From: Larry Woodman <lwoodman@redhat.com>
Subject: [RHEL5 Patch] Feature request - RHEL5.1 shared page table for hugetlb  page
Date: Thu, 03 May 2007 15:50:28 -0400
Bugzilla: 222753
Message-Id: <463A3D04.5090406@redhat.com>
Changelog: [mm] shared page table for hugetlb  page


The attached upstream patch shares page table pages for hugetlbfs.

The shared page table is useful in the situation where a large number of
independent processes map the same large shared memory segments.  The 
amount of memory saved from processes sharing page table pages is quite 
significant.  In addition, shared page tables with hugetlbfs allows faster 
TLB refill and smaller cache pollution upon TLB miss.

With page table sharing, the pte entries are shared among many processes
so the cache consumption is much smaller so the application gets a much 
higher cache hit ratio.  One other effect with the hardware page walker 
is that the cache hit ratio on the pte will be higher which helps reduce 
the tlb miss latency.  These two effects contribute to higher application 
performance.

Addresses BZ 222753




--- linux-2.6.18.noarch/arch/i386/mm/hugetlbpage.c.orig
+++ linux-2.6.18.noarch/arch/i386/mm/hugetlbpage.c
@@ -17,6 +17,113 @@
 #include <asm/tlb.h>
 #include <asm/tlbflush.h>
 
+static unsigned long page_table_shareable(struct vm_area_struct *svma,
+				struct vm_area_struct *vma,
+				unsigned long addr, pgoff_t idx)
+{
+	unsigned long saddr = ((idx - svma->vm_pgoff) << PAGE_SHIFT) +
+				svma->vm_start;
+	unsigned long sbase = saddr & PUD_MASK;
+	unsigned long s_end = sbase + PUD_SIZE;
+
+	/*
+	 * match the virtual addresses, permission and the alignment of the
+	 * page table page.
+	 */
+	if (pmd_index(addr) != pmd_index(saddr) ||
+	    vma->vm_flags != svma->vm_flags ||
+	    sbase < svma->vm_start || svma->vm_end < s_end)
+		return 0;
+
+	return saddr;
+}
+
+static int vma_shareable(struct vm_area_struct *vma, unsigned long addr)
+{
+	unsigned long base = addr & PUD_MASK;
+	unsigned long end = base + PUD_SIZE;
+
+	/*
+	 * check on proper vm_flags and page table alignment
+	 */
+	if (vma->vm_flags & VM_MAYSHARE &&
+	    vma->vm_start <= base && end <= vma->vm_end)
+		return 1;
+	return 0;
+}
+
+/*
+ * search for a shareable pmd page for hugetlb.
+ */
+static void huge_pmd_share(struct mm_struct *mm, unsigned long addr, pud_t *pud)
+{
+	struct vm_area_struct *vma = find_vma(mm, addr);
+	struct address_space *mapping = vma->vm_file->f_mapping;
+	pgoff_t idx = ((addr - vma->vm_start) >> PAGE_SHIFT) +
+			vma->vm_pgoff;
+	struct prio_tree_iter iter;
+	struct vm_area_struct *svma;
+	unsigned long saddr;
+	pte_t *spte = NULL;
+
+	if (!vma_shareable(vma, addr))
+		return;
+
+	spin_lock(&mapping->i_mmap_lock);
+	vma_prio_tree_foreach(svma, &iter, &mapping->i_mmap, idx, idx) {
+		if (svma == vma)
+			continue;
+
+		saddr = page_table_shareable(svma, vma, addr, idx);
+		if (saddr) {
+			spte = huge_pte_offset(svma->vm_mm, saddr);
+			if (spte) {
+				get_page(virt_to_page(spte));
+				break;
+			}
+		}
+	}
+
+	if (!spte)
+		goto out;
+
+	spin_lock(&mm->page_table_lock);
+	if (pud_none(*pud))
+		pud_populate(mm, pud, (unsigned long) spte & PAGE_MASK);
+	else
+		put_page(virt_to_page(spte));
+	spin_unlock(&mm->page_table_lock);
+out:
+	spin_unlock(&mapping->i_mmap_lock);
+}
+
+/*
+ * unmap huge page backed by shared pte.
+ *
+ * Hugetlb pte page is ref counted at the time of mapping.  If pte is shared
+ * indicated by page_count > 1, unmap is achieved by clearing pud and
+ * decrementing the ref count. If count == 1, the pte page is not shared.
+ *
+ * called with vma->vm_mm->page_table_lock held.
+ *
+ * returns: 1 successfully unmapped a shared pte page
+ *	    0 the underlying pte page is not shared, or it is the last user
+ */
+int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep)
+{
+	pgd_t *pgd = pgd_offset(mm, *addr);
+	pud_t *pud = pud_offset(pgd, *addr);
+
+	BUG_ON(page_count(virt_to_page(ptep)) == 0);
+	if (page_count(virt_to_page(ptep)) == 1)
+		return 0;
+
+	pud_clear(pud);
+	put_page(virt_to_page(ptep));
+	*addr = ALIGN(*addr, HPAGE_SIZE * PTRS_PER_PTE) - HPAGE_SIZE;
+	return 1;
+}
+
 pte_t *huge_pte_alloc(struct mm_struct *mm, unsigned long addr)
 {
 	pgd_t *pgd;
@@ -25,8 +132,11 @@ pte_t *huge_pte_alloc(struct mm_struct *
 
