
2026-05-13 13:04 ` [PATCH v2 21/69] mm: Rename vmemmap optimization macros around folio semantics Muchun Song
2026-05-13 13:04 ` [PATCH v2 22/69] mm/sparse: Drop power-of-2 size requirement for struct mem_section Muchun Song
2026-05-13 13:04 ` [PATCH v2 23/69] mm/sparse-vmemmap: track compound page order in " Muchun Song
2026-05-13 13:04 ` [PATCH v2 24/69] mm/mm_init: Skip initializing shared vmemmap tail pages Muchun Song
2026-05-13 13:04 ` [PATCH v2 25/69] mm/sparse-vmemmap: Initialize shared tail vmemmap pages on allocation Muchun Song
2026-05-13 13:04 ` [PATCH v2 26/69] mm/sparse-vmemmap: Support section-based vmemmap accounting Muchun Song
2026-05-13 13:04 ` [PATCH v2 27/69] mm/sparse-vmemmap: Support section-based vmemmap optimization Muchun Song
2026-05-13 13:04 ` [PATCH v2 28/69] mm/hugetlb: Use generic vmemmap optimization macros Muchun Song
2026-05-13 13:04 ` [PATCH v2 29/69] mm/sparse: Mark memblocks present earlier Muchun Song
2026-05-13 13:04 ` [PATCH v2 30/69] mm/hugetlb: Switch HugeTLB to section-based vmemmap optimization Muchun Song
2026-05-13 13:04 ` [PATCH v2 31/69] mm/sparse: Remove section_map_size() Muchun Song
2026-05-13 13:05 ` [PATCH v2 32/69] mm/mm_init: Factor out pfn_to_zone() as a shared helper Muchun Song
2026-05-13 13:05 ` [PATCH v2 33/69] mm/sparse: Remove SPARSEMEM_VMEMMAP_PREINIT Muchun Song
2026-05-13 13:05 ` [PATCH v2 34/69] mm/sparse: Inline usemap allocation into sparse_init_nid() Muchun Song
2026-05-13 13:05 ` [PATCH v2 35/69] mm/hugetlb: Remove HUGE_BOOTMEM_HVO Muchun Song
2026-05-13 13:05 ` [PATCH v2 36/69] mm/hugetlb: Remove HUGE_BOOTMEM_CMA Muchun Song
2026-05-13 13:05 ` Muchun Song [this message]
2026-05-13 13:05 ` [PATCH v2 38/69] mm/sparse-vmemmap: Introduce CONFIG_SPARSEMEM_VMEMMAP_OPTIMIZATION Muchun Song
2026-05-13 13:05 ` [PATCH v2 39/69] mm/sparse-vmemmap: Switch DAX to vmemmap_shared_tail_page() Muchun Song
2026-05-13 13:05 ` [PATCH v2 40/69] powerpc/mm: " Muchun Song
2026-05-13 13:05 ` [PATCH v2 41/69] mm/sparse-vmemmap: Drop the extra tail page from DAX reservation Muchun Song
2026-05-13 13:05 ` [PATCH v2 42/69] mm/sparse-vmemmap: Switch DAX to section-based vmemmap optimization Muchun Song
2026-05-13 13:05 ` [PATCH v2 43/69] mm/sparse-vmemmap: Unify DAX and HugeTLB population paths Muchun Song
2026-05-13 13:05 ` [PATCH v2 44/69] mm/sparse-vmemmap: Remove the unused ptpfn argument Muchun Song
2026-05-13 13:05 ` [PATCH v2 45/69] powerpc/mm: Make vmemmap_populate_compound_pages() static Muchun Song
2026-05-13 13:05 ` [PATCH v2 46/69] mm/sparse-vmemmap: Map shared vmemmap tail pages read-only Muchun Song
2026-05-13 13:20 ` [PATCH v2 47/69] powerpc/mm: " Muchun Song
2026-05-13 13:20   ` [PATCH v2 48/69] mm/sparse-vmemmap: Inline vmemmap_populate_address() into its caller Muchun Song
2026-05-13 13:20   ` [PATCH v2 49/69] mm/hugetlb_vmemmap: Remove vmemmap_wrprotect_hvo() Muchun Song
2026-05-13 13:20   ` [PATCH v2 50/69] mm/sparse: Simplify section_nr_vmemmap_pages() Muchun Song
