SCAN_SUCCEED,
SCAN_PMD_NULL,
SCAN_EXCEED_NONE_PTE,
+ SCAN_EXCEED_SWAP_PTE,
+ SCAN_EXCEED_SHARED_PTE,
SCAN_PTE_NON_PRESENT,
SCAN_PTE_UFFD_WP,
SCAN_PAGE_RO,
SCAN_DEL_PAGE_LRU,
SCAN_ALLOC_HUGE_PAGE_FAIL,
SCAN_CGROUP_CHARGE_FAIL,
- SCAN_EXCEED_SWAP_PTE,
SCAN_TRUNCATED,
SCAN_PAGE_HAS_PRIVATE,
};
*/
static unsigned int khugepaged_max_ptes_none __read_mostly;
static unsigned int khugepaged_max_ptes_swap __read_mostly;
+static unsigned int khugepaged_max_ptes_shared __read_mostly;
#define MM_SLOTS_HASH_BITS 10
static __read_mostly DEFINE_HASHTABLE(mm_slots_hash, MM_SLOTS_HASH_BITS);
__ATTR(max_ptes_swap, 0644, khugepaged_max_ptes_swap_show,
khugepaged_max_ptes_swap_store);
+static ssize_t khugepaged_max_ptes_shared_show(struct kobject *kobj,
+ struct kobj_attribute *attr,
+ char *buf)
+{
+ return sprintf(buf, "%u\n", khugepaged_max_ptes_shared);
+}
+
+static ssize_t khugepaged_max_ptes_shared_store(struct kobject *kobj,
+ struct kobj_attribute *attr,
+ const char *buf, size_t count)
+{
+ int err;
+ unsigned long max_ptes_shared;
+
+ err = kstrtoul(buf, 10, &max_ptes_shared);
+ if (err || max_ptes_shared > HPAGE_PMD_NR-1)
+ return -EINVAL;
+
+ khugepaged_max_ptes_shared = max_ptes_shared;
+
+ return count;
+}
+
+static struct kobj_attribute khugepaged_max_ptes_shared_attr =
+ __ATTR(max_ptes_shared, 0644, khugepaged_max_ptes_shared_show,
+ khugepaged_max_ptes_shared_store);
+
static struct attribute *khugepaged_attr[] = {
&khugepaged_defrag_attr.attr,
&khugepaged_max_ptes_none_attr.attr,
+ &khugepaged_max_ptes_swap_attr.attr,
+ &khugepaged_max_ptes_shared_attr.attr,
&pages_to_scan_attr.attr,
&pages_collapsed_attr.attr,
&full_scans_attr.attr,
&scan_sleep_millisecs_attr.attr,
&alloc_sleep_millisecs_attr.attr,
- &khugepaged_max_ptes_swap_attr.attr,
NULL,
};
khugepaged_pages_to_scan = HPAGE_PMD_NR * 8;
khugepaged_max_ptes_none = HPAGE_PMD_NR - 1;
khugepaged_max_ptes_swap = HPAGE_PMD_NR / 8;
+ khugepaged_max_ptes_shared = HPAGE_PMD_NR / 2;
return 0;
}
* under mmap sem read mode). Stop here (after we
* return all pagetables will be destroyed) until
* khugepaged has finished working on the pagetables
- * under the mmap_sem.
+ * under the mmap_lock.
