aboutsummaryrefslogtreecommitdiff
path: root/mm/huge_memory.c
diff options
context:
space:
mode:
Diffstat (limited to 'mm/huge_memory.c')
-rw-r--r--mm/huge_memory.c674
1 files changed, 411 insertions, 263 deletions
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index 7489884682d..33514d88fef 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -5,6 +5,8 @@
* the COPYING file in the top-level directory.
*/
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
#include <linux/mm.h>
#include <linux/sched.h>
#include <linux/highmem.h>
@@ -27,11 +29,12 @@
#include "internal.h"
/*
- * By default transparent hugepage support is enabled for all mappings
- * and khugepaged scans all mappings. Defrag is only invoked by
- * khugepaged hugepage allocations and by page faults inside
- * MADV_HUGEPAGE regions to avoid the risk of slowing down short lived
- * allocations.
+ * By default transparent hugepage support is disabled in order that avoid
+ * to risk increase the memory footprint of applications without a guaranteed
+ * benefit. When transparent hugepage support is enabled, is for all mappings,
+ * and khugepaged scans all mappings.
+ * Defrag is invoked by khugepaged hugepage allocations and by page faults
+ * for all hugepage allocations.
*/
unsigned long transparent_hugepage_flags __read_mostly =
#ifdef CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS
@@ -129,8 +132,14 @@ static int set_recommended_min_free_kbytes(void)
(unsigned long) nr_free_buffer_pages() / 20);
recommended_min <<= (PAGE_SHIFT-10);
- if (recommended_min > min_free_kbytes)
+ if (recommended_min > min_free_kbytes) {
+ if (user_min_free_kbytes >= 0)
+ pr_info("raising min_free_kbytes from %d to %lu "
+ "to help transparent hugepage allocations\n",
+ min_free_kbytes, recommended_min);
+
min_free_kbytes = recommended_min;
+ }
setup_per_zone_wmarks();
return 0;
}
@@ -144,8 +153,7 @@ static int start_khugepaged(void)
khugepaged_thread = kthread_run(khugepaged, NULL,
"khugepaged");
if (unlikely(IS_ERR(khugepaged_thread))) {
- printk(KERN_ERR
- "khugepaged: kthread_run(khugepaged) failed\n");
+ pr_err("khugepaged: kthread_run(khugepaged) failed\n");
err = PTR_ERR(khugepaged_thread);
khugepaged_thread = NULL;
}
@@ -577,19 +585,19 @@ static int __init hugepage_init_sysfs(struct kobject **hugepage_kobj)
*hugepage_kobj = kobject_create_and_add("transparent_hugepage", mm_kobj);
if (unlikely(!*hugepage_kobj)) {
- printk(KERN_ERR "hugepage: failed to create transparent hugepage kobject\n");
+ pr_err("failed to create transparent hugepage kobject\n");
return -ENOMEM;
}
err = sysfs_create_group(*hugepage_kobj, &hugepage_attr_group);
if (err) {
- printk(KERN_ERR "hugepage: failed to register transparent hugepage group\n");
+ pr_err("failed to register transparent hugepage group\n");
goto delete_obj;
}
err = sysfs_create_group(*hugepage_kobj, &khugepaged_attr_group);
if (err) {
- printk(KERN_ERR "hugepage: failed to register transparent hugepage group\n");
+ pr_err("failed to register transparent hugepage group\n");
goto remove_hp_group;
}
@@ -654,7 +662,7 @@ out:
hugepage_exit_sysfs(hugepage_kobj);
return err;
}
-module_init(hugepage_init)
+subsys_initcall(hugepage_init);
static int __init setup_transparent_hugepage(char *str)
{
@@ -682,8 +690,7 @@ static int __init setup_transparent_hugepage(char *str)
}
out:
if (!ret)
- printk(KERN_WARNING
- "transparent_hugepage= cannot parse, ignored\n");
+ pr_warn("transparent_hugepage= cannot parse, ignored\n");
return ret;
}
__setup("transparent_hugepage=", setup_transparent_hugepage);
@@ -709,8 +716,9 @@ static int __do_huge_pmd_anonymous_page(struct mm_struct *mm,
struct page *page)
{
pgtable_t pgtable;
+ spinlock_t *ptl;
- VM_BUG_ON(!PageCompound(page));
+ VM_BUG_ON_PAGE(!PageCompound(page), page);
pgtable = pte_alloc_one(mm, haddr);
if (unlikely(!pgtable))
return VM_FAULT_OOM;
@@ -723,9 +731,9 @@ static int __do_huge_pmd_anonymous_page(struct mm_struct *mm,
*/
__SetPageUptodate(page);
- spin_lock(&mm->page_table_lock);
+ ptl = pmd_lock(mm, pmd);
if (unlikely(!pmd_none(*pmd))) {
- spin_unlock(&mm->page_table_lock);
+ spin_unlock(ptl);
mem_cgroup_uncharge_page(page);
put_page(page);
pte_free(mm, pgtable);
@@ -737,8 +745,8 @@ static int __do_huge_pmd_anonymous_page(struct mm_struct *mm,
pgtable_trans_huge_deposit(mm, pmd, pgtable);
set_pmd_at(mm, haddr, pmd, entry);
add_mm_counter(mm, MM_ANONPAGES, HPAGE_PMD_NR);
- mm->nr_ptes++;
- spin_unlock(&mm->page_table_lock);
+ atomic_long_inc(&mm->nr_ptes);
+ spin_unlock(ptl);
}
return 0;
@@ -758,14 +766,7 @@ static inline struct page *alloc_hugepage_vma(int defrag,
HPAGE_PMD_ORDER, vma, haddr, nd);
}
-#ifndef CONFIG_NUMA
