Skip to contentSkip to conversation

mm/slab: take n->list_lock in __slab_try_return_freelist() to avoid race

PATCH Superseded Harry Yoo (Meta) 2 participants 3 messages Sep 2 – Sep 3

Subsystems (2)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Harry Yoo (Meta)

author maintainer
#0 lore raw

PATCH v2 1/1 mm/slab: take n->list_lock in __slab_try_return_freelist() to avoid race

Commit ba7425312607 ("mm, slab: add an optimistic __slab_try_return_freelist()") incorrectly assumed that nobody has freed an object to the slab as long as slab->freelist is NULL and cmpxchg succeeds.

However, as reported by Hyunwoo Kim [1], other CPUs might have freed an object to the slab, insert the slab to the partial list, then allocated an object from the slab, and be in the middle of removing the slab from the list under n->list_lock.

Since __refill_objects_node() puts the slab back on pc.slabs outside n->list_lock, it might insert the slab into that list while the slab is concurrently being removed from n->partial. This led to a list corruption, as reported by Hyunwoo Kim [1]:

list_add corruption. next->prev should be prev
(ffff888100000248), but was dead000000000122.
(next=ffffea000416e410).
kernel BUG at lib/list_debug.c:29!

  Oops: invalid opcode: 0000 [#1] SMP NOPTI
  CPU: 1 UID: 65534 PID: 144 Comm: poc Not tainted
  7.2.0-16172-gcf72cbb39da8-dirty #1 PREEMPT(lazy)
  RIP: 0010:__list_add_valid_or_report+0x80/0xd0
  ...
  Call Trace:
   alloc_from_new_slab+0x183/0x300
   ___slab_alloc+0x31c/0x890
   __kmalloc_noprof+0x3d4/0x800
   lsm_blob_alloc+0x2d/0x50
   security_msg_msg_alloc+0x26/0x90
   load_msg+0x1aa/0x210
   do_msgsnd+0x91/0x800
   do_syscall_64+0x109/0x5d0
   entry_SYSCALL_64_after_hwframe+0x77/0x7f
  ...
  Kernel panic - not syncing: Fatal exception

This is a classic ABA problem where cmpxchg succeeds but the state has changed since __refill_objects_node() took the freelist from the slab.

As Vlastimil Babka mentioned [2], it should be rare to return more than one slab (due to the racy read of slab->counters in get_partial_node_bulk()). Therefore, instead of introducing additional complexity, acquire and release n->list_lock twice in the worst case.

Return the slab directly to the partial list and hold n->list_lock across the cmpxchg and add_partial(). This is similar to the initial version of commit ba7425312607 [3]. This is enough to avoid the race as the list manipulation is serialized by n->list_lock. While at it, bring back unlikely() hint now that the condition is unlikely.

---
Changes in v2:
- EDITME: describe what is new in this series revision.
- EDITME: use bulletpoints and terse descriptions.
- Link to v1: https://lore.kernel.org/r/apPa-cGLcyt90l-E@v4bel

---
 mm/slub.c | 17 ++++++++++++-----
 1 file changed, 12 insertions(+), 5 deletions(-)
mm/slub.c +12 -5
index f9b56cb439e7..5cbbacb8ee32 100644
@@ -5684,6 +5684,8 @@ static bool __slab_try_return_freelist(struct kmem_cache *s, struct slab *slab,
 				       void *head, int cnt)
 {
 	struct freelist_counters old, new;
+	struct kmem_cache_node *n;
+	unsigned long flags;
 
 	old.freelist = slab->freelist;
 	old.counters = slab->counters;
@@ -5695,9 +5697,16 @@ static bool __slab_try_return_freelist(struct kmem_cache *s, struct slab *slab,
 	new.counters = old.counters;
 	new.inuse -= cnt;
 
-	if (!slab_update_freelist(s, slab, &old, &new, "__slab_try_return_freelist"))
+	n = get_node(s, slab_nid(slab));
+	spin_lock_irqsave(&n->list_lock, flags);
+
+	if (!slab_update_freelist(s, slab, &old, &new, "__slab_try_return_freelist")) {
+		spin_unlock_irqrestore(&n->list_lock, flags);
 		return false;
+	}
 
+	add_partial(n, slab, ADD_TO_TAIL);
+	spin_unlock_irqrestore(&n->list_lock, flags);
 	return true;
 }
 
@@ -7296,10 +7305,8 @@ __refill_objects_node(struct kmem_cache *s, void **p, gfp_t gfp, unsigned int mi
 			void *head = object;
 			void *tail;
 
-			if (__slab_try_return_freelist(s, slab, head, count)) {
-				list_add(&slab->slab_list, &pc.slabs);
+			if (__slab_try_return_freelist(s, slab, head, count))
 				break;
-			}
 
 			do {
 				tail = object;
@@ -7312,7 +7319,7 @@ __refill_objects_node(struct kmem_cache *s, void **p, gfp_t gfp, unsigned int mi
 			break;
 	}
 
-	if (!list_empty(&pc.slabs)) {
+	if (unlikely(!list_empty(&pc.slabs))) {
 		spin_lock_irqsave(&n->list_lock, flags);
 
 		list_for_each_entry(slab, &pc.slabs, slab_list)

---
base-commit: cee9395acd8043be0644b25c34bfa86623f2b935
change-id: 20260902-slab-fix-aba-d60f39fc9dee

Best regards,

Signature · 2 lines
Cheers,
Harry / Hyeonggon
#1 lore raw
Quoted 10 lines · Harry Yoo (Meta)

Commit ba7425312607 ("mm, slab: add an optimistic __slab_try_return_freelist()") incorrectly assumed that nobody has freed an object to the slab as long as slab->freelist is NULL and cmpxchg succeeds.