*/
- down_write(&mm->mmap_sem);
- up_write(&mm->mmap_sem);
+ mmap_write_lock(mm);
+ mmap_write_unlock(mm);
}
}
static void release_pte_page(struct page *page)
{
- dec_node_page_state(page, NR_ISOLATED_ANON + page_is_file_lru(page));
+ mod_node_page_state(page_pgdat(page),
+ NR_ISOLATED_ANON + page_is_file_lru(page),
+ -compound_nr(page));
unlock_page(page);
putback_lru_page(page);
}
-static void release_pte_pages(pte_t *pte, pte_t *_pte)
+static void release_pte_pages(pte_t *pte, pte_t *_pte,
+ struct list_head *compound_pagelist)
{
+ struct page *page, *tmp;
+
while (--_pte >= pte) {
pte_t pteval = *_pte;
- if (!pte_none(pteval) && !is_zero_pfn(pte_pfn(pteval)))
- release_pte_page(pte_page(pteval));
+
+ page = pte_page(pteval);
+ if (!pte_none(pteval) && !is_zero_pfn(pte_pfn(pteval)) &&
+ !PageCompound(page))
+ release_pte_page(page);
+ }
+
+ list_for_each_entry_safe(page, tmp, compound_pagelist, lru) {
+ list_del(&page->lru);
+ release_pte_page(page);
}
}
+static bool is_refcount_suitable(struct page *page)
+{
+ int expected_refcount;
+
+ expected_refcount = total_mapcount(page);
+ if (PageSwapCache(page))
+ expected_refcount += compound_nr(page);
+
+ return page_count(page) == expected_refcount;
+}
+
static int __collapse_huge_page_isolate(struct vm_area_struct *vma,
unsigned long address,
- pte_t *pte)
+ pte_t *pte,
+ struct list_head *compound_pagelist)
{
struct page *page = NULL;
pte_t *_pte;
- int none_or_zero = 0, result = 0, referenced = 0;
+ int none_or_zero = 0, shared = 0, result = 0, referenced = 0;
bool writable = false;
for (_pte = pte; _pte < pte+HPAGE_PMD_NR;
goto out;
}
- /* TODO: teach khugepaged to collapse THP mapped with pte */
- if (PageCompound(page)) {
- result = SCAN_PAGE_COMPOUND;
+ VM_BUG_ON_PAGE(!PageAnon(page), page);
+
+ if (page_mapcount(page) > 1 &&
+ ++shared > khugepaged_max_ptes_shared) {
+ result = SCAN_EXCEED_SHARED_PTE;
goto out;
}
- VM_BUG_ON_PAGE(!PageAnon(page), page);
+ if (PageCompound(page)) {
+ struct page *p;
+ page = compound_head(page);
+
+ /*
+ * Check if we have dealt with the compound page
+ * already
+ */
+ list_for_each_entry(p, compound_pagelist, lru) {
+ if (page == p)
+ goto next;
+ }
+ }
/*
* We can do it before isolate_lru_page because the
}
/*
- * cannot use mapcount: can't collapse if there's a gup pin.
- * The page must only be referenced by the scanned process
- * and page swap cache.
+ * Check if the page has any GUP (or other external) pins.
+ *
+ * The page table that maps the page has been already unlinked
+ * from the page table tree and this process cannot get
+ * an additinal pin on the page.
+ *
+ * New pins can come later if the page is shared across fork,
+ * but not from this process. The other process cannot write to
+ * the page, only trigger CoW.
*/
- if (page_count(page) != 1 + PageSwapCache(page)) {
+ if (!is_refcount_suitable(page)) {
unlock_page(page);
result = SCAN_PAGE_COUNT;
goto out;
}
- if (pte_write(pteval)) {
- writable = true;
- } else {
- if (PageSwapCache(page) &&
- !reuse_swap_page(page, NULL)) {
- unlock_page(page);
- result = SCAN_SWAP_CACHE_PAGE;
- goto out;
- }
+ if (!pte_write(pteval) && PageSwapCache(page) &&
+ !reuse_swap_page(page, NULL)) {
/*
- * Page is not in the swap cache. It can be collapsed
- * into a THP.
+ * Page is in the swap cache and cannot be re-used.
+ * It cannot be collapsed into a THP.