-static inline struct page *alloc_hugepage(int defrag)
-{
- return alloc_pages(alloc_hugepage_gfpmask(defrag, 0),
- HPAGE_PMD_ORDER);
-}
-#endif
-
+/* Caller must hold page table lock. */
static bool set_huge_zero_page(pgtable_t pgtable, struct mm_struct *mm,
struct vm_area_struct *vma, unsigned long haddr, pmd_t *pmd,
struct page *zero_page)
@@ -778,7 +779,7 @@ static bool set_huge_zero_page(pgtable_t pgtable, struct mm_struct *mm,
entry = pmd_mkhuge(entry);
pgtable_trans_huge_deposit(mm, pmd, pgtable);
set_pmd_at(mm, haddr, pmd, entry);
- mm->nr_ptes++;
+ atomic_long_inc(&mm->nr_ptes);
return true;
}
@@ -797,6 +798,7 @@ int do_huge_pmd_anonymous_page(struct mm_struct *mm, struct vm_area_struct *vma,
return VM_FAULT_OOM;
if (!(flags & FAULT_FLAG_WRITE) &&
transparent_hugepage_use_zero_page()) {
+ spinlock_t *ptl;
pgtable_t pgtable;
struct page *zero_page;
bool set;
@@ -809,10 +811,10 @@ int do_huge_pmd_anonymous_page(struct mm_struct *mm, struct vm_area_struct *vma,
count_vm_event(THP_FAULT_FALLBACK);
return VM_FAULT_FALLBACK;
}
- spin_lock(&mm->page_table_lock);
+ ptl = pmd_lock(mm, pmd);
set = set_huge_zero_page(pgtable, mm, vma, haddr, pmd,
zero_page);
- spin_unlock(&mm->page_table_lock);
+ spin_unlock(ptl);
if (!set) {
pte_free(mm, pgtable);
put_huge_zero_page();
@@ -825,7 +827,7 @@ int do_huge_pmd_anonymous_page(struct mm_struct *mm, struct vm_area_struct *vma,
count_vm_event(THP_FAULT_FALLBACK);
return VM_FAULT_FALLBACK;
}
- if (unlikely(mem_cgroup_newpage_charge(page, mm, GFP_KERNEL))) {
+ if (unlikely(mem_cgroup_charge_anon(page, mm, GFP_KERNEL))) {
put_page(page);
count_vm_event(THP_FAULT_FALLBACK);
return VM_FAULT_FALLBACK;
@@ -845,6 +847,7 @@ int copy_huge_pmd(struct mm_struct *dst_mm, struct mm_struct *src_mm,
pmd_t *dst_pmd, pmd_t *src_pmd, unsigned long addr,
struct vm_area_struct *vma)
{
+ spinlock_t *dst_ptl, *src_ptl;
struct page *src_page;
pmd_t pmd;
pgtable_t pgtable;
@@ -855,8 +858,9 @@ int copy_huge_pmd(struct mm_struct *dst_mm, struct mm_struct *src_mm,
if (unlikely(!pgtable))
goto out;
- spin_lock(&dst_mm->page_table_lock);
- spin_lock_nested(&src_mm->page_table_lock, SINGLE_DEPTH_NESTING);
+ dst_ptl = pmd_lock(dst_mm, dst_pmd);
+ src_ptl = pmd_lockptr(src_mm, src_pmd);
+ spin_lock_nested(src_ptl, SINGLE_DEPTH_NESTING);
ret = -EAGAIN;
pmd = *src_pmd;
@@ -865,7 +869,7 @@ int copy_huge_pmd(struct mm_struct *dst_mm, struct mm_struct *src_mm,
goto out_unlock;
}
/*
- * mm->page_table_lock is enough to be sure that huge zero pmd is not
+ * When page table lock is held, the huge zero pmd should not be
* under splitting since we don't split the page itself, only pmd to
* a page table.
*/
@@ -884,17 +888,18 @@ int copy_huge_pmd(struct mm_struct *dst_mm, struct mm_struct *src_mm,
ret = 0;
goto out_unlock;
}
+
if (unlikely(pmd_trans_splitting(pmd))) {
/* split huge page running from under us */
- spin_unlock(&src_mm->page_table_lock);
- spin_unlock(&dst_mm->page_table_lock);
+ spin_unlock(src_ptl);
+ spin_unlock(dst_ptl);
pte_free(dst_mm, pgtable);
wait_split_huge_page(vma->anon_vma, src_pmd); /* src_vma */
goto out;
}
src_page = pmd_page(pmd);
- VM_BUG_ON(!PageHead(src_page));
+ VM_BUG_ON_PAGE(!PageHead(src_page), src_page);
get_page(src_page);
page_dup_rmap(src_page);
add_mm_counter(dst_mm, MM_ANONPAGES, HPAGE_PMD_NR);
@@ -903,12 +908,12 @@ int copy_huge_pmd(struct mm_struct *dst_mm, struct mm_struct *src_mm,
pmd = pmd_mkold(pmd_wrprotect(pmd));
pgtable_trans_huge_deposit(dst_mm, dst_pmd, pgtable);
set_pmd_at(dst_mm, addr, dst_pmd, pmd);
- dst_mm->nr_ptes++;
+ atomic_long_inc(&dst_mm->nr_ptes);
ret = 0;
out_unlock:
- spin_unlock(&src_mm->page_table_lock);
- spin_unlock(&dst_mm->page_table_lock);
+ spin_unlock(src_ptl);
+ spin_unlock(dst_ptl);
out:
return ret;
}
@@ -919,10 +924,11 @@ void huge_pmd_set_accessed(struct mm_struct *mm,
pmd_t *pmd, pmd_t orig_pmd,
int dirty)
{
+ spinlock_t *ptl;
pmd_t entry;
unsigned long haddr;
- spin_lock(&mm->page_table_lock);
+ ptl = pmd_lock(mm, pmd);
if (unlikely(!pmd_same(*pmd, orig_pmd)))
goto unlock;
@@ -932,81 +938,38 @@ void huge_pmd_set_accessed(struct mm_struct *mm,
update_mmu_cache_pmd(vma, address, pmd);
unlock:
- spin_unlock(&mm->page_table_lock);
+ spin_unlock(ptl);
}
-static int do_huge_pmd_wp_zero_page_fallback(struct mm_struct *mm,
- struct vm_area_struct *vma, unsigned long address,
- pmd_t *pmd, pmd_t orig_pmd, unsigned long haddr)
+/*
+ * Save CONFIG_DEBUG_PAGEALLOC from faulting falsely on tail pages
+ * during copy_user_huge_page()'s copy_page_rep(): in the case when
+ * the source page gets split and a tail freed before copy completes.
+ * Called under pmd_lock of checked pmd, so safe from splitting itself.