However, as reported by Hyunwoo Kim [1], other CPUs might have freed an object to the slab, insert the slab to the partial list, then allocated an object from the slab, and be in the middle of removing the slab from the list under n->list_lock.

Quoted 4 lines

Since __refill_objects_node() puts the slab back on pc.slabs outside n->list_lock, it might insert the slab into that list while the slab is concurrently being removed from n->partial. This led to a list corruption, as reported by Hyunwoo Kim [1]:

"as reported by ..." was already said above, we could keep just the [1] or drop completely?

Quoted 30 lines

list_add corruption. next->prev should be prev
(ffff888100000248), but was dead000000000122.
(next=ffffea000416e410).
kernel BUG at lib/list_debug.c:29!

  Oops: invalid opcode: 0000 [#1] SMP NOPTI
  CPU: 1 UID: 65534 PID: 144 Comm: poc Not tainted
  7.2.0-16172-gcf72cbb39da8-dirty #1 PREEMPT(lazy)
  RIP: 0010:__list_add_valid_or_report+0x80/0xd0
  ...
  Call Trace:
   alloc_from_new_slab+0x183/0x300
   ___slab_alloc+0x31c/0x890
   __kmalloc_noprof+0x3d4/0x800
   lsm_blob_alloc+0x2d/0x50
   security_msg_msg_alloc+0x26/0x90
   load_msg+0x1aa/0x210
   do_msgsnd+0x91/0x800
   do_syscall_64+0x109/0x5d0
   entry_SYSCALL_64_after_hwframe+0x77/0x7f
  ...
  Kernel panic - not syncing: Fatal exception

This is a classic ABA problem where cmpxchg succeeds but the state has changed since __refill_objects_node() took the freelist from the slab.

As Vlastimil Babka mentioned [2], it should be rare to return more than one slab (due to the racy read of slab->counters in get_partial_node_bulk()). Therefore, instead of introducing additional complexity, acquire and release n->list_lock twice in the worst case.

Quoted 5 lines

Return the slab directly to the partial list and hold n->list_lock across the cmpxchg and add_partial(). This is similar to the initial version of commit ba7425312607 [3]. This is enough to avoid the race as the list manipulation is serialized by n->list_lock. While at it, bring back unlikely() hint now that the condition is unlikely.

Ah nicely spotted.

Quoted 30 lines

---
Changes in v2:
- EDITME: describe what is new in this series revision.
- EDITME: use bulletpoints and terse descriptions.
- Link to v1: https://lore.kernel.org/r/apPa-cGLcyt90l-E@v4bel

---
 mm/slub.c | 17 ++++++++++++-----
 1 file changed, 12 insertions(+), 5 deletions(-)
mm/slub.c +2 -0
index f9b56cb439e7..5cbbacb8ee32 100644
@@ -5684,6 +5684,8 @@ static bool __slab_try_return_freelist(struct kmem_cache *s, struct slab *slab,
 				       void *head, int cnt)
 {
 	struct freelist_counters old, new;
+	struct kmem_cache_node *n;
+	unsigned long flags;
 
 	old.freelist = slab->freelist;
 	old.counters = slab->counters;

mm/slub.c static bool __slab_try_return_freelist(struct kmem_cache *s, struct slab *slab, · line 5700

Quoted 5 lines
mm/slub.c +1 -1
@@ -5695,9 +5697,16 @@ static bool __slab_try_return_freelist(struct kmem_cache *s, struct slab *slab,
 	new.counters = old.counters;
 	new.inuse -= cnt;
 
-	if (!slab_update_freelist(s, slab, &old, &new, "__slab_try_return_freelist"))
+	n = get_node(s, slab_nid(slab));

Wonder if we could just pass the 'n' to __slab_try_return_freelist() that we already have in the caller.