*/
+ unlock_page(page);
+ result = SCAN_SWAP_CACHE_PAGE;
+ goto out;
}
/*
result = SCAN_DEL_PAGE_LRU;
goto out;
}
- inc_node_page_state(page,
- NR_ISOLATED_ANON + page_is_file_lru(page));
+ mod_node_page_state(page_pgdat(page),
+ NR_ISOLATED_ANON + page_is_file_lru(page),
+ compound_nr(page));
VM_BUG_ON_PAGE(!PageLocked(page), page);
VM_BUG_ON_PAGE(PageLRU(page), page);
+ if (PageCompound(page))
+ list_add_tail(&page->lru, compound_pagelist);
+next:
/* There should be enough young pte to collapse the page */
if (pte_young(pteval) ||
page_is_young(page) || PageReferenced(page) ||
mmu_notifier_test_young(vma->vm_mm, address))
referenced++;
+
+ if (pte_write(pteval))
+ writable = true;
}
if (likely(writable)) {
if (likely(referenced)) {
}
out:
- release_pte_pages(pte, _pte);
+ release_pte_pages(pte, _pte, compound_pagelist);
trace_mm_collapse_huge_page_isolate(page, none_or_zero,
referenced, writable, result);
return 0;
static void __collapse_huge_page_copy(pte_t *pte, struct page *page,
struct vm_area_struct *vma,
unsigned long address,
- spinlock_t *ptl)
+ spinlock_t *ptl,
+ struct list_head *compound_pagelist)
{
+ struct page *src_page, *tmp;
pte_t *_pte;
for (_pte = pte; _pte < pte + HPAGE_PMD_NR;
_pte++, page++, address += PAGE_SIZE) {
pte_t pteval = *_pte;
- struct page *src_page;
if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) {
clear_user_highpage(page, address);
} else {
src_page = pte_page(pteval);
copy_user_highpage(page, src_page, address, vma);
- VM_BUG_ON_PAGE(page_mapcount(src_page) != 1, src_page);
- release_pte_page(src_page);
+ if (!PageCompound(src_page))
+ release_pte_page(src_page);
/*
* ptl mostly unnecessary, but preempt has to
* be disabled to update the per-cpu stats
free_page_and_swap_cache(src_page);
}
}
+
+ list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) {
+ list_del(&src_page->lru);
+ release_pte_page(src_page);
+ }
}
static void khugepaged_alloc_sleep(void)
#endif
/*
- * If mmap_sem temporarily dropped, revalidate vma
- * before taking mmap_sem.
+ * If mmap_lock temporarily dropped, revalidate vma
+ * before taking mmap_lock.
* Return 0 if succeeds, otherwise return none-zero
* value (scan code).
*/
* Only done if khugepaged_scan_pmd believes it is worthwhile.
*
* Called and returns without pte mapped or spinlocks held,
- * but with mmap_sem held to protect against vma changes.
+ * but with mmap_lock held to protect against vma changes.
*/
static bool __collapse_huge_page_swapin(struct mm_struct *mm,
swapped_in++;
ret = do_swap_page(&vmf);
- /* do_swap_page returns VM_FAULT_RETRY with released mmap_sem */
+ /* do_swap_page returns VM_FAULT_RETRY with released mmap_lock */
if (ret & VM_FAULT_RETRY) {
- down_read(&mm->mmap_sem);
+ mmap_read_lock(mm);
if (hugepage_vma_revalidate(mm, address, &vmf.vma)) {
/* vma is no longer available, don't continue to swapin */
trace_mm_collapse_huge_page_swapin(mm, swapped_in, referenced, 0);
struct page **hpage,
int node, int referenced, int unmapped)
{
+ LIST_HEAD(compound_pagelist);
pmd_t *pmd, _pmd;
pte_t *pte;
pgtable_t pgtable;
struct page *new_page;
spinlock_t *pmd_ptl, *pte_ptl;
int isolated = 0, result = 0;
- struct mem_cgroup *memcg;
struct vm_area_struct *vma;
struct mmu_notifier_range range;
gfp_t gfp;
gfp = alloc_hugepage_khugepaged_gfpmask() | __GFP_THISNODE;
/*
- * Before allocating the hugepage, release the mmap_sem read lock.
+ * Before allocating the hugepage, release the mmap_lock read lock.
* The allocation can take potentially a long time if it involves
- * sync compaction, and we do not need to hold the mmap_sem during
+ * sync compaction, and we do not need to hold the mmap_lock during
* that. We will recheck the vma after taking it again in write mode.