+ */
+static void get_user_huge_page(struct page *page)
{
- pgtable_t pgtable;
- pmd_t _pmd;
- struct page *page;
- int i, ret = 0;
- unsigned long mmun_start; /* For mmu_notifiers */
- unsigned long mmun_end; /* For mmu_notifiers */
+ if (IS_ENABLED(CONFIG_DEBUG_PAGEALLOC)) {
+ struct page *endpage = page + HPAGE_PMD_NR;
- page = alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, address);
- if (!page) {
- ret |= VM_FAULT_OOM;
- goto out;
- }
-
- if (mem_cgroup_newpage_charge(page, mm, GFP_KERNEL)) {
- put_page(page);
- ret |= VM_FAULT_OOM;
- goto out;
+ atomic_add(HPAGE_PMD_NR, &page->_count);
+ while (++page < endpage)
+ get_huge_page_tail(page);
+ } else {
+ get_page(page);
}
+}
- clear_user_highpage(page, address);
- __SetPageUptodate(page);
-
- mmun_start = haddr;
- mmun_end = haddr + HPAGE_PMD_SIZE;
- mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end);
-
- spin_lock(&mm->page_table_lock);
- if (unlikely(!pmd_same(*pmd, orig_pmd)))
- goto out_free_page;
-
- pmdp_clear_flush(vma, haddr, pmd);
- /* leave pmd empty until pte is filled */
-
- pgtable = pgtable_trans_huge_withdraw(mm, pmd);
- pmd_populate(mm, &_pmd, pgtable);
+static void put_user_huge_page(struct page *page)
+{
+ if (IS_ENABLED(CONFIG_DEBUG_PAGEALLOC)) {
+ struct page *endpage = page + HPAGE_PMD_NR;
- for (i = 0; i < HPAGE_PMD_NR; i++, haddr += PAGE_SIZE) {
- pte_t *pte, entry;
- if (haddr == (address & PAGE_MASK)) {
- entry = mk_pte(page, vma->vm_page_prot);
- entry = maybe_mkwrite(pte_mkdirty(entry), vma);
- page_add_new_anon_rmap(page, vma, haddr);
- } else {
- entry = pfn_pte(my_zero_pfn(haddr), vma->vm_page_prot);
- entry = pte_mkspecial(entry);
- }
- pte = pte_offset_map(&_pmd, haddr);
- VM_BUG_ON(!pte_none(*pte));
- set_pte_at(mm, haddr, pte, entry);
- pte_unmap(pte);
+ while (page < endpage)
+ put_page(page++);
+ } else {
+ put_page(page);
}
- smp_wmb(); /* make pte visible before pmd */
- pmd_populate(mm, pmd, pgtable);
- spin_unlock(&mm->page_table_lock);
- put_huge_zero_page();
- inc_mm_counter(mm, MM_ANONPAGES);
-
- mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
-
- ret |= VM_FAULT_WRITE;
-out:
- return ret;
-out_free_page:
- spin_unlock(&mm->page_table_lock);
- mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
- mem_cgroup_uncharge_page(page);
- put_page(page);
- goto out;
}
static int do_huge_pmd_wp_page_fallback(struct mm_struct *mm,
@@ -1016,6 +979,7 @@ static int do_huge_pmd_wp_page_fallback(struct mm_struct *mm,
struct page *page,
unsigned long haddr)
{
+ spinlock_t *ptl;
pgtable_t pgtable;
pmd_t _pmd;
int ret = 0, i;
@@ -1035,7 +999,7 @@ static int do_huge_pmd_wp_page_fallback(struct mm_struct *mm,
__GFP_OTHER_NODE,
vma, address, page_to_nid(page));
if (unlikely(!pages[i] ||
- mem_cgroup_newpage_charge(pages[i], mm,
+ mem_cgroup_charge_anon(pages[i], mm,
GFP_KERNEL))) {
if (pages[i])
put_page(pages[i]);
@@ -1062,10 +1026,10 @@ static int do_huge_pmd_wp_page_fallback(struct mm_struct *mm,
mmun_end = haddr + HPAGE_PMD_SIZE;
mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end);
- spin_lock(&mm->page_table_lock);
+ ptl = pmd_lock(mm, pmd);
if (unlikely(!pmd_same(*pmd, orig_pmd)))
goto out_free_pages;
- VM_BUG_ON(!PageHead(page));
+ VM_BUG_ON_PAGE(!PageHead(page), page);
pmdp_clear_flush(vma, haddr, pmd);
/* leave pmd empty until pte is filled */
@@ -1088,7 +1052,7 @@ static int do_huge_pmd_wp_page_fallback(struct mm_struct *mm,
smp_wmb(); /* make pte visible before pmd */
pmd_populate(mm, pmd, pgtable);
page_remove_rmap(page);
- spin_unlock(&mm->page_table_lock);
+ spin_unlock(ptl);
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
@@ -1099,7 +1063,7 @@ out:
return ret;
out_free_pages:
- spin_unlock(&mm->page_table_lock);
+ spin_unlock(ptl);
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
mem_cgroup_uncharge_start();
for (i = 0; i < HPAGE_PMD_NR; i++) {
@@ -1114,22 +1078,24 @@ out_free_pages:
int do_huge_pmd_wp_page(struct mm_struct *mm, struct vm_area_struct *vma,
unsigned long address, pmd_t *pmd, pmd_t orig_pmd)
{
+ spinlock_t *ptl;
int ret = 0;
struct page *page = NULL, *new_page;
unsigned long haddr;
unsigned long mmun_start; /* For mmu_notifiers */
unsigned long mmun_end; /* For mmu_notifiers */
+ ptl = pmd_lockptr(mm, pmd);
VM_BUG_ON(!vma->anon_vma);
haddr = address & HPAGE_PMD_MASK;
if (is_huge_zero_pmd(orig_pmd))
goto alloc;
- spin_lock(&mm->page_table_lock);
+ spin_lock(ptl);
if (unlikely(!pmd_same(*pmd, orig_pmd)))
goto out_unlock;
page = pmd_page(orig_pmd);
- VM_BUG_ON(!PageCompound(page) || !PageHead(page));
+ VM_BUG_ON_PAGE(!PageCompound(page) || !PageHead(page), page);
if (page_mapcount(page) == 1) {
pmd_t entry;
entry = pmd_mkyoung(orig_pmd);
@@ -1139,8 +1105,8 @@ int do_huge_pmd_wp_page(struct mm_struct *mm, struct vm_area_struct *vma,
ret |= VM_FAULT_WRITE;
goto out_unlock;
}
- get_page(page);
- spin_unlock(&mm->page_table_lock);
+ get_user_huge_page(page);
+ spin_unlock(ptl);
alloc:
if (transparent_hugepage_enabled(vma) &&
!transparent_hugepage_debug_cow())
@@ -1150,34 +1116,37 @@ alloc:
new_page = NULL;
if (unlikely(!new_page)) {
- if (is_huge_zero_pmd(orig_pmd)) {
- ret = do_huge_pmd_wp_zero_page_fallback(mm, vma,
- address, pmd, orig_pmd, haddr);
+ if (!page) {
+ split_huge_page_pmd(vma, address, pmd);
+ ret |= VM_FAULT_FALLBACK;
} else {
ret = do_huge_pmd_wp_page_fallback(mm, vma, address,
pmd, orig_pmd, page, haddr);
- if (ret & VM_FAULT_OOM)
+ if (ret & VM_FAULT_OOM) {
split_huge_page(page);
- put_page(page);
+ ret |= VM_FAULT_FALLBACK;
+ }
+ put_user_huge_page(page);
}
count_vm_event(THP_FAULT_FALLBACK);
goto out;
}