 	pgd = pgd_offset(mm, addr);
 	pud = pud_alloc(mm, pgd, addr);
-	if (pud)
+	if (pud) {
+		if (pud_none(*pud))
+			huge_pmd_share(mm, addr, pud);
 		pte = (pte_t *) pmd_alloc(mm, pud, addr);
+	}
 	BUG_ON(pte && !pte_none(*pte) && !pte_huge(*pte));
 
 	return pte;
--- linux-2.6.18.noarch/arch/ia64/mm/hugetlbpage.c.orig
+++ linux-2.6.18.noarch/arch/ia64/mm/hugetlbpage.c
@@ -64,6 +64,11 @@ huge_pte_offset (struct mm_struct *mm, u
 	return pte;
 }
 
+int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep)
+{
+	return 0;
+}
+
 #define mk_pte_huge(entry) { pte_val(entry) |= _PAGE_P; }
 
 /*
--- linux-2.6.18.noarch/arch/powerpc/mm/hugetlbpage.c.orig
+++ linux-2.6.18.noarch/arch/powerpc/mm/hugetlbpage.c
@@ -146,6 +146,11 @@ pte_t *huge_pte_alloc(struct mm_struct *
 	return hugepte_offset(hpdp, addr);
 }
 
+int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep)
+{
+	return 0;
+}
+
 static void free_hugepte_range(struct mmu_gather *tlb, hugepd_t *hpdp)
 {
 	pte_t *hugepte = hugepd_page(*hpdp);
--- linux-2.6.18.noarch/arch/sh64/mm/hugetlbpage.c.orig
+++ linux-2.6.18.noarch/arch/sh64/mm/hugetlbpage.c
@@ -53,6 +53,11 @@ pte_t *huge_pte_offset(struct mm_struct 
 	return pte;
 }
 
+int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep)
+{
+	return 0;
+}
+
 void set_huge_pte_at(struct mm_struct *mm, unsigned long addr,
 		     pte_t *ptep, pte_t entry)
 {
--- linux-2.6.18.noarch/arch/sh/mm/hugetlbpage.c.orig
+++ linux-2.6.18.noarch/arch/sh/mm/hugetlbpage.c
@@ -83,6 +83,11 @@ pte_t huge_ptep_get_and_clear(struct mm_
 	return entry;
 }
 
+int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep)
+{
+	return 0;
+}
+
 struct page *follow_huge_addr(struct mm_struct *mm,
 			      unsigned long address, int write)
 {
--- linux-2.6.18.noarch/arch/sparc64/mm/hugetlbpage.c.orig
+++ linux-2.6.18.noarch/arch/sparc64/mm/hugetlbpage.c
@@ -235,6 +235,11 @@ pte_t *huge_pte_offset(struct mm_struct 
 	return pte;
 }
 
+int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep)
+{
+	return 0;
+}
+
 void set_huge_pte_at(struct mm_struct *mm, unsigned long addr,
 		     pte_t *ptep, pte_t entry)
 {
--- linux-2.6.18.noarch/fs/hugetlbfs/inode.c.orig
+++ linux-2.6.18.noarch/fs/hugetlbfs/inode.c
@@ -293,7 +293,7 @@ hugetlb_vmtruncate_list(struct prio_tree
 		if (h_vm_pgoff >= h_pgoff)
 			v_offset = 0;
 
-		unmap_hugepage_range(vma,
+		__unmap_hugepage_range(vma,
 				vma->vm_start + v_offset, vma->vm_end);
 	}
 }
--- linux-2.6.18.noarch/include/linux/hugetlb.h.orig
+++ linux-2.6.18.noarch/include/linux/hugetlb.h
@@ -17,6 +17,7 @@ int hugetlb_sysctl_handler(struct ctl_ta
 int copy_hugetlb_page_range(struct mm_struct *, struct mm_struct *, struct vm_area_struct *);
 int follow_hugetlb_page(struct mm_struct *, struct vm_area_struct *, struct page **, struct vm_area_struct **, unsigned long *, int *, int);
 void unmap_hugepage_range(struct vm_area_struct *, unsigned long, unsigned long);
+void __unmap_hugepage_range(struct vm_area_struct *, unsigned long, unsigned long);
 int hugetlb_prefault(struct address_space *, struct vm_area_struct *);
 int hugetlb_report_meminfo(char *);
 int hugetlb_report_node_meminfo(int, char *);
@@ -34,6 +35,7 @@ extern int sysctl_hugetlb_shm_group;
 
