mm: make compound_head() robust

Hugh has pointed that compound_head() call can be unsafe in some
context. There's one example:

	CPU0					CPU1

isolate_migratepages_block()
  page_count()
    compound_head()
      !!PageTail() == true
					put_page()
					  tail->first_page = NULL
      head = tail->first_page
					alloc_pages(__GFP_COMP)
					   prep_compound_page()
					     tail->first_page = head
					     __SetPageTail(p);
      !!PageTail() == true
    <head == NULL dereferencing>

The race is pure theoretical. I don't it's possible to trigger it in
practice. But who knows.

We can fix the race by changing how encode PageTail() and compound_head()
within struct page to be able to update them in one shot.

The patch introduces page->compound_head into third double word block in
front of compound_dtor and compound_order. Bit 0 encodes PageTail() and
the rest bits are pointer to head page if bit zero is set.

The patch moves page->pmd_huge_pte out of word, just in case if an
architecture defines pgtable_t into something what can have the bit 0
set.

hugetlb_cgroup uses page->lru.next in the second tail page to store
pointer struct hugetlb_cgroup. The patch switch it to use page->private
in the second tail page instead. The space is free since ->first_page is
removed from the union.

The patch also opens possibility to remove HUGETLB_CGROUP_MIN_ORDER
limitation, since there's now space in first tail page to store struct
hugetlb_cgroup pointer. But that's out of scope of the patch.

That means page->compound_head shares storage space with:

 - page->lru.next;
 - page->next;
 - page->rcu_head.next;

That's too long list to be absolutely sure, but looks like nobody uses
bit 0 of the word.

page->rcu_head.next guaranteed[1] to have bit 0 clean as long as we use
call_rcu(), call_rcu_bh(), call_rcu_sched(), or call_srcu(). But future
call_rcu_lazy() is not allowed as it makes use of the bit and we can
get false positive PageTail().

[1] http://lkml.kernel.org/g/20150827163634.GD4029@linux.vnet.ibm.com

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This commit is contained in:
Kirill A. Shutemov 2015-11-06 16:29:54 -08:00 committed by Linus Torvalds
parent f1e61557f0
commit 1d798ca3f1
15 changed files with 89 additions and 182 deletions

View file

@ -54,8 +54,8 @@ everything required is done by pgtable_page_ctor() and pgtable_page_dtor(),
which must be called on PTE table allocation / freeing.
Make sure the architecture doesn't use slab allocator for page table
allocation: slab uses page->slab_cache and page->first_page for its pages.
These fields share storage with page->ptl.
allocation: slab uses page->slab_cache for its pages.
This field shares storage with page->ptl.
PMD split lock only makes sense if you have more than two page table
levels.

View file

@ -169,7 +169,6 @@ CONFIG_FLATMEM_MANUAL=y
# CONFIG_SPARSEMEM_MANUAL is not set
CONFIG_FLATMEM=y
CONFIG_FLAT_NODE_MEM_MAP=y
CONFIG_PAGEFLAGS_EXTENDED=y
CONFIG_SPLIT_PTLOCK_CPUS=4
# CONFIG_PHYS_ADDR_T_64BIT is not set
CONFIG_ZONE_DMA_FLAG=1

View file

@ -32,7 +32,7 @@ static inline struct hugetlb_cgroup *hugetlb_cgroup_from_page(struct page *page)
if (compound_order(page) < HUGETLB_CGROUP_MIN_ORDER)
return NULL;
return (struct hugetlb_cgroup *)page[2].lru.next;
return (struct hugetlb_cgroup *)page[2].private;
}
static inline
@ -42,7 +42,7 @@ int set_hugetlb_cgroup(struct page *page, struct hugetlb_cgroup *h_cg)
if (compound_order(page) < HUGETLB_CGROUP_MIN_ORDER)
return -1;
page[2].lru.next = (void *)h_cg;
page[2].private = (unsigned long)h_cg;
return 0;
}