- if (unlikely(mem_cgroup_newpage_charge(new_page, mm, GFP_KERNEL))) {
+ if (unlikely(mem_cgroup_charge_anon(new_page, mm, GFP_KERNEL))) {
put_page(new_page);
if (page) {
split_huge_page(page);
- put_page(page);
- }
+ put_user_huge_page(page);
+ } else
+ split_huge_page_pmd(vma, address, pmd);
+ ret |= VM_FAULT_FALLBACK;
count_vm_event(THP_FAULT_FALLBACK);
- ret |= VM_FAULT_OOM;
goto out;
}
count_vm_event(THP_FAULT_ALLOC);
- if (is_huge_zero_pmd(orig_pmd))
+ if (!page)
clear_huge_page(new_page, haddr, HPAGE_PMD_NR);
else
copy_user_huge_page(new_page, page, haddr, vma, HPAGE_PMD_NR);
@@ -1187,11 +1156,11 @@ alloc:
mmun_end = haddr + HPAGE_PMD_SIZE;
mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end);
- spin_lock(&mm->page_table_lock);
+ spin_lock(ptl);
if (page)
- put_page(page);
+ put_user_huge_page(page);
if (unlikely(!pmd_same(*pmd, orig_pmd))) {
- spin_unlock(&mm->page_table_lock);
+ spin_unlock(ptl);
mem_cgroup_uncharge_page(new_page);
put_page(new_page);
goto out_mn;
@@ -1203,23 +1172,23 @@ alloc:
page_add_new_anon_rmap(new_page, vma, haddr);
set_pmd_at(mm, haddr, pmd, entry);
update_mmu_cache_pmd(vma, address, pmd);
- if (is_huge_zero_pmd(orig_pmd)) {
+ if (!page) {
add_mm_counter(mm, MM_ANONPAGES, HPAGE_PMD_NR);
put_huge_zero_page();
} else {
- VM_BUG_ON(!PageHead(page));
+ VM_BUG_ON_PAGE(!PageHead(page), page);
page_remove_rmap(page);
put_page(page);
}
ret |= VM_FAULT_WRITE;
}
- spin_unlock(&mm->page_table_lock);
+ spin_unlock(ptl);
out_mn:
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
out:
return ret;
out_unlock:
- spin_unlock(&mm->page_table_lock);
+ spin_unlock(ptl);
return ret;
}
@@ -1231,7 +1200,7 @@ struct page *follow_trans_huge_pmd(struct vm_area_struct *vma,
struct mm_struct *mm = vma->vm_mm;
struct page *page = NULL;
- assert_spin_locked(&mm->page_table_lock);
+ assert_spin_locked(pmd_lockptr(mm, pmd));
if (flags & FOLL_WRITE && !pmd_write(*pmd))
goto out;
@@ -1240,8 +1209,12 @@ struct page *follow_trans_huge_pmd(struct vm_area_struct *vma,
if ((flags & FOLL_DUMP) && is_huge_zero_pmd(*pmd))
return ERR_PTR(-EFAULT);
+ /* Full NUMA hinting faults to serialise migration in fault paths */
+ if ((flags & FOLL_NUMA) && pmd_numa(*pmd))
+ goto out;
+
page = pmd_page(*pmd);
- VM_BUG_ON(!PageHead(page));
+ VM_BUG_ON_PAGE(!PageHead(page), page);
if (flags & FOLL_TOUCH) {
pmd_t _pmd;
/*
@@ -1266,7 +1239,7 @@ struct page *follow_trans_huge_pmd(struct vm_area_struct *vma,
}
}
page += (addr & ~HPAGE_PMD_MASK) >> PAGE_SHIFT;
- VM_BUG_ON(!PageCompound(page));
+ VM_BUG_ON_PAGE(!PageCompound(page), page);
if (flags & FOLL_GET)
get_page_foll(page);
@@ -1278,73 +1251,133 @@ out:
int do_huge_pmd_numa_page(struct mm_struct *mm, struct vm_area_struct *vma,
unsigned long addr, pmd_t pmd, pmd_t *pmdp)
{
+ spinlock_t *ptl;
+ struct anon_vma *anon_vma = NULL;
struct page *page;
unsigned long haddr = addr & HPAGE_PMD_MASK;
- int target_nid;
- int current_nid = -1;
- bool migrated;
+ int page_nid = -1, this_nid = numa_node_id();
+ int target_nid, last_cpupid = -1;
+ bool page_locked;
+ bool migrated = false;
+ int flags = 0;
- spin_lock(&mm->page_table_lock);
+ ptl = pmd_lock(mm, pmdp);
if (unlikely(!pmd_same(pmd, *pmdp)))
goto out_unlock;
+ /*
+ * If there are potential migrations, wait for completion and retry
+ * without disrupting NUMA hinting information. Do not relock and
+ * check_same as the page may no longer be mapped.
+ */
+ if (unlikely(pmd_trans_migrating(*pmdp))) {
+ spin_unlock(ptl);
+ wait_migrate_huge_page(vma->anon_vma, pmdp);
+ goto out;
+ }
+
page = pmd_page(pmd);
- get_page(page);
- current_nid = page_to_nid(page);
+ BUG_ON(is_huge_zero_page(page));
+ page_nid = page_to_nid(page);
+ last_cpupid = page_cpupid_last(page);
count_vm_numa_event(NUMA_HINT_FAULTS);
- if (current_nid == numa_node_id())
+ if (page_nid == this_nid) {
count_vm_numa_event(NUMA_HINT_FAULTS_LOCAL);
+ flags |= TNF_FAULT_LOCAL;
+ }
+ /*
+ * Avoid grouping on DSO/COW pages in specific and RO pages
+ * in general, RO pages shouldn't hurt as much anyway since
+ * they can be in shared cache state.
+ */
+ if (!pmd_write(pmd))
+ flags |= TNF_NO_GROUP;
+
+ /*
+ * Acquire the page lock to serialise THP migrations but avoid dropping
+ * page_table_lock if at all possible
+ */
+ page_locked = trylock_page(page);
target_nid = mpol_misplaced(page, vma, haddr);
if (target_nid == -1) {
- put_page(page);
- goto clear_pmdnuma;
+ /* If the page was locked, there are no parallel migrations */
+ if (page_locked)
+ goto clear_pmdnuma;
+ }
+
+ /* Migration could have started since the pmd_trans_migrating check */
+ if (!page_locked) {
+ spin_unlock(ptl);
+ wait_on_page_locked(page);
+ page_nid = -1;
+ goto out;
}
- /* Acquire the page lock to serialise THP migrations */
- spin_unlock(&mm->page_table_lock);
- lock_page(page);
+ /*
+ * Page is misplaced. Page lock serialises migrations. Acquire anon_vma
+ * to serialises splits
+ */
+ get_page(page);
+ spin_unlock(ptl);
+ anon_vma = page_lock_anon_vma_read(page);
- /* Confirm the PTE did not while locked */
- spin_lock(&mm->page_table_lock);
+ /* Confirm the PMD did not change while page_table_lock was released */
+ spin_lock(ptl);
if (unlikely(!pmd_same(pmd, *pmdp))) {
unlock_page(page);
put_page(page);
+ page_nid = -1;
goto out_unlock;
}
- spin_unlock(&mm->page_table_lock);
- /* Migrate the THP to the requested node */
+ /* Bail if we fail to protect against THP splits for any reason */
+ if (unlikely(!anon_vma)) {
+ put_page(page);
+ page_nid = -1;
+ goto clear_pmdnuma;
+ }
+
+ /*
+ * Migrate the THP to the requested node, returns with page unlocked
+ * and pmd_numa cleared.