 pte_t *huge_pte_alloc(struct mm_struct *mm, unsigned long addr);
 pte_t *huge_pte_offset(struct mm_struct *mm, unsigned long addr);
+int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep);
 struct page *follow_huge_addr(struct mm_struct *mm, unsigned long address,
 			      int write);
 struct page *follow_huge_pmd(struct mm_struct *mm, unsigned long address,
--- linux-2.6.18.noarch/mm/hugetlb.c.orig
+++ linux-2.6.18.noarch/mm/hugetlb.c
@@ -344,7 +344,6 @@ int copy_hugetlb_page_range(struct mm_st
 			entry = *src_pte;
 			ptepage = pte_page(entry);
 			get_page(ptepage);
-			add_mm_counter(dst, file_rss, HPAGE_SIZE / PAGE_SIZE);
 			set_huge_pte_at(dst, addr, dst_pte, entry);
 		}
 		spin_unlock(&src->page_table_lock);
@@ -356,8 +355,8 @@ nomem:
 	return -ENOMEM;
 }
 
-void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
-			  unsigned long end)
+void __unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
+			    unsigned long end)
 {
 	struct mm_struct *mm = vma->vm_mm;
 	unsigned long address;
@@ -370,28 +369,43 @@ void unmap_hugepage_range(struct vm_area
 	BUG_ON(end & ~HPAGE_MASK);
 
 	spin_lock(&mm->page_table_lock);
-
-	/* Update high watermark before we lower rss */
-	update_hiwater_rss(mm);
-
 	for (address = start; address < end; address += HPAGE_SIZE) {
 		ptep = huge_pte_offset(mm, address);
 		if (!ptep)
 			continue;
 
+		if (huge_pmd_unshare(mm, &address, ptep))
+			continue;
+
 		pte = huge_ptep_get_and_clear(mm, address, ptep);
 		if (pte_none(pte))
 			continue;
 
 		page = pte_page(pte);
 		put_page(page);
-		add_mm_counter(mm, file_rss, (int) -(HPAGE_SIZE / PAGE_SIZE));
 	}
-
 	spin_unlock(&mm->page_table_lock);
 	flush_tlb_range(vma, start, end);
 }
 
+void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
+			  unsigned long end)
+{
+	/*
+	 * It is undesirable to test vma->vm_file as it should be non-null
+	 * for valid hugetlb area. However, vm_file will be NULL in the error
+	 * cleanup path of do_mmap_pgoff. When hugetlbfs ->mmap method fails,
+	 * do_mmap_pgoff() nullifies vma->vm_file before calling this function
+	 * to clean up. Since no pte has actually been setup, it is safe to
+	 * do nothing in this case.
+	 */
+	if (vma->vm_file) {
+		spin_lock(&vma->vm_file->f_mapping->i_mmap_lock);
+		__unmap_hugepage_range(vma, start, end);
+		spin_unlock(&vma->vm_file->f_mapping->i_mmap_lock);
+	}
+}
+
 static int hugetlb_cow(struct mm_struct *mm, struct vm_area_struct *vma,
 			unsigned long address, pte_t *ptep, pte_t pte)
 {
@@ -491,7 +505,6 @@ retry:
 	if (!pte_none(*ptep))
 		goto backout;
 
-	add_mm_counter(mm, file_rss, HPAGE_SIZE / PAGE_SIZE);
 	new_pte = make_huge_pte(vma, page, ((vma->vm_flags & VM_WRITE)
 				&& (vma->vm_flags & VM_SHARED)));
 	set_huge_pte_at(mm, address, ptep, new_pte);
@@ -629,11 +642,14 @@ void hugetlb_change_protection(struct vm
 	BUG_ON(address >= end);
 	flush_cache_range(vma, address, end);
 
+	spin_lock(&vma->vm_file->f_mapping->i_mmap_lock);
 	spin_lock(&mm->page_table_lock);
 	for (; address < end; address += HPAGE_SIZE) {
 		ptep = huge_pte_offset(mm, address);
 		if (!ptep)
 			continue;
+		if (huge_pmd_unshare(mm, &address, ptep))
+			continue;
 		if (!pte_none(*ptep)) {
 			pte = huge_ptep_get_and_clear(mm, address, ptep);
 			pte = pte_mkhuge(pte_modify(pte, newprot));
@@ -642,6 +658,7 @@ void hugetlb_change_protection(struct vm
 		}
 	}
 	spin_unlock(&mm->page_table_lock);
+	spin_unlock(&vma->vm_file->f_mapping->i_mmap_lock);
 
 	flush_tlb_range(vma, start, end);
 }