*/
- up_read(&mm->mmap_sem);
+ mmap_read_unlock(mm);
new_page = khugepaged_alloc_page(hpage, gfp, node);
if (!new_page) {
result = SCAN_ALLOC_HUGE_PAGE_FAIL;
goto out_nolock;
}
- if (unlikely(mem_cgroup_try_charge(new_page, mm, gfp, &memcg, true))) {
+ if (unlikely(mem_cgroup_charge(new_page, mm, gfp))) {
result = SCAN_CGROUP_CHARGE_FAIL;
goto out_nolock;
}
+ count_memcg_page_event(new_page, THP_COLLAPSE_ALLOC);
- down_read(&mm->mmap_sem);
+ mmap_read_lock(mm);
result = hugepage_vma_revalidate(mm, address, &vma);
if (result) {
- mem_cgroup_cancel_charge(new_page, memcg, true);
- up_read(&mm->mmap_sem);
+ mmap_read_unlock(mm);
goto out_nolock;
}
pmd = mm_find_pmd(mm, address);
if (!pmd) {
result = SCAN_PMD_NULL;
- mem_cgroup_cancel_charge(new_page, memcg, true);
- up_read(&mm->mmap_sem);
+ mmap_read_unlock(mm);
goto out_nolock;
}
/*
- * __collapse_huge_page_swapin always returns with mmap_sem locked.
- * If it fails, we release mmap_sem and jump out_nolock.
+ * __collapse_huge_page_swapin always returns with mmap_lock locked.
+ * If it fails, we release mmap_lock and jump out_nolock.
* Continuing to collapse causes inconsistency.
*/
if (unmapped && !__collapse_huge_page_swapin(mm, vma, address,
pmd, referenced)) {
- mem_cgroup_cancel_charge(new_page, memcg, true);
- up_read(&mm->mmap_sem);
+ mmap_read_unlock(mm);
goto out_nolock;
}
- up_read(&mm->mmap_sem);
+ mmap_read_unlock(mm);
/*
* Prevent all access to pagetables with the exception of
* gup_fast later handled by the ptep_clear_flush and the VM
* handled by the anon_vma lock + PG_lock.
*/
- down_write(&mm->mmap_sem);
+ mmap_write_lock(mm);
result = SCAN_ANY_PROCESS;
if (!mmget_still_valid(mm))
goto out;
mmu_notifier_invalidate_range_end(&range);
spin_lock(pte_ptl);
- isolated = __collapse_huge_page_isolate(vma, address, pte);
+ isolated = __collapse_huge_page_isolate(vma, address, pte,
+ &compound_pagelist);
spin_unlock(pte_ptl);
if (unlikely(!isolated)) {
*/
anon_vma_unlock_write(vma->anon_vma);
- __collapse_huge_page_copy(pte, new_page, vma, address, pte_ptl);
+ __collapse_huge_page_copy(pte, new_page, vma, address, pte_ptl,
+ &compound_pagelist);
pte_unmap(pte);
__SetPageUptodate(new_page);
pgtable = pmd_pgtable(_pmd);
spin_lock(pmd_ptl);
BUG_ON(!pmd_none(*pmd));
page_add_new_anon_rmap(new_page, vma, address, true);
- mem_cgroup_commit_charge(new_page, memcg, false, true);
- count_memcg_events(memcg, THP_COLLAPSE_ALLOC, 1);
lru_cache_add_active_or_unevictable(new_page, vma);
pgtable_trans_huge_deposit(mm, pmd, pgtable);
set_pmd_at(mm, address, pmd, _pmd);
khugepaged_pages_collapsed++;
result = SCAN_SUCCEED;
out_up_write:
- up_write(&mm->mmap_sem);
+ mmap_write_unlock(mm);
out_nolock:
+ if (!IS_ERR_OR_NULL(*hpage))
+ mem_cgroup_uncharge(*hpage);
trace_mm_collapse_huge_page(mm, isolated, result);
return;
out:
- mem_cgroup_cancel_charge(new_page, memcg, true);
goto out_up_write;
}
{
pmd_t *pmd;
pte_t *pte, *_pte;
- int ret = 0, none_or_zero = 0, result = 0, referenced = 0;
+ int ret = 0, result = 0, referenced = 0;
+ int none_or_zero = 0, shared = 0;
struct page *page = NULL;
unsigned long _address;
spinlock_t *ptl;
goto out_unmap;
}
- /* TODO: teach khugepaged to collapse THP mapped with pte */
- if (PageCompound(page)) {
- result = SCAN_PAGE_COMPOUND;
+ if (page_mapcount(page) > 1 &&
+ ++shared > khugepaged_max_ptes_shared) {
+ result = SCAN_EXCEED_SHARED_PTE;
goto out_unmap;
}
+ page = compound_head(page);
+
/*
* Record which node the original page is from and save this
* information to khugepaged_node_load[].