+ */
+ spin_unlock(ptl);
migrated = migrate_misplaced_transhuge_page(mm, vma,
pmdp, pmd, addr, page, target_nid);
- if (!migrated)
- goto check_same;
-
- task_numa_fault(target_nid, HPAGE_PMD_NR, true);
- return 0;
+ if (migrated) {
+ flags |= TNF_MIGRATED;
+ page_nid = target_nid;
+ }
-check_same:
- spin_lock(&mm->page_table_lock);
- if (unlikely(!pmd_same(pmd, *pmdp)))
- goto out_unlock;
+ goto out;
clear_pmdnuma:
+ BUG_ON(!PageLocked(page));
pmd = pmd_mknonnuma(pmd);
set_pmd_at(mm, haddr, pmdp, pmd);
VM_BUG_ON(pmd_numa(*pmdp));
update_mmu_cache_pmd(vma, addr, pmdp);
+ unlock_page(page);
out_unlock:
- spin_unlock(&mm->page_table_lock);
- if (current_nid != -1)
- task_numa_fault(current_nid, HPAGE_PMD_NR, false);
+ spin_unlock(ptl);
+
+out:
+ if (anon_vma)
+ page_unlock_anon_vma_read(anon_vma);
+
+ if (page_nid != -1)
+ task_numa_fault(last_cpupid, page_nid, HPAGE_PMD_NR, flags);
+
return 0;
}
int zap_huge_pmd(struct mmu_gather *tlb, struct vm_area_struct *vma,
pmd_t *pmd, unsigned long addr)
{
+ spinlock_t *ptl;
int ret = 0;
- if (__pmd_trans_huge_lock(pmd, vma) == 1) {
+ if (__pmd_trans_huge_lock(pmd, vma, &ptl) == 1) {
struct page *page;
pgtable_t pgtable;
pmd_t orig_pmd;
@@ -1358,17 +1391,17 @@ int zap_huge_pmd(struct mmu_gather *tlb, struct vm_area_struct *vma,
tlb_remove_pmd_tlb_entry(tlb, pmd, addr);
pgtable = pgtable_trans_huge_withdraw(tlb->mm, pmd);
if (is_huge_zero_pmd(orig_pmd)) {
- tlb->mm->nr_ptes--;
- spin_unlock(&tlb->mm->page_table_lock);
+ atomic_long_dec(&tlb->mm->nr_ptes);
+ spin_unlock(ptl);
put_huge_zero_page();
} else {
page = pmd_page(orig_pmd);
page_remove_rmap(page);
- VM_BUG_ON(page_mapcount(page) < 0);
+ VM_BUG_ON_PAGE(page_mapcount(page) < 0, page);
add_mm_counter(tlb->mm, MM_ANONPAGES, -HPAGE_PMD_NR);
- VM_BUG_ON(!PageHead(page));
- tlb->mm->nr_ptes--;
- spin_unlock(&tlb->mm->page_table_lock);
+ VM_BUG_ON_PAGE(!PageHead(page), page);
+ atomic_long_dec(&tlb->mm->nr_ptes);
+ spin_unlock(ptl);
tlb_remove_page(tlb, page);
}
pte_free(tlb->mm, pgtable);
@@ -1381,14 +1414,15 @@ int mincore_huge_pmd(struct vm_area_struct *vma, pmd_t *pmd,
unsigned long addr, unsigned long end,
unsigned char *vec)
{
+ spinlock_t *ptl;
int ret = 0;
- if (__pmd_trans_huge_lock(pmd, vma) == 1) {
+ if (__pmd_trans_huge_lock(pmd, vma, &ptl) == 1) {
/*
* All logical pages in the range are present
* if backed by a huge page.
*/
- spin_unlock(&vma->vm_mm->page_table_lock);
+ spin_unlock(ptl);
memset(vec, 1, (end - addr) >> PAGE_SHIFT);
ret = 1;
}
@@ -1401,6 +1435,7 @@ int move_huge_pmd(struct vm_area_struct *vma, struct vm_area_struct *new_vma,
unsigned long new_addr, unsigned long old_end,
pmd_t *old_pmd, pmd_t *new_pmd)
{
+ spinlock_t *old_ptl, *new_ptl;
int ret = 0;
pmd_t pmd;
@@ -1421,41 +1456,72 @@ int move_huge_pmd(struct vm_area_struct *vma, struct vm_area_struct *new_vma,
goto out;
}
- ret = __pmd_trans_huge_lock(old_pmd, vma);
+ /*
+ * We don't have to worry about the ordering of src and dst
+ * ptlocks because exclusive mmap_sem prevents deadlock.
+ */
+ ret = __pmd_trans_huge_lock(old_pmd, vma, &old_ptl);
if (ret == 1) {
+ new_ptl = pmd_lockptr(mm, new_pmd);
+ if (new_ptl != old_ptl)
+ spin_lock_nested(new_ptl, SINGLE_DEPTH_NESTING);
pmd = pmdp_get_and_clear(mm, old_addr, old_pmd);
VM_BUG_ON(!pmd_none(*new_pmd));
+
+ if (pmd_move_must_withdraw(new_ptl, old_ptl)) {
+ pgtable_t pgtable;
+ pgtable = pgtable_trans_huge_withdraw(mm, old_pmd);
+ pgtable_trans_huge_deposit(mm, new_pmd, pgtable);
+ }
set_pmd_at(mm, new_addr, new_pmd, pmd_mksoft_dirty(pmd));
- spin_unlock(&mm->page_table_lock);
+ if (new_ptl != old_ptl)
+ spin_unlock(new_ptl);
+ spin_unlock(old_ptl);
}
out:
return ret;
}
+/*
+ * Returns
+ * - 0 if PMD could not be locked
+ * - 1 if PMD was locked but protections unchange and TLB flush unnecessary
+ * - HPAGE_PMD_NR is protections changed and TLB flush necessary
+ */
int change_huge_pmd(struct vm_area_struct *vma, pmd_t *pmd,
unsigned long addr, pgprot_t newprot, int prot_numa)
{
struct mm_struct *mm = vma->vm_mm;
+ spinlock_t *ptl;
int ret = 0;
- if (__pmd_trans_huge_lock(pmd, vma) == 1) {
+ if (__pmd_trans_huge_lock(pmd, vma, &ptl) == 1) {
pmd_t entry;
- entry = pmdp_get_and_clear(mm, addr, pmd);
+ ret = 1;
if (!prot_numa) {
+ entry = pmdp_get_and_clear(mm, addr, pmd);
+ if (pmd_numa(entry))
+ entry = pmd_mknonnuma(entry);
entry = pmd_modify(entry, newprot);
+ ret = HPAGE_PMD_NR;
+ set_pmd_at(mm, addr, pmd, entry);
BUG_ON(pmd_write(entry));
} else {
struct page *page = pmd_page(*pmd);
- /* only check non-shared pages */
- if (page_mapcount(page) == 1 &&
+ /*
+ * Do not trap faults against the zero page. The
+ * read-only data is likely to be read-cached on the
+ * local CPU cache and it is less useful to know about
+ * local vs remote hits on the zero page.
+ */
+ if (!is_huge_zero_page(page) &&
!pmd_numa(*pmd)) {
- entry = pmd_mknuma(entry);
+ pmdp_set_numa(mm, addr, pmd);
+ ret = HPAGE_PMD_NR;
}
}
- set_pmd_at(mm, addr, pmd, entry);
- spin_unlock(&vma->vm_mm->page_table_lock);
- ret = 1;
+ spin_unlock(ptl);
}
return ret;
@@ -1468,12 +1534,13 @@ int change_huge_pmd(struct vm_area_struct *vma, pmd_t *pmd,
* Note that if it returns 1, this routine returns without unlocking page
* table locks. So callers must unlock them.