View file

@ -430,46 +430,6 @@ static inline void compound_unlock_irqrestore(struct page *page,
#endif
}
static inline struct page *compound_head_by_tail(struct page *tail)
{
struct page *head = tail->first_page;
/*
* page->first_page may be a dangling pointer to an old
* compound page, so recheck that it is still a tail
* page before returning.
*/
smp_rmb();
if (likely(PageTail(tail)))
return head;
return tail;
}
/*
* Since either compound page could be dismantled asynchronously in THP
* or we access asynchronously arbitrary positioned struct page, there
* would be tail flag race. To handle this race, we should call
* smp_rmb() before checking tail flag. compound_head_by_tail() did it.
*/
static inline struct page *compound_head(struct page *page)
{
if (unlikely(PageTail(page)))
return compound_head_by_tail(page);
return page;
}
/*
* If we access compound page synchronously such as access to
* allocated page, there is no need to handle tail flag race, so we can
* check tail flag directly without any synchronization primitive.
*/
static inline struct page *compound_head_fast(struct page *page)
{
if (unlikely(PageTail(page)))
return page->first_page;
return page;
}
/*
* The atomic page->_mapcount, starts from -1: so that transitions
* both from it and to it can be tracked, using atomic_inc_and_test
@ -518,7 +478,7 @@ static inline void get_huge_page_tail(struct page *page)
VM_BUG_ON_PAGE(!PageTail(page), page);
VM_BUG_ON_PAGE(page_mapcount(page) < 0, page);
VM_BUG_ON_PAGE(atomic_read(&page->_count) != 0, page);
if (compound_tail_refcounted(page->first_page))
if (compound_tail_refcounted(compound_head(page)))
atomic_inc(&page->_mapcount);
}
@ -541,13 +501,7 @@ static inline struct page *virt_to_head_page(const void *x)
{
struct page *page = virt_to_page(x);
/*
* We don't need to worry about synchronization of tail flag
* when we call virt_to_head_page() since it is only called for
* already allocated page and this page won't be freed until
* this virt_to_head_page() is finished. So use _fast variant.
*/
return compound_head_fast(page);
return compound_head(page);
}
/*
@ -1586,8 +1540,7 @@ static inline bool ptlock_init(struct page *page)
* with 0. Make sure nobody took it in use in between.
*
* It can happen if arch try to use slab for page table allocation:
* slab code uses page->slab_cache and page->first_page (for tail
* pages), which share storage with page->ptl.
* slab code uses page->slab_cache, which share storage with page->ptl.
*/
VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page);
if (!ptlock_alloc(page))

View file

@ -111,7 +111,13 @@ struct page {
};
};
/* Third double word block */
/*
* Third double word block
*
* WARNING: bit 0 of the first word encode PageTail(). That means
* the rest users of the storage space MUST NOT use the bit to
* avoid collision and false-positive PageTail().
*/
union {
struct list_head lru; /* Pageout list, eg. active_list
* protected by zone->lru_lock !
@ -132,14 +138,23 @@ struct page {
struct rcu_head rcu_head; /* Used by SLAB
* when destroying via RCU
*/
/* First tail page of compound page */
/* Tail pages of compound page */
struct {
unsigned long compound_head; /* If bit zero is set */
/* First tail page only */
unsigned short int compound_dtor;
unsigned short int compound_order;
};
#if defined(CONFIG_TRANSPARENT_HUGEPAGE) && USE_SPLIT_PMD_PTLOCKS
pgtable_t pmd_huge_pte; /* protected by page->ptl */
struct {
unsigned long __pad; /* do not overlay pmd_huge_pte
* with compound_head to avoid
* possible bit 0 collision.
*/
pgtable_t pmd_huge_pte; /* protected by page->ptl */
};
#endif
};
@ -160,7 +175,6 @@ struct page {
#endif
#endif
struct kmem_cache *slab_cache; /* SL[AU]B: Pointer to slab */
struct page *first_page; /* Compound tail pages */
};
#ifdef CONFIG_MEMCG