}
/*
- * cannot use mapcount: can't collapse if there's a gup pin.
- * The page must only be referenced by the scanned process
- * and page swap cache.
+ * Check if the page has any GUP (or other external) pins.
+ *
+ * Here the check is racy it may see totmal_mapcount > refcount
+ * in some cases.
+ * For example, one process with one forked child process.
+ * The parent has the PMD split due to MADV_DONTNEED, then
+ * the child is trying unmap the whole PMD, but khugepaged
+ * may be scanning the parent between the child has
+ * PageDoubleMap flag cleared and dec the mapcount. So
+ * khugepaged may see total_mapcount > refcount.
+ *
+ * But such case is ephemeral we could always retry collapse
+ * later. However it may report false positive if the page
+ * has excessive GUP pins (i.e. 512). Anyway the same check
+ * will be done again later the risk seems low.
*/
- if (page_count(page) != 1 + PageSwapCache(page)) {
+ if (!is_refcount_suitable(page)) {
result = SCAN_PAGE_COUNT;
goto out_unmap;
}
pte_unmap_unlock(pte, ptl);
if (ret) {
node = khugepaged_find_target_node();
- /* collapse_huge_page will return with the mmap_sem released */
+ /* collapse_huge_page will return with the mmap_lock released */
collapse_huge_page(mm, address, hpage, node,
referenced, unmapped);
}
if (likely(mm_slot->nr_pte_mapped_thp == 0))
return 0;
- if (!down_write_trylock(&mm->mmap_sem))
+ if (!mmap_write_trylock(mm))
return -EBUSY;
if (unlikely(khugepaged_test_exit(mm)))
out:
mm_slot->nr_pte_mapped_thp = 0;
- up_write(&mm->mmap_sem);
+ mmap_write_unlock(mm);
return 0;
}
/*
* Check vma->anon_vma to exclude MAP_PRIVATE mappings that
* got written to. These VMAs are likely not worth investing
- * down_write(mmap_sem) as PMD-mapping is likely to be split
+ * mmap_write_lock(mm) as PMD-mapping is likely to be split
* later.
*
* Not that vma->anon_vma check is racy: it can be set up after
- * the check but before we took mmap_sem by the fault path.
+ * the check but before we took mmap_lock by the fault path.
* But page lock would prevent establishing any new ptes of the
* page, so we are safe.
*
if (!pmd)
continue;
/*
- * We need exclusive mmap_sem to retract page table.
+ * We need exclusive mmap_lock to retract page table.
*
* We use trylock due to lock inversion: we need to acquire
- * mmap_sem while holding page lock. Fault path does it in
+ * mmap_lock while holding page lock. Fault path does it in
* reverse order. Trylock is a way to avoid deadlock.