*/
-int __pmd_trans_huge_lock(pmd_t *pmd, struct vm_area_struct *vma)
+int __pmd_trans_huge_lock(pmd_t *pmd, struct vm_area_struct *vma,
+ spinlock_t **ptl)
{
- spin_lock(&vma->vm_mm->page_table_lock);
+ *ptl = pmd_lock(vma->vm_mm, pmd);
if (likely(pmd_trans_huge(*pmd))) {
if (unlikely(pmd_trans_splitting(*pmd))) {
- spin_unlock(&vma->vm_mm->page_table_lock);
+ spin_unlock(*ptl);
wait_split_huge_page(vma->anon_vma, pmd);
return -1;
} else {
@@ -1482,27 +1549,44 @@ int __pmd_trans_huge_lock(pmd_t *pmd, struct vm_area_struct *vma)
return 1;
}
}
- spin_unlock(&vma->vm_mm->page_table_lock);
+ spin_unlock(*ptl);
return 0;
}
+/*
+ * This function returns whether a given @page is mapped onto the @address
+ * in the virtual space of @mm.
+ *
+ * When it's true, this function returns *pmd with holding the page table lock
+ * and passing it back to the caller via @ptl.
+ * If it's false, returns NULL without holding the page table lock.
+ */
pmd_t *page_check_address_pmd(struct page *page,
struct mm_struct *mm,
unsigned long address,
- enum page_check_address_pmd_flag flag)
+ enum page_check_address_pmd_flag flag,
+ spinlock_t **ptl)
{
- pmd_t *pmd, *ret = NULL;
+ pgd_t *pgd;
+ pud_t *pud;
+ pmd_t *pmd;
if (address & ~HPAGE_PMD_MASK)
- goto out;
+ return NULL;
- pmd = mm_find_pmd(mm, address);
- if (!pmd)
- goto out;
- if (pmd_none(*pmd))
- goto out;
+ pgd = pgd_offset(mm, address);
+ if (!pgd_present(*pgd))
+ return NULL;
+ pud = pud_offset(pgd, address);
+ if (!pud_present(*pud))
+ return NULL;
+ pmd = pmd_offset(pud, address);
+
+ *ptl = pmd_lock(mm, pmd);
+ if (!pmd_present(*pmd))
+ goto unlock;
if (pmd_page(*pmd) != page)
- goto out;
+ goto unlock;
/*
* split_vma() may create temporary aliased mappings. There is
* no risk as long as all huge pmd are found and have their
@@ -1512,14 +1596,15 @@ pmd_t *page_check_address_pmd(struct page *page,
*/
if (flag == PAGE_CHECK_ADDRESS_PMD_NOTSPLITTING_FLAG &&
pmd_trans_splitting(*pmd))
- goto out;
+ goto unlock;
if (pmd_trans_huge(*pmd)) {
VM_BUG_ON(flag == PAGE_CHECK_ADDRESS_PMD_SPLITTING_FLAG &&
!pmd_trans_splitting(*pmd));
- ret = pmd;
+ return pmd;
}
-out:
- return ret;
+unlock:
+ spin_unlock(*ptl);
+ return NULL;
}
static int __split_huge_page_splitting(struct page *page,
@@ -1527,6 +1612,7 @@ static int __split_huge_page_splitting(struct page *page,
unsigned long address)
{
struct mm_struct *mm = vma->vm_mm;
+ spinlock_t *ptl;
pmd_t *pmd;
int ret = 0;
/* For mmu_notifiers */
@@ -1534,9 +1620,8 @@ static int __split_huge_page_splitting(struct page *page,
const unsigned long mmun_end = address + HPAGE_PMD_SIZE;
mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end);
- spin_lock(&mm->page_table_lock);
pmd = page_check_address_pmd(page, mm, address,
- PAGE_CHECK_ADDRESS_PMD_NOTSPLITTING_FLAG);
+ PAGE_CHECK_ADDRESS_PMD_NOTSPLITTING_FLAG, &ptl);
if (pmd) {
/*
* We can't temporarily set the pmd to null in order
@@ -1547,8 +1632,8 @@ static int __split_huge_page_splitting(struct page *page,
*/
pmdp_splitting_flush(vma, address, pmd);
ret = 1;
+ spin_unlock(ptl);
}
- spin_unlock(&mm->page_table_lock);
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
return ret;
@@ -1636,7 +1721,7 @@ static void __split_huge_page_refcount(struct page *page,
page_tail->mapping = page->mapping;
page_tail->index = page->index + i;
- page_nid_xchg_last(page_tail, page_nid_last(page));
+ page_cpupid_xchg_last(page_tail, page_cpupid_last(page));
BUG_ON(!PageAnon(page_tail));
BUG_ON(!PageUptodate(page_tail));
@@ -1679,14 +1764,14 @@ static int __split_huge_page_map(struct page *page,
unsigned long address)
{
struct mm_struct *mm = vma->vm_mm;
+ spinlock_t *ptl;
pmd_t *pmd, _pmd;
int ret = 0, i;
pgtable_t pgtable;
unsigned long haddr;
- spin_lock(&mm->page_table_lock);
pmd = page_check_address_pmd(page, mm, address,
- PAGE_CHECK_ADDRESS_PMD_SPLITTING_FLAG);
+ PAGE_CHECK_ADDRESS_PMD_SPLITTING_FLAG, &ptl);
if (pmd) {
pgtable = pgtable_trans_huge_withdraw(mm, pmd);
pmd_populate(mm, &_pmd, pgtable);
@@ -1741,8 +1826,8 @@ static int __split_huge_page_map(struct page *page,
pmdp_invalidate(vma, address, pmd);
pmd_populate(mm, pmd, pgtable);
ret = 1;
+ spin_unlock(ptl);
}
- spin_unlock(&mm->page_table_lock);
return ret;
}
@@ -1776,10 +1861,11 @@ static void __split_huge_page(struct page *page,
* the newly established pmd of the child later during the
* walk, to be able to set it as pmd_trans_splitting too.
*/
- if (mapcount != page_mapcount(page))
- printk(KERN_ERR "mapcount %d page_mapcount %d\n",
- mapcount, page_mapcount(page));
- BUG_ON(mapcount != page_mapcount(page));
+ if (mapcount != page_mapcount(page)) {
+ pr_err("mapcount %d page_mapcount %d\n",
+ mapcount, page_mapcount(page));
+ BUG();
+ }
__split_huge_page_refcount(page, list);
@@ -1790,10 +1876,11 @@ static void __split_huge_page(struct page *page,
BUG_ON(is_vma_temporary_stack(vma));
mapcount2 += __split_huge_page_map(page, vma, addr);
}
- if (mapcount != mapcount2)
- printk(KERN_ERR "mapcount %d mapcount2 %d page_mapcount %d\n",
- mapcount, mapcount2, page_mapcount(page));
- BUG_ON(mapcount != mapcount2);
+ if (mapcount != mapcount2) {
+ pr_err("mapcount %d mapcount2 %d page_mapcount %d\n",
+ mapcount, mapcount2, page_mapcount(page));
+ BUG();
+ }
}
/*
@@ -1839,22 +1926,27 @@ out:
return ret;
}
-#define VM_NO_THP (VM_SPECIAL|VM_MIXEDMAP|VM_HUGETLB|VM_SHARED|VM_MAYSHARE)
+#define VM_NO_THP (VM_SPECIAL | VM_HUGETLB | VM_SHARED | VM_MAYSHARE)
int hugepage_madvise(struct vm_area_struct *vma,
unsigned long *vm_flags, int advice)
{
- struct mm_struct *mm = vma->vm_mm;
-
switch (advice) {
case MADV_HUGEPAGE:
+#ifdef CONFIG_S390
+ /*
+ * qemu blindly sets MADV_HUGEPAGE on all allocations, but s390
+ * can't handle this properly after s390_enable_sie, so we simply
+ * ignore the madvise to prevent qemu from causing a SIGSEGV.