View file

@ -86,12 +86,7 @@ enum pageflags {
PG_private, /* If pagecache, has fs-private data */
PG_private_2, /* If pagecache, has fs aux data */
PG_writeback, /* Page is under writeback */
#ifdef CONFIG_PAGEFLAGS_EXTENDED
PG_head, /* A head page */
PG_tail, /* A tail page */
#else
PG_compound, /* A compound page */
#endif
PG_swapcache, /* Swap page: swp_entry_t in private */
PG_mappedtodisk, /* Has blocks allocated on-disk */
PG_reclaim, /* To be reclaimed asap */
@ -398,21 +393,35 @@ static inline void set_page_writeback_keepwrite(struct page *page)
test_set_page_writeback_keepwrite(page);
}
#ifdef CONFIG_PAGEFLAGS_EXTENDED
/*
* System with lots of page flags available. This allows separate
* flags for PageHead() and PageTail() checks of compound pages so that bit
* tests can be used in performance sensitive paths. PageCompound is
* generally not used in hot code paths except arch/powerpc/mm/init_64.c
* and arch/powerpc/kvm/book3s_64_vio_hv.c which use it to detect huge pages
* and avoid handling those in real mode.
*/
__PAGEFLAG(Head, head) CLEARPAGEFLAG(Head, head)
__PAGEFLAG(Tail, tail)
static inline int PageTail(struct page *page)
{
return READ_ONCE(page->compound_head) & 1;
}
static inline void set_compound_head(struct page *page, struct page *head)
{
WRITE_ONCE(page->compound_head, (unsigned long)head + 1);
}
static inline void clear_compound_head(struct page *page)
{
WRITE_ONCE(page->compound_head, 0);
}
static inline struct page *compound_head(struct page *page)
{
unsigned long head = READ_ONCE(page->compound_head);
if (unlikely(head & 1))
return (struct page *) (head - 1);
return page;
}
static inline int PageCompound(struct page *page)
{
return page->flags & ((1L << PG_head) | (1L << PG_tail));
return PageHead(page) || PageTail(page);
}
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
@ -425,59 +434,6 @@ static inline void ClearPageCompound(struct page *page)
#define PG_head_mask ((1L << PG_head))
#else
/*
* Reduce page flag use as much as possible by overlapping
* compound page flags with the flags used for page cache pages. Possible
* because PageCompound is always set for compound pages and not for
* pages on the LRU and/or pagecache.
*/
TESTPAGEFLAG(Compound, compound)
__SETPAGEFLAG(Head, compound) __CLEARPAGEFLAG(Head, compound)
/*
* PG_reclaim is used in combination with PG_compound to mark the
* head and tail of a compound page. This saves one page flag
* but makes it impossible to use compound pages for the page cache.
* The PG_reclaim bit would have to be used for reclaim or readahead
* if compound pages enter the page cache.
*
* PG_compound & PG_reclaim => Tail page
* PG_compound & ~PG_reclaim => Head page
*/
#define PG_head_mask ((1L << PG_compound))
#define PG_head_tail_mask ((1L << PG_compound) | (1L << PG_reclaim))
static inline int PageHead(struct page *page)
{
return ((page->flags & PG_head_tail_mask) == PG_head_mask);
}
static inline int PageTail(struct page *page)
{
return ((page->flags & PG_head_tail_mask) == PG_head_tail_mask);
}
static inline void __SetPageTail(struct page *page)
{
page->flags |= PG_head_tail_mask;
}
static inline void __ClearPageTail(struct page *page)
{
page->flags &= ~PG_head_tail_mask;
}
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
static inline void ClearPageCompound(struct page *page)
{
BUG_ON((page->flags & PG_head_tail_mask) != (1 << PG_compound));
clear_bit(PG_compound, &page->flags);
}
#endif
#endif /* !PAGEFLAGS_EXTENDED */
#ifdef CONFIG_HUGETLB_PAGE
int PageHuge(struct page *page);
int PageHeadHuge(struct page *page);