Quoted 39 lines
mm/slub.c +9 -5
+	spin_lock_irqsave(&n->list_lock, flags);
+
+	if (!slab_update_freelist(s, slab, &old, &new, "__slab_try_return_freelist")) {
+		spin_unlock_irqrestore(&n->list_lock, flags);
 		return false;
+	}
 
+	add_partial(n, slab, ADD_TO_TAIL);
+	spin_unlock_irqrestore(&n->list_lock, flags);
 	return true;
 }
 
@@ -7296,10 +7305,8 @@ __refill_objects_node(struct kmem_cache *s, void **p, gfp_t gfp, unsigned int mi
 			void *head = object;
 			void *tail;
 
-			if (__slab_try_return_freelist(s, slab, head, count)) {
-				list_add(&slab->slab_list, &pc.slabs);
+			if (__slab_try_return_freelist(s, slab, head, count))
 				break;
-			}
 
 			do {
 				tail = object;
@@ -7312,7 +7319,7 @@ __refill_objects_node(struct kmem_cache *s, void **p, gfp_t gfp, unsigned int mi
 			break;
 	}
 
-	if (!list_empty(&pc.slabs)) {
+	if (unlikely(!list_empty(&pc.slabs))) {
 		spin_lock_irqsave(&n->list_lock, flags);
 
 		list_for_each_entry(slab, &pc.slabs, slab_list)
 
---

base-commit: cee9395acd8043be0644b25c34bfa86623f2b935
change-id: 20260902-slab-fix-aba-d60f39fc9dee

Best regards,

author maintainer
#2 lore raw
Quoted 11 lines · Vlastimil Babka (SUSE)
Quoted 10 lines · Harry Yoo (Meta)

Commit ba7425312607 ("mm, slab: add an optimistic __slab_try_return_freelist()") incorrectly assumed that nobody has freed an object to the slab as long as slab->freelist is NULL and cmpxchg succeeds.

However, as reported by Hyunwoo Kim [1], other CPUs might have freed an object to the slab, insert the slab to the partial list, then allocated an object from the slab, and be in the middle of removing the slab from the list under n->list_lock.

Quoted 7 lines
Quoted 4 lines

Since __refill_objects_node() puts the slab back on pc.slabs outside n->list_lock, it might insert the slab into that list while the slab is concurrently being removed from n->partial. This led to a list corruption, as reported by Hyunwoo Kim [1]:

"as reported by ..." was already said above, we could keep just the [1] or drop completely?

I didn't want to steal the credit for splat ;) but yeah keeping [1] should be enough.

Quoted 30 lines
Quoted 30 lines

list_add corruption. next->prev should be prev
(ffff888100000248), but was dead000000000122.
(next=ffffea000416e410).
kernel BUG at lib/list_debug.c:29!

  Oops: invalid opcode: 0000 [#1] SMP NOPTI
  CPU: 1 UID: 65534 PID: 144 Comm: poc Not tainted
  7.2.0-16172-gcf72cbb39da8-dirty #1 PREEMPT(lazy)
  RIP: 0010:__list_add_valid_or_report+0x80/0xd0
  ...
  Call Trace:
   alloc_from_new_slab+0x183/0x300
   ___slab_alloc+0x31c/0x890
   __kmalloc_noprof+0x3d4/0x800
   lsm_blob_alloc+0x2d/0x50
   security_msg_msg_alloc+0x26/0x90
   load_msg+0x1aa/0x210
   do_msgsnd+0x91/0x800
   do_syscall_64+0x109/0x5d0
   entry_SYSCALL_64_after_hwframe+0x77/0x7f
  ...
  Kernel panic - not syncing: Fatal exception

This is a classic ABA problem where cmpxchg succeeds but the state has changed since __refill_objects_node() took the freelist from the slab.

As Vlastimil Babka mentioned [2], it should be rare to return more than one slab (due to the racy read of slab->counters in get_partial_node_bulk()). Therefore, instead of introducing additional complexity, acquire and release n->list_lock twice in the worst case.

Quoted 7 lines
Quoted 5 lines

Return the slab directly to the partial list and hold n->list_lock across the cmpxchg and add_partial(). This is similar to the initial version of commit ba7425312607 [3]. This is enough to avoid the race as the list manipulation is serialized by n->list_lock. While at it, bring back unlikely() hint now that the condition is unlikely.

Ah nicely spotted.

Thanks ;-)

Quoted 17 lines
Quoted 17 lines
mm/slub.c +2 -0
index f9b56cb439e7..5cbbacb8ee32 100644
@@ -5684,6 +5684,8 @@ static bool __slab_try_return_freelist(struct kmem_cache *s, struct slab *slab,
 				       void *head, int cnt)
 {
 	struct freelist_counters old, new;
+	struct kmem_cache_node *n;
+	unsigned long flags;
 
 	old.freelist = slab->freelist;
 	old.counters = slab->counters;
@@ -5695,9 +5697,16 @@ static bool __slab_try_return_freelist(struct kmem_cache *s, struct slab *slab,
 	new.counters = old.counters;
 	new.inuse -= cnt;
 
Quoted 5 lines

mm/slub.c static bool __slab_try_return_freelist(struct kmem_cache *s, struct slab *slab, · line 5700

Quoted 2 lines
mm/slub.c +1 -1
-	if (!slab_update_freelist(s, slab, &old, &new, "__slab_try_return_freelist"))
+	n = get_node(s, slab_nid(slab));

Wonder if we could just pass the 'n' to __slab_try_return_freelist() that we already have in the caller.

Sounds good, will do.

Signature · 2 lines
Cheers,
Harry / Hyeonggon