+ */
+ if (mm_has_pgste(vma->vm_mm))
+ return 0;
+#endif
/*
* Be somewhat over-protective like KSM for now!
*/
if (*vm_flags & (VM_HUGEPAGE | VM_NO_THP))
return -EINVAL;
- if (mm->def_flags & VM_NOHUGEPAGE)
- return -EINVAL;
*vm_flags &= ~VM_NOHUGEPAGE;
*vm_flags |= VM_HUGEPAGE;
/*
@@ -2054,9 +2146,9 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma,
if (unlikely(!page))
goto out;
- VM_BUG_ON(PageCompound(page));
- BUG_ON(!PageAnon(page));
- VM_BUG_ON(!PageSwapBacked(page));
+ VM_BUG_ON_PAGE(PageCompound(page), page);
+ VM_BUG_ON_PAGE(!PageAnon(page), page);
+ VM_BUG_ON_PAGE(!PageSwapBacked(page), page);
/* cannot use mapcount: can't collapse if there's a gup pin */
if (page_count(page) != 1)
@@ -2079,8 +2171,8 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma,
}
/* 0 stands for page_is_file_cache(page) == false */
inc_zone_page_state(page, NR_ISOLATED_ANON + 0);
- VM_BUG_ON(!PageLocked(page));
- VM_BUG_ON(PageLRU(page));
+ VM_BUG_ON_PAGE(!PageLocked(page), page);
+ VM_BUG_ON_PAGE(PageLRU(page), page);
/* If there is no mapped pte young don't collapse the page */
if (pte_young(pteval) || PageReferenced(page) ||
@@ -2110,7 +2202,7 @@ static void __collapse_huge_page_copy(pte_t *pte, struct page *page,
} else {
src_page = pte_page(pteval);
copy_user_highpage(page, src_page, address, vma);
- VM_BUG_ON(page_mapcount(src_page) != 1);
+ VM_BUG_ON_PAGE(page_mapcount(src_page) != 1, src_page);
release_pte_page(src_page);
/*
* ptl mostly unnecessary, but preempt has to
@@ -2139,7 +2231,34 @@ static void khugepaged_alloc_sleep(void)
msecs_to_jiffies(khugepaged_alloc_sleep_millisecs));
}
+static int khugepaged_node_load[MAX_NUMNODES];
+
#ifdef CONFIG_NUMA
+static int khugepaged_find_target_node(void)
+{
+ static int last_khugepaged_target_node = NUMA_NO_NODE;
+ int nid, target_node = 0, max_value = 0;
+
+ /* find first node with max normal pages hit */
+ for (nid = 0; nid < MAX_NUMNODES; nid++)
+ if (khugepaged_node_load[nid] > max_value) {
+ max_value = khugepaged_node_load[nid];
+ target_node = nid;
+ }
+
+ /* do some balance if several nodes have the same hit record */
+ if (target_node <= last_khugepaged_target_node)
+ for (nid = last_khugepaged_target_node + 1; nid < MAX_NUMNODES;
+ nid++)
+ if (max_value == khugepaged_node_load[nid]) {
+ target_node = nid;
+ break;
+ }
+
+ last_khugepaged_target_node = target_node;
+ return target_node;
+}
+
static bool khugepaged_prealloc_page(struct page **hpage, bool *wait)
{
if (IS_ERR(*hpage)) {
@@ -2162,7 +2281,7 @@ static struct page
struct vm_area_struct *vma, unsigned long address,
int node)
{
- VM_BUG_ON(*hpage);
+ VM_BUG_ON_PAGE(*hpage, *hpage);
/*
* Allocate the page while the vma is still valid and under
* the mmap_sem read mode so there is no memory allocation
@@ -2173,9 +2292,8 @@ static struct page
* mmap_sem in read mode is good idea also to allow greater
* scalability.
*/
- *hpage = alloc_hugepage_vma(khugepaged_defrag(), vma, address,
- node, __GFP_OTHER_NODE);
-
+ *hpage = alloc_pages_exact_node(node, alloc_hugepage_gfpmask(
+ khugepaged_defrag(), __GFP_OTHER_NODE), HPAGE_PMD_ORDER);
/*
* After allocating the hugepage, release the mmap_sem read lock in
* preparation for taking it in write mode.
@@ -2191,6 +2309,17 @@ static struct page
return *hpage;
}
#else
+static int khugepaged_find_target_node(void)
+{
+ return 0;
+}
+
+static inline struct page *alloc_hugepage(int defrag)
+{
+ return alloc_pages(alloc_hugepage_gfpmask(defrag, 0),
+ HPAGE_PMD_ORDER);
+}
+
static struct page *khugepaged_alloc_hugepage(bool *wait)
{
struct page *hpage;
@@ -2257,7 +2386,7 @@ static void collapse_huge_page(struct mm_struct *mm,
pte_t *pte;
pgtable_t pgtable;
struct page *new_page;
- spinlock_t *ptl;
+ spinlock_t *pmd_ptl, *pte_ptl;
int isolated;
unsigned long hstart, hend;
unsigned long mmun_start; /* For mmu_notifiers */
@@ -2270,7 +2399,7 @@ static void collapse_huge_page(struct mm_struct *mm,
if (!new_page)
return;
- if (unlikely(mem_cgroup_newpage_charge(new_page, mm, GFP_KERNEL)))
+ if (unlikely(mem_cgroup_charge_anon(new_page, mm, GFP_KERNEL)))
return;
/*
@@ -2294,18 +2423,16 @@ static void collapse_huge_page(struct mm_struct *mm,
pmd = mm_find_pmd(mm, address);
if (!pmd)
goto out;
- if (pmd_trans_huge(*pmd))
- goto out;
anon_vma_lock_write(vma->anon_vma);
pte = pte_offset_map(pmd, address);
- ptl = pte_lockptr(mm, pmd);
+ pte_ptl = pte_lockptr(mm, pmd);
mmun_start = address;
mmun_end = address + HPAGE_PMD_SIZE;
mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end);
- spin_lock(&mm->page_table_lock); /* probably unnecessary */
+ pmd_ptl = pmd_lock(mm, pmd); /* probably unnecessary */
/*
* After this gup_fast can't run anymore. This also removes
* any huge TLB entry from the CPU so we won't allow
@@ -2313,16 +2440,16 @@ static void collapse_huge_page(struct mm_struct *mm,
* to avoid the risk of CPU bugs in that area.