View file

@ -200,18 +200,6 @@ config MEMORY_HOTREMOVE
depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE
depends on MIGRATION
#
# If we have space for more page flags then we can enable additional
# optimizations and functionality.
#
# Regular Sparsemem takes page flag bits for the sectionid if it does not
# use a virtual memmap. Disable extended page flags for 32 bit platforms
# that require the use of a sectionid in the page flags.
#
config PAGEFLAGS_EXTENDED
def_bool y
depends on 64BIT || SPARSEMEM_VMEMMAP || !SPARSEMEM
# Heavily threaded applications may benefit from splitting the mm-wide
# page_table_lock, so that faults on different parts of the user address
# space can be handled with less contention: split it at this NR_CPUS.

View file

@ -25,12 +25,7 @@ static const struct trace_print_flags pageflag_names[] = {
{1UL << PG_private, "private" },
{1UL << PG_private_2, "private_2" },
{1UL << PG_writeback, "writeback" },
#ifdef CONFIG_PAGEFLAGS_EXTENDED
{1UL << PG_head, "head" },
{1UL << PG_tail, "tail" },
#else
{1UL << PG_compound, "compound" },
#endif
{1UL << PG_swapcache, "swapcache" },
{1UL << PG_mappedtodisk, "mappedtodisk" },
{1UL << PG_reclaim, "reclaim" },

View file

@ -1755,8 +1755,7 @@ static void __split_huge_page_refcount(struct page *page,
(1L << PG_unevictable)));
page_tail->flags |= (1L << PG_dirty);
/* clear PageTail before overwriting first_page */
smp_wmb();
clear_compound_head(page_tail);
if (page_is_young(page))
set_page_young(page_tail);

View file

@ -1001,9 +1001,8 @@ static void destroy_compound_gigantic_page(struct page *page,
struct page *p = page + 1;
for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) {
__ClearPageTail(p);
clear_compound_head(p);
set_page_refcounted(p);
p->first_page = NULL;
}
set_compound_order(page, 0);
@ -1276,10 +1275,7 @@ static void prep_compound_gigantic_page(struct page *page, unsigned long order)
*/
__ClearPageReserved(p);
set_page_count(p, 0);
p->first_page = page;
/* Make sure p->first_page is always valid for PageTail() */
smp_wmb();
__SetPageTail(p);
set_compound_head(p, page);
}
}

View file

@ -385,7 +385,7 @@ void __init hugetlb_cgroup_file_init(void)
/*
* Add cgroup control files only if the huge page consists
* of more than two normal pages. This is because we use
* page[2].lru.next for storing cgroup details.
* page[2].private for storing cgroup details.
*/
if (huge_page_order(h) >= HUGETLB_CGROUP_MIN_ORDER)
__hugetlb_cgroup_file_init(hstate_index(h));

View file

@ -80,9 +80,9 @@ static inline void __get_page_tail_foll(struct page *page,
* speculative page access (like in
* page_cache_get_speculative()) on tail pages.
*/
VM_BUG_ON_PAGE(atomic_read(&page->first_page->_count) <= 0, page);
VM_BUG_ON_PAGE(atomic_read(&compound_head(page)->_count) <= 0, page);
if (get_page_head)
atomic_inc(&page->first_page->_count);
atomic_inc(&compound_head(page)->_count);
get_huge_page_tail(page);
}

View file

@ -776,8 +776,6 @@ static int me_huge_page(struct page *p, unsigned long pfn)
#define lru (1UL << PG_lru)
#define swapbacked (1UL << PG_swapbacked)
#define head (1UL << PG_head)
#define tail (1UL << PG_tail)
#define compound (1UL << PG_compound)
#define slab (1UL << PG_slab)
#define reserved (1UL << PG_reserved)
@ -800,12 +798,7 @@ static struct page_state {
*/
{ slab, slab, MF_MSG_SLAB, me_kernel },
#ifdef CONFIG_PAGEFLAGS_EXTENDED
{ head, head, MF_MSG_HUGE, me_huge_page },
{ tail, tail, MF_MSG_HUGE, me_huge_page },
#else
{ compound, compound, MF_MSG_HUGE, me_huge_page },
#endif
{ sc|dirty, sc|dirty, MF_MSG_DIRTY_SWAPCACHE, me_swapcache_dirty },
{ sc|dirty, sc, MF_MSG_CLEAN_SWAPCACHE, me_swapcache_clean },