*/
- if (down_write_trylock(&vma->vm_mm->mmap_sem)) {
+ if (mmap_write_trylock(vma->vm_mm)) {
spinlock_t *ptl = pmd_lock(vma->vm_mm, pmd);
/* assume page table is clear */
_pmd = pmdp_collapse_flush(vma, addr, pmd);
spin_unlock(ptl);
- up_write(&vma->vm_mm->mmap_sem);
+ mmap_write_unlock(vma->vm_mm);
mm_dec_nr_ptes(vma->vm_mm);
pte_free(vma->vm_mm, pmd_pgtable(_pmd));
} else {
struct address_space *mapping = file->f_mapping;
gfp_t gfp;
struct page *new_page;
- struct mem_cgroup *memcg;
pgoff_t index, end = start + HPAGE_PMD_NR;
LIST_HEAD(pagelist);
XA_STATE_ORDER(xas, &mapping->i_pages, start, HPAGE_PMD_ORDER);
goto out;
}
- if (unlikely(mem_cgroup_try_charge(new_page, mm, gfp, &memcg, true))) {
+ if (unlikely(mem_cgroup_charge(new_page, mm, gfp))) {
result = SCAN_CGROUP_CHARGE_FAIL;
goto out;
}
+ count_memcg_page_event(new_page, THP_COLLAPSE_ALLOC);
/* This will be less messy when we use multi-index entries */
do {
break;
xas_unlock_irq(&xas);
if (!xas_nomem(&xas, GFP_KERNEL)) {
- mem_cgroup_cancel_charge(new_page, memcg, true);
result = SCAN_FAIL;
goto out;
}
}
if (nr_none) {
- struct zone *zone = page_zone(new_page);
-
- __mod_node_page_state(zone->zone_pgdat, NR_FILE_PAGES, nr_none);
+ __mod_lruvec_page_state(new_page, NR_FILE_PAGES, nr_none);
if (is_shmem)
- __mod_node_page_state(zone->zone_pgdat,
- NR_SHMEM, nr_none);
+ __mod_lruvec_page_state(new_page, NR_SHMEM, nr_none);
}
xa_locked:
SetPageUptodate(new_page);
page_ref_add(new_page, HPAGE_PMD_NR - 1);
- mem_cgroup_commit_charge(new_page, memcg, false, true);
-
- if (is_shmem) {
+ if (is_shmem)
set_page_dirty(new_page);
- lru_cache_add_anon(new_page);
- } else {
- lru_cache_add_file(new_page);
- }
- count_memcg_events(memcg, THP_COLLAPSE_ALLOC, 1);
+ lru_cache_add(new_page);
/*
* Remove pte page tables, so we can re-fault the page as huge.
VM_BUG_ON(nr_none);
xas_unlock_irq(&xas);
- mem_cgroup_cancel_charge(new_page, memcg, true);
new_page->mapping = NULL;
}
unlock_page(new_page);
out:
VM_BUG_ON(!list_empty(&pagelist));
+ if (!IS_ERR_OR_NULL(*hpage))
+ mem_cgroup_uncharge(*hpage);
/* TODO: tracepoints */
}
* the next mm on the list.
*/
vma = NULL;
- if (unlikely(!down_read_trylock(&mm->mmap_sem)))
- goto breakouterloop_mmap_sem;
+ if (unlikely(!mmap_read_trylock(mm)))
+ goto breakouterloop_mmap_lock;
if (likely(!khugepaged_test_exit(mm)))
vma = find_vma(mm, khugepaged_scan.address);
pgoff_t pgoff = linear_page_index(vma,
khugepaged_scan.address);
- up_read(&mm->mmap_sem);
+ mmap_read_unlock(mm);
ret = 1;
khugepaged_scan_file(mm, file, pgoff, hpage);
fput(file);
khugepaged_scan.address += HPAGE_PMD_SIZE;
progress += HPAGE_PMD_NR;
if (ret)
- /* we released mmap_sem so break loop */
- goto breakouterloop_mmap_sem;
+ /* we released mmap_lock so break loop */
+ goto breakouterloop_mmap_lock;
if (progress >= pages)
goto breakouterloop;
}
}
breakouterloop:
- up_read(&mm->mmap_sem); /* exit_mmap will destroy ptes after this */
-breakouterloop_mmap_sem:
+ mmap_read_unlock(mm); /* exit_mmap will destroy ptes after this */
+breakouterloop_mmap_lock:
spin_lock(&khugepaged_mm_lock);
VM_BUG_ON(khugepaged_scan.mm_slot != mm_slot);