*/
_pmd = pmdp_clear_flush(vma, address, pmd);
- spin_unlock(&mm->page_table_lock);
+ spin_unlock(pmd_ptl);
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
- spin_lock(ptl);
+ spin_lock(pte_ptl);
isolated = __collapse_huge_page_isolate(vma, address, pte);
- spin_unlock(ptl);
+ spin_unlock(pte_ptl);
if (unlikely(!isolated)) {
pte_unmap(pte);
- spin_lock(&mm->page_table_lock);
+ spin_lock(pmd_ptl);
BUG_ON(!pmd_none(*pmd));
/*
* We can only use set_pmd_at when establishing
@@ -2330,7 +2457,7 @@ static void collapse_huge_page(struct mm_struct *mm,
* points to regular pagetables. Use pmd_populate for that
*/
pmd_populate(mm, pmd, pmd_pgtable(_pmd));
- spin_unlock(&mm->page_table_lock);
+ spin_unlock(pmd_ptl);
anon_vma_unlock_write(vma->anon_vma);
goto out;
}
@@ -2341,7 +2468,7 @@ static void collapse_huge_page(struct mm_struct *mm,
*/
anon_vma_unlock_write(vma->anon_vma);
- __collapse_huge_page_copy(pte, new_page, vma, address, ptl);
+ __collapse_huge_page_copy(pte, new_page, vma, address, pte_ptl);
pte_unmap(pte);
__SetPageUptodate(new_page);
pgtable = pmd_pgtable(_pmd);
@@ -2356,13 +2483,13 @@ static void collapse_huge_page(struct mm_struct *mm,
*/
smp_wmb();
- spin_lock(&mm->page_table_lock);
+ spin_lock(pmd_ptl);
BUG_ON(!pmd_none(*pmd));
page_add_new_anon_rmap(new_page, vma, address);
pgtable_trans_huge_deposit(mm, pmd, pgtable);
set_pmd_at(mm, address, pmd, _pmd);
update_mmu_cache_pmd(vma, address, pmd);
- spin_unlock(&mm->page_table_lock);
+ spin_unlock(pmd_ptl);
*hpage = NULL;
@@ -2394,9 +2521,8 @@ static int khugepaged_scan_pmd(struct mm_struct *mm,
pmd = mm_find_pmd(mm, address);
if (!pmd)
goto out;
- if (pmd_trans_huge(*pmd))
- goto out;
+ memset(khugepaged_node_load, 0, sizeof(khugepaged_node_load));
pte = pte_offset_map_lock(mm, pmd, address, &ptl);
for (_address = address, _pte = pte; _pte < pte+HPAGE_PMD_NR;
_pte++, _address += PAGE_SIZE) {
@@ -2413,13 +2539,14 @@ static int khugepaged_scan_pmd(struct mm_struct *mm,
if (unlikely(!page))
goto out_unmap;
/*
- * Chose the node of the first page. This could
- * be more sophisticated and look at more pages,
- * but isn't for now.
+ * Record which node the original page is from and save this
+ * information to khugepaged_node_load[].
+ * Khupaged will allocate hugepage from the node has the max
+ * hit record.
*/
- if (node == NUMA_NO_NODE)
- node = page_to_nid(page);
- VM_BUG_ON(PageCompound(page));
+ node = page_to_nid(page);
+ khugepaged_node_load[node]++;
+ VM_BUG_ON_PAGE(PageCompound(page), page);
if (!PageLRU(page) || PageLocked(page) || !PageAnon(page))
goto out_unmap;
/* cannot use mapcount: can't collapse if there's a gup pin */
@@ -2433,9 +2560,11 @@ static int khugepaged_scan_pmd(struct mm_struct *mm,
ret = 1;
out_unmap:
pte_unmap_unlock(pte, ptl);
- if (ret)
+ if (ret) {
+ node = khugepaged_find_target_node();
/* collapse_huge_page will return with the mmap_sem released */
collapse_huge_page(mm, address, hpage, vma, node);
+ }
out:
return ret;
}
@@ -2640,7 +2769,7 @@ static int khugepaged(void *none)
struct mm_slot *mm_slot;
set_freezable();
- set_user_nice(current, 19);
+ set_user_nice(current, MAX_NICE);
while (!kthread_should_stop()) {
khugepaged_do_scan();
@@ -2687,6 +2816,7 @@ static void __split_huge_zero_page_pmd(struct vm_area_struct *vma,
void __split_huge_page_pmd(struct vm_area_struct *vma, unsigned long address,
pmd_t *pmd)
{
+ spinlock_t *ptl;
struct page *page;
struct mm_struct *mm = vma->vm_mm;
unsigned long haddr = address & HPAGE_PMD_MASK;
@@ -2697,29 +2827,37 @@ void __split_huge_page_pmd(struct vm_area_struct *vma, unsigned long address,
mmun_start = haddr;
mmun_end = haddr + HPAGE_PMD_SIZE;
+again:
mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end);
- spin_lock(&mm->page_table_lock);
+ ptl = pmd_lock(mm, pmd);
if (unlikely(!pmd_trans_huge(*pmd))) {
- spin_unlock(&mm->page_table_lock);
+ spin_unlock(ptl);
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
return;
}
if (is_huge_zero_pmd(*pmd)) {
__split_huge_zero_page_pmd(vma, haddr, pmd);
- spin_unlock(&mm->page_table_lock);
+ spin_unlock(ptl);
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
return;
}
page = pmd_page(*pmd);
- VM_BUG_ON(!page_count(page));
+ VM_BUG_ON_PAGE(!page_count(page), page);
get_page(page);
- spin_unlock(&mm->page_table_lock);
+ spin_unlock(ptl);
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
split_huge_page(page);
put_page(page);
- BUG_ON(pmd_trans_huge(*pmd));
+
+ /*
+ * We don't always have down_write of mmap_sem here: a racing
+ * do_huge_pmd_wp_page() might have copied-on-write to another
+ * huge page before our split_huge_page() got the anon_vma lock.
+ */
+ if (unlikely(pmd_trans_huge(*pmd)))
+ goto again;
}
void split_huge_page_pmd_mm(struct mm_struct *mm, unsigned long address,
@@ -2735,12 +2873,22 @@ void split_huge_page_pmd_mm(struct mm_struct *mm, unsigned long address,
static void split_huge_page_address(struct mm_struct *mm,
unsigned long address)
{
+ pgd_t *pgd;
+ pud_t *pud;
pmd_t *pmd;
VM_BUG_ON(!(address & ~HPAGE_PMD_MASK));
- pmd = mm_find_pmd(mm, address);
- if (!pmd)
+ pgd = pgd_offset(mm, address);
+ if (!pgd_present(*pgd))
+ return;
+
+ pud = pud_offset(pgd, address);
+ if (!pud_present(*pud))
+ return;
+
+ pmd = pmd_offset(pud, address);
+ if (!pmd_present(*pmd))
return;
/*
* Caller holds the mmap_sem write mode, so a huge pmd cannot
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2025-11-05 08:19:59 +0000