View file

@ -445,15 +445,15 @@ static void bad_page(struct page *page, const char *reason,
/*
* Higher-order pages are called "compound pages". They are structured thusly:
*
* The first PAGE_SIZE page is called the "head page".
* The first PAGE_SIZE page is called the "head page" and have PG_head set.
*
* The remaining PAGE_SIZE pages are called "tail pages".
* The remaining PAGE_SIZE pages are called "tail pages". PageTail() is encoded
* in bit 0 of page->compound_head. The rest of bits is pointer to head page.
*
* All pages have PG_compound set. All tail pages have their ->first_page
* pointing at the head page.
* The first tail page's ->compound_dtor holds the offset in array of compound
* page destructors. See compound_page_dtors.
*
* The first tail page's ->lru.next holds the address of the compound page's
* put_page() function. Its ->lru.prev holds the order of allocation.
* The first tail page's ->compound_order holds the order of allocation.
* This usage means that zero-order pages may not be compound.
*/
@ -473,10 +473,7 @@ void prep_compound_page(struct page *page, unsigned long order)
for (i = 1; i < nr_pages; i++) {
struct page *p = page + i;
set_page_count(p, 0);
p->first_page = page;
/* Make sure p->first_page is always valid for PageTail() */
smp_wmb();
__SetPageTail(p);
set_compound_head(p, page);
}
}
@ -854,17 +851,30 @@ static void free_one_page(struct zone *zone,
static int free_tail_pages_check(struct page *head_page, struct page *page)
{
if (!IS_ENABLED(CONFIG_DEBUG_VM))
return 0;
int ret = 1;
/*
* We rely page->lru.next never has bit 0 set, unless the page
* is PageTail(). Let's make sure that's true even for poisoned ->lru.
*/
BUILD_BUG_ON((unsigned long)LIST_POISON1 & 1);
if (!IS_ENABLED(CONFIG_DEBUG_VM)) {
ret = 0;
goto out;
}
if (unlikely(!PageTail(page))) {
bad_page(page, "PageTail not set", 0);
return 1;
goto out;
}
if (unlikely(page->first_page != head_page)) {
bad_page(page, "first_page not consistent", 0);
return 1;
if (unlikely(compound_head(page) != head_page)) {
bad_page(page, "compound_head not consistent", 0);
goto out;
}
return 0;
ret = 0;
out:
clear_compound_head(page);
return ret;
}
static void __meminit __init_single_page(struct page *page, unsigned long pfn,
@ -931,6 +941,10 @@ void __meminit reserve_bootmem_region(unsigned long start, unsigned long end)
struct page *page = pfn_to_page(start_pfn);
init_reserved_page(start_pfn);
/* Avoid false-positive PageTail() */
INIT_LIST_HEAD(&page->lru);
SetPageReserved(page);
}
}

View file

@ -201,7 +201,7 @@ void put_refcounted_compound_page(struct page *page_head, struct page *page)
__put_single_page(page);
return;
}
VM_BUG_ON_PAGE(page_head != page->first_page, page);
VM_BUG_ON_PAGE(page_head != compound_head(page), page);
/*
* We can release the refcount taken by
* get_page_unless_zero() now that
@ -262,7 +262,7 @@ static void put_compound_page(struct page *page)
* Case 3 is possible, as we may race with
* __split_huge_page_refcount tearing down a THP page.
*/
page_head = compound_head_by_tail(page);
page_head = compound_head(page);
if (!__compound_tail_refcounted(page_head))
put_unrefcounted_compound_page(page_head, page);
else