2026-05-13 13:20   ` [PATCH v2 51/69] mm/sparse-vmemmap: Introduce vmemmap_nr_struct_pages() Muchun Song
2026-05-13 13:20   ` [PATCH v2 52/69] powerpc/mm: Drop powerpc vmemmap_can_optimize() Muchun Song
2026-05-13 13:20   ` [PATCH v2 53/69] mm/sparse-vmemmap: Drop vmemmap_can_optimize() Muchun Song
2026-05-13 13:20   ` [PATCH v2 54/69] mm/sparse-vmemmap: Drop @pgmap from vmemmap population APIs Muchun Song
2026-05-13 13:20   ` [PATCH v2 55/69] mm/sparse: Decouple section activation from ZONE_DEVICE Muchun Song
2026-05-13 13:20   ` [PATCH v2 56/69] mm: Redefine HVO as Hugepage Vmemmap Optimization Muchun Song
2026-05-13 13:20   ` [PATCH v2 57/69] mm/sparse-vmemmap: Consolidate HVO enable checks Muchun Song
2026-05-13 13:20   ` [PATCH v2 58/69] mm/hugetlb: Make HVO optimizable checks depend on generic logic Muchun Song
2026-05-13 13:20   ` [PATCH v2 59/69] mm/sparse-vmemmap: Localize init_compound_tail() Muchun Song
2026-05-13 13:20   ` [PATCH v2 60/69] mm/mm_init: Check zone consistency on optimized vmemmap sections Muchun Song
2026-05-13 13:20   ` [PATCH v2 61/69] mm/hugetlb: Drop boot-time HVO handling for gigantic folios Muchun Song
2026-05-13 13:20   ` [PATCH v2 62/69] mm/hugetlb: Simplify hugetlb_folio_init_vmemmap() Muchun Song
2026-05-13 13:20   ` [PATCH v2 63/69] mm/hugetlb: Initialize the full bootmem hugepage in hugetlb code Muchun Song
2026-05-13 13:20   ` [PATCH v2 64/69] mm/mm_init: Factor out compound page initialization Muchun Song
2026-05-13 13:20   ` [PATCH v2 65/69] mm/mm_init: Make __init_single_page() static Muchun Song
2026-05-13 13:20   ` [PATCH v2 66/69] mm/cma: Move CMA pageblock initialization into cma_activate_area() Muchun Song
2026-05-13 13:20   ` [PATCH v2 67/69] mm/cma: Move init_cma_pageblock() into cma.c Muchun Song
2026-05-13 13:20   ` [PATCH v2 68/69] mm/mm_init: Initialize pageblock migratetype in memmap init helpers Muchun Song
2026-05-13 13:20   ` [PATCH v2 69/69] Documentation/mm: Rewrite vmemmap_dedup.rst for unified HVO Muchun Song
2026-05-13 17:46 ` [PATCH v2 00/69] mm: Generalize HVO for HugeTLB and device DAX Andrew Morton
2026-05-13 18:26   ` Oscar Salvador

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260513130542.35604-38-songmuchun@bytedance.com \
    --to=songmuchun@bytedance.com \
    --cc=Liam.Howlett@oracle.com \
    --cc=ackerleytng@google.com \
    --cc=akpm@linux-foundation.org \
    --cc=aneesh.kumar@linux.ibm.com \
    --cc=chleroy@kernel.org \
    --cc=david@kernel.org \
    --cc=fvdl@google.com \
    --cc=joao.m.martins@oracle.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=linuxppc-dev@lists.ozlabs.org \
    --cc=ljs@kernel.org \
    --cc=maddy@linux.ibm.com \
    --cc=mhocko@suse.com \
    --cc=mpe@ellerman.id.au \
    --cc=muchun.song@linux.dev \
    --cc=npiggin@gmail.com \
    --cc=osalvador@suse.de \
    --cc=rppt@kernel.org \
    --cc=surenb@google.com \
    --cc=vbabka@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox