Skip to contentSkip to conversation

writeback, cgroup: fix null-ptr-deref write in bdi_split_work_to_wbs

PATCH Applied Baokun Li 5 participants 8 messages Apr 10, 2023 – May 4, 2023

Subsystems (4)
lore
Share & tools
lore

Baokun Li

author
#0 lore raw

PATCH v2 1/1 writeback, cgroup: fix null-ptr-deref write in bdi_split_work_to_wbs

KASAN report null-ptr-deref:
==================================================================

BUG: KASAN: null-ptr-deref in bdi_split_work_to_wbs+0x5c5/0x7b0
Write of size 8 at addr 0000000000000000 by task sync/943
CPU: 5 PID: 943 Comm: sync Tainted: 6.3.0-rc5-next-20230406-dirty #461
Call Trace:
 <TASK>
 dump_stack_lvl+0x7f/0xc0
 print_report+0x2ba/0x340
 kasan_report+0xc4/0x120
 kasan_check_range+0x1b7/0x2e0
 __kasan_check_write+0x24/0x40
 bdi_split_work_to_wbs+0x5c5/0x7b0
 sync_inodes_sb+0x195/0x630
 sync_inodes_one_sb+0x3a/0x50
 iterate_supers+0x106/0x1b0
 ksys_sync+0x98/0x160

[...]
==================================================================

The race that causes the above issue is as follows:

cpu1 cpu2
-------------------------|------------------------- inode_switch_wbs
INIT_WORK(&isw->work, inode_switch_wbs_work_fn) queue_rcu_work(isw_wq, &isw->work)
// queue_work async
inode_switch_wbs_work_fn
wb_put_many(old_wb, nr_switched)

    percpu_ref_put_many
     ref->data->release(ref)
     cgwb_release
      queue_work(cgwb_release_wq, &wb->release_work)
      // queue_work async
       &wb->release_work
       cgwb_release_workfn
                            ksys_sync
                             iterate_supers
                              sync_inodes_one_sb
                               sync_inodes_sb
                                bdi_split_work_to_wbs
                                 kmalloc(sizeof(*work), GFP_ATOMIC)
                                 // alloc memory failed
        percpu_ref_exit
         ref->data = NULL
         kfree(data)
                                 wb_get(wb)
                                  percpu_ref_get(&wb->refcnt)
                                   percpu_ref_get_many(ref, 1)
                                    atomic_long_add(nr, &ref->data->count)
                                     atomic64_add(i, v)
                                     // trigger null-ptr-deref

bdi_split_work_to_wbs() traverses &bdi->wb_list to split work into all wbs. If the allocation of new work fails, the on-stack fallback will be used and the reference count of the current wb is increased afterwards. If cgroup writeback membership switches occur before getting the reference count and the current wb is released as old_wd, then calling wb_get() or wb_put() will trigger the null pointer dereference above.

This issue was introduced in v4.3-rc7 (see fix tag1). Both sync_inodes_sb() and __writeback_inodes_sb_nr() calls to bdi_split_work_to_wbs() can trigger this issue. For scenarios called via sync_inodes_sb(), originally commit 7fc5854f8c6e ("writeback: synchronize sync(2) against cgroup writeback membership switches") reduced the possibility of the issue by adding wb_switch_rwsem, but in v5.14-rc1 (see fix tag2) removed the "inode_io_list_del_locked(inode, old_wb)" from inode_switch_wbs_work_fn() so that wb->state contains WB_has_dirty_io, thus old_wb is not skipped when traversing wbs in bdi_split_work_to_wbs(), and the issue becomes easily reproducible again.

To solve this problem, percpu_ref_exit() is called under RCU protection to avoid race between cgwb_release_workfn() and bdi_split_work_to_wbs(). Moreover, replace wb_get() with wb_tryget() in bdi_split_work_to_wbs(), and skip the current wb if wb_tryget() fails because the wb has already been shutdown.

Fixes b817525a4a80 ("writeback: bdi_writeback iteration must not skip dying ones")
Fixes f3b6a6df38aa ("writeback, cgroup: keep list of inodes attached to bdi_writeback")

---
V1->V2:
Use RCU instead of wb_switch_rwsem to avoid race.

 fs/fs-writeback.c | 17 ++++++++++-------
 mm/backing-dev.c  | 12 ++++++++++--
 2 files changed, 20 insertions(+), 9 deletions(-)
fs/fs-writeback.c +10 -7
index 195dc23e0d83..1db3e3c24b43 100644
@@ -978,6 +978,16 @@ static void bdi_split_work_to_wbs(struct backing_dev_info *bdi,
 			continue;
 		}
 
+		/*
+		 * If wb_tryget fails, the wb has been shutdown, skip it.
+		 *
+		 * Pin @wb so that it stays on @bdi->wb_list.  This allows
+		 * continuing iteration from @wb after dropping and
+		 * regrabbing rcu read lock.
+		 */
+		if (!wb_tryget(wb))
+			continue;
+
 		/* alloc failed, execute synchronously using on-stack fallback */
 		work = &fallback_work;
 		*work = *base_work;
@@ -986,13 +996,6 @@ static void bdi_split_work_to_wbs(struct backing_dev_info *bdi,
 		work->done = &fallback_work_done;
 
 		wb_queue_work(wb, work);
-
-		/*
-		 * Pin @wb so that it stays on @bdi->wb_list.  This allows
-		 * continuing iteration from @wb after dropping and
-		 * regrabbing rcu read lock.
-		 */
-		wb_get(wb);
 		last_wb = wb;
 
 		rcu_read_unlock();
mm/backing-dev.c +10 -2
index ad011308cebe..43b48750b491 100644
@@ -507,6 +507,15 @@ static LIST_HEAD(offline_cgwbs);
 static void cleanup_offline_cgwbs_workfn(struct work_struct *work);
 static DECLARE_WORK(cleanup_offline_cgwbs_work, cleanup_offline_cgwbs_workfn);
 
+static void cgwb_free_rcu(struct rcu_head *rcu_head)
+{
+	struct bdi_writeback *wb = container_of(rcu_head,
+			struct bdi_writeback, rcu);
+
+	percpu_ref_exit(&wb->refcnt);
+	kfree(wb);
+}
+
 static void cgwb_release_workfn(struct work_struct *work)
 {
 	struct bdi_writeback *wb = container_of(work, struct bdi_writeback,
@@ -529,11 +538,10 @@ static void cgwb_release_workfn(struct work_struct *work)
 	list_del(&wb->offline_node);
 	spin_unlock_irq(&cgwb_lock);
 
-	percpu_ref_exit(&wb->refcnt);
 	wb_exit(wb);
 	bdi_put(bdi);
 	WARN_ON_ONCE(!list_empty(&wb->b_attached));
-	kfree_rcu(wb, rcu);
+	call_rcu(&wb->rcu, cgwb_free_rcu);
 }
 
 static void cgwb_release(struct percpu_ref *refcnt)
Signature · 1 line
2.31.1
maintainer
#1 lore raw
Quoted 11 lines · On Mon, 10 Apr 2023 21:08:26 +0800 Baokun Li <[email protected]> wrote:

...

To solve this problem, percpu_ref_exit() is called under RCU protection to avoid race between cgwb_release_workfn() and bdi_split_work_to_wbs(). Moreover, replace wb_get() with wb_tryget() in bdi_split_work_to_wbs(), and skip the current wb if wb_tryget() fails because the wb has already been shutdown.

Fixes b817525a4a80 ("writeback: bdi_writeback iteration must not skip dying ones")
Fixes f3b6a6df38aa ("writeback, cgroup: keep list of inodes attached to bdi_writeback")

Two Fixes: is awkward. The Fixes: serves a guide to which kernel versions should be patched, but those two commits are six years apart.

So... how far back should this fix be backported?

Quoted 2 lines
 fs/fs-writeback.c | 17 ++++++++++-------
 mm/backing-dev.c  | 12 ++++++++++--

Jens, which tree do you think should carry this?

author
#2 lore raw
Quoted 10 lines · Andrew Morton
Quoted 8 lines · On Mon, 10 Apr 2023 21:08:26 +0800 Baokun Li <[email protected]> wrote:

...

To solve this problem, percpu_ref_exit() is called under RCU protection to avoid race between cgwb_release_workfn() and bdi_split_work_to_wbs(). Moreover, replace wb_get() with wb_tryget() in bdi_split_work_to_wbs(), and skip the current wb if wb_tryget() fails because the wb has already been shutdown.

Quoted 6 lines
Quoted 2 lines
Fixes b817525a4a80 ("writeback: bdi_writeback iteration must not skip dying ones")
Fixes f3b6a6df38aa ("writeback, cgroup: keep list of inodes attached to bdi_writeback")

Two Fixes: is awkward. The Fixes: serves a guide to which kernel versions should be patched, but those two commits are six years apart.

So... how far back should this fix be backported?

This issue was introduced in v4.3-rc7 by commit b817525a4a80 ("writeback: bdi_writeback iteration must not skip dying ones"), so anything that has this commit incorporated is problematic.

Another fix tag patch invalidates a previously unintentional fix, and then the problem becomes more easily reproducible. This fix tag can actually be removed, and is added here so that people who see the patch will know what happened.

Signature · 3 lines
With Best Regards,
Baokun Li
.
reviewer Reviewed-by on v2 1/1
#3 lore raw
In reply to Baokun Li
Quoted 78 lines · Baokun Li

KASAN report null-ptr-deref:
==================================================================

BUG: KASAN: null-ptr-deref in bdi_split_work_to_wbs+0x5c5/0x7b0
Write of size 8 at addr 0000000000000000 by task sync/943
CPU: 5 PID: 943 Comm: sync Tainted: 6.3.0-rc5-next-20230406-dirty #461
Call Trace:
 <TASK>
 dump_stack_lvl+0x7f/0xc0
 print_report+0x2ba/0x340
 kasan_report+0xc4/0x120
 kasan_check_range+0x1b7/0x2e0
 __kasan_check_write+0x24/0x40
 bdi_split_work_to_wbs+0x5c5/0x7b0
 sync_inodes_sb+0x195/0x630
 sync_inodes_one_sb+0x3a/0x50
 iterate_supers+0x106/0x1b0
 ksys_sync+0x98/0x160

[...]
==================================================================

The race that causes the above issue is as follows:

cpu1 cpu2
-------------------------|------------------------- inode_switch_wbs
INIT_WORK(&isw->work, inode_switch_wbs_work_fn) queue_rcu_work(isw_wq, &isw->work)
// queue_work async
inode_switch_wbs_work_fn
wb_put_many(old_wb, nr_switched)

    percpu_ref_put_many
     ref->data->release(ref)
     cgwb_release
      queue_work(cgwb_release_wq, &wb->release_work)
      // queue_work async
       &wb->release_work
       cgwb_release_workfn
                            ksys_sync
                             iterate_supers
                              sync_inodes_one_sb
                               sync_inodes_sb
                                bdi_split_work_to_wbs
                                 kmalloc(sizeof(*work), GFP_ATOMIC)
                                 // alloc memory failed
        percpu_ref_exit
         ref->data = NULL
         kfree(data)
                                 wb_get(wb)
                                  percpu_ref_get(&wb->refcnt)
                                   percpu_ref_get_many(ref, 1)
                                    atomic_long_add(nr, &ref->data->count)
                                     atomic64_add(i, v)
                                     // trigger null-ptr-deref

bdi_split_work_to_wbs() traverses &bdi->wb_list to split work into all wbs. If the allocation of new work fails, the on-stack fallback will be used and the reference count of the current wb is increased afterwards. If cgroup writeback membership switches occur before getting the reference count and the current wb is released as old_wd, then calling wb_get() or wb_put() will trigger the null pointer dereference above.

This issue was introduced in v4.3-rc7 (see fix tag1). Both sync_inodes_sb() and __writeback_inodes_sb_nr() calls to bdi_split_work_to_wbs() can trigger this issue. For scenarios called via sync_inodes_sb(), originally commit 7fc5854f8c6e ("writeback: synchronize sync(2) against cgroup writeback membership switches") reduced the possibility of the issue by adding wb_switch_rwsem, but in v5.14-rc1 (see fix tag2) removed the "inode_io_list_del_locked(inode, old_wb)" from inode_switch_wbs_work_fn() so that wb->state contains WB_has_dirty_io, thus old_wb is not skipped when traversing wbs in bdi_split_work_to_wbs(), and the issue becomes easily reproducible again.

To solve this problem, percpu_ref_exit() is called under RCU protection to avoid race between cgwb_release_workfn() and bdi_split_work_to_wbs(). Moreover, replace wb_get() with wb_tryget() in bdi_split_work_to_wbs(), and skip the current wb if wb_tryget() fails because the wb has already been shutdown.

Quoted 7 lines
Fixes b817525a4a80 ("writeback: bdi_writeback iteration must not skip dying ones")
Fixes f3b6a6df38aa ("writeback, cgroup: keep list of inodes attached to bdi_writeback")

---
V1->V2:
Use RCU instead of wb_switch_rwsem to avoid race.

The cgwb shutdown code is really messy. But your change looks good to me and I don't see an easier way around this race. Feel free to add:

Honza

Quoted 76 lines
 fs/fs-writeback.c | 17 ++++++++++-------
 mm/backing-dev.c  | 12 ++++++++++--
 2 files changed, 20 insertions(+), 9 deletions(-)
fs/fs-writeback.c +10 -7
index 195dc23e0d83..1db3e3c24b43 100644
@@ -978,6 +978,16 @@ static void bdi_split_work_to_wbs(struct backing_dev_info *bdi,
 			continue;
 		}
 
+		/*
+		 * If wb_tryget fails, the wb has been shutdown, skip it.
+		 *
+		 * Pin @wb so that it stays on @bdi->wb_list.  This allows
+		 * continuing iteration from @wb after dropping and
+		 * regrabbing rcu read lock.
+		 */
+		if (!wb_tryget(wb))
+			continue;
+
 		/* alloc failed, execute synchronously using on-stack fallback */
 		work = &fallback_work;
 		*work = *base_work;
@@ -986,13 +996,6 @@ static void bdi_split_work_to_wbs(struct backing_dev_info *bdi,
 		work->done = &fallback_work_done;
 
 		wb_queue_work(wb, work);
-
-		/*
-		 * Pin @wb so that it stays on @bdi->wb_list.  This allows
-		 * continuing iteration from @wb after dropping and
-		 * regrabbing rcu read lock.
-		 */
-		wb_get(wb);
 		last_wb = wb;
 
 		rcu_read_unlock();
mm/backing-dev.c +10 -2
index ad011308cebe..43b48750b491 100644
@@ -507,6 +507,15 @@ static LIST_HEAD(offline_cgwbs);
 static void cleanup_offline_cgwbs_workfn(struct work_struct *work);
 static DECLARE_WORK(cleanup_offline_cgwbs_work, cleanup_offline_cgwbs_workfn);
 
+static void cgwb_free_rcu(struct rcu_head *rcu_head)
+{
+	struct bdi_writeback *wb = container_of(rcu_head,
+			struct bdi_writeback, rcu);
+
+	percpu_ref_exit(&wb->refcnt);
+	kfree(wb);
+}
+
 static void cgwb_release_workfn(struct work_struct *work)
 {
 	struct bdi_writeback *wb = container_of(work, struct bdi_writeback,
@@ -529,11 +538,10 @@ static void cgwb_release_workfn(struct work_struct *work)
 	list_del(&wb->offline_node);
 	spin_unlock_irq(&cgwb_lock);
 
-	percpu_ref_exit(&wb->refcnt);
 	wb_exit(wb);
 	bdi_put(bdi);
 	WARN_ON_ONCE(!list_empty(&wb->b_attached));
-	kfree_rcu(wb, rcu);
+	call_rcu(&wb->rcu, cgwb_free_rcu);
 }
 
 static void cgwb_release(struct percpu_ref *refcnt)
Signature · 1 line
2.31.1
Signature · 2 lines
Jan Kara <[email protected]>
SUSE Labs, CR
Acked-by on v2 1/1
#4 lore raw
In reply to Baokun Li
Quoted 78 lines · Baokun Li

KASAN report null-ptr-deref:
==================================================================

BUG: KASAN: null-ptr-deref in bdi_split_work_to_wbs+0x5c5/0x7b0
Write of size 8 at addr 0000000000000000 by task sync/943
CPU: 5 PID: 943 Comm: sync Tainted: 6.3.0-rc5-next-20230406-dirty #461
Call Trace:
 <TASK>
 dump_stack_lvl+0x7f/0xc0
 print_report+0x2ba/0x340
 kasan_report+0xc4/0x120
 kasan_check_range+0x1b7/0x2e0
 __kasan_check_write+0x24/0x40
 bdi_split_work_to_wbs+0x5c5/0x7b0
 sync_inodes_sb+0x195/0x630
 sync_inodes_one_sb+0x3a/0x50
 iterate_supers+0x106/0x1b0
 ksys_sync+0x98/0x160

[...]
==================================================================

The race that causes the above issue is as follows:

cpu1 cpu2
-------------------------|------------------------- inode_switch_wbs
INIT_WORK(&isw->work, inode_switch_wbs_work_fn) queue_rcu_work(isw_wq, &isw->work)
// queue_work async
inode_switch_wbs_work_fn
wb_put_many(old_wb, nr_switched)

    percpu_ref_put_many
     ref->data->release(ref)
     cgwb_release
      queue_work(cgwb_release_wq, &wb->release_work)
      // queue_work async
       &wb->release_work
       cgwb_release_workfn
                            ksys_sync
                             iterate_supers
                              sync_inodes_one_sb
                               sync_inodes_sb
                                bdi_split_work_to_wbs
                                 kmalloc(sizeof(*work), GFP_ATOMIC)
                                 // alloc memory failed
        percpu_ref_exit
         ref->data = NULL
         kfree(data)
                                 wb_get(wb)
                                  percpu_ref_get(&wb->refcnt)
                                   percpu_ref_get_many(ref, 1)
                                    atomic_long_add(nr, &ref->data->count)
                                     atomic64_add(i, v)
                                     // trigger null-ptr-deref

bdi_split_work_to_wbs() traverses &bdi->wb_list to split work into all wbs. If the allocation of new work fails, the on-stack fallback will be used and the reference count of the current wb is increased afterwards. If cgroup writeback membership switches occur before getting the reference count and the current wb is released as old_wd, then calling wb_get() or wb_put() will trigger the null pointer dereference above.

This issue was introduced in v4.3-rc7 (see fix tag1). Both sync_inodes_sb() and __writeback_inodes_sb_nr() calls to bdi_split_work_to_wbs() can trigger this issue. For scenarios called via sync_inodes_sb(), originally commit 7fc5854f8c6e ("writeback: synchronize sync(2) against cgroup writeback membership switches") reduced the possibility of the issue by adding wb_switch_rwsem, but in v5.14-rc1 (see fix tag2) removed the "inode_io_list_del_locked(inode, old_wb)" from inode_switch_wbs_work_fn() so that wb->state contains WB_has_dirty_io, thus old_wb is not skipped when traversing wbs in bdi_split_work_to_wbs(), and the issue becomes easily reproducible again.

To solve this problem, percpu_ref_exit() is called under RCU protection to avoid race between cgwb_release_workfn() and bdi_split_work_to_wbs(). Moreover, replace wb_get() with wb_tryget() in bdi_split_work_to_wbs(), and skip the current wb if wb_tryget() fails because the wb has already been shutdown.

Quoted 4 lines
Fixes b817525a4a80 ("writeback: bdi_writeback iteration must not skip dying ones")
Fixes f3b6a6df38aa ("writeback, cgroup: keep list of inodes attached to bdi_writeback")

Thanks.

Signature · 1 line
tejun
maintainer
#5 lore raw
In reply to Baokun Li
Quoted 72 lines · On Mon, 10 Apr 2023 21:08:26 +0800 Baokun Li <[email protected]> wrote:

KASAN report null-ptr-deref:
==================================================================

BUG: KASAN: null-ptr-deref in bdi_split_work_to_wbs+0x5c5/0x7b0
Write of size 8 at addr 0000000000000000 by task sync/943
CPU: 5 PID: 943 Comm: sync Tainted: 6.3.0-rc5-next-20230406-dirty #461
Call Trace:
 <TASK>
 dump_stack_lvl+0x7f/0xc0
 print_report+0x2ba/0x340
 kasan_report+0xc4/0x120
 kasan_check_range+0x1b7/0x2e0
 __kasan_check_write+0x24/0x40
 bdi_split_work_to_wbs+0x5c5/0x7b0
 sync_inodes_sb+0x195/0x630
 sync_inodes_one_sb+0x3a/0x50
 iterate_supers+0x106/0x1b0
 ksys_sync+0x98/0x160

[...]
==================================================================

The race that causes the above issue is as follows:

cpu1 cpu2
-------------------------|------------------------- inode_switch_wbs
INIT_WORK(&isw->work, inode_switch_wbs_work_fn) queue_rcu_work(isw_wq, &isw->work)
// queue_work async
inode_switch_wbs_work_fn
wb_put_many(old_wb, nr_switched)

    percpu_ref_put_many
     ref->data->release(ref)
     cgwb_release
      queue_work(cgwb_release_wq, &wb->release_work)
      // queue_work async
       &wb->release_work
       cgwb_release_workfn
                            ksys_sync
                             iterate_supers
                              sync_inodes_one_sb
                               sync_inodes_sb
                                bdi_split_work_to_wbs
                                 kmalloc(sizeof(*work), GFP_ATOMIC)
                                 // alloc memory failed
        percpu_ref_exit
         ref->data = NULL
         kfree(data)
                                 wb_get(wb)
                                  percpu_ref_get(&wb->refcnt)
                                   percpu_ref_get_many(ref, 1)
                                    atomic_long_add(nr, &ref->data->count)
                                     atomic64_add(i, v)
                                     // trigger null-ptr-deref

bdi_split_work_to_wbs() traverses &bdi->wb_list to split work into all wbs. If the allocation of new work fails, the on-stack fallback will be used and the reference count of the current wb is increased afterwards. If cgroup writeback membership switches occur before getting the reference count and the current wb is released as old_wd, then calling wb_get() or wb_put() will trigger the null pointer dereference above.

This issue was introduced in v4.3-rc7 (see fix tag1). Both sync_inodes_sb() and __writeback_inodes_sb_nr() calls to bdi_split_work_to_wbs() can trigger this issue. For scenarios called via sync_inodes_sb(), originally commit 7fc5854f8c6e ("writeback: synchronize sync(2) against cgroup writeback membership switches") reduced the possibility of the issue by adding wb_switch_rwsem, but in v5.14-rc1 (see fix tag2) removed the "inode_io_list_del_locked(inode, old_wb)" from inode_switch_wbs_work_fn() so that wb->state contains WB_has_dirty_io, thus old_wb is not skipped when traversing wbs in bdi_split_work_to_wbs(), and the issue becomes easily reproducible again.

Quoted 8 lines

To solve this problem, percpu_ref_exit() is called under RCU protection to avoid race between cgwb_release_workfn() and bdi_split_work_to_wbs(). Moreover, replace wb_get() with wb_tryget() in bdi_split_work_to_wbs(), and skip the current wb if wb_tryget() fails because the wb has already been shutdown.

Fixes b817525a4a80 ("writeback: bdi_writeback iteration must not skip dying ones")
Fixes f3b6a6df38aa ("writeback, cgroup: keep list of inodes attached to bdi_writeback")

Cc Roman for this second commit.

Quoted 1 line

Having two Fixes: is awkward. These serve as a guide to tell -stable maintainers which kernels need the fix. Can we be more precise?

Quoted 38 lines
fs/fs-writeback.c +10 -7
@@ -978,6 +978,16 @@ static void bdi_split_work_to_wbs(struct backing_dev_info *bdi,
 			continue;
 		}
 
+		/*
+		 * If wb_tryget fails, the wb has been shutdown, skip it.
+		 *
+		 * Pin @wb so that it stays on @bdi->wb_list.  This allows
+		 * continuing iteration from @wb after dropping and
+		 * regrabbing rcu read lock.
+		 */
+		if (!wb_tryget(wb))
+			continue;
+
 		/* alloc failed, execute synchronously using on-stack fallback */
 		work = &fallback_work;
 		*work = *base_work;
@@ -986,13 +996,6 @@ static void bdi_split_work_to_wbs(struct backing_dev_info *bdi,
 		work->done = &fallback_work_done;
 
 		wb_queue_work(wb, work);
-
-		/*
-		 * Pin @wb so that it stays on @bdi->wb_list.  This allows
-		 * continuing iteration from @wb after dropping and
-		 * regrabbing rcu read lock.
-		 */
-		wb_get(wb);
 		last_wb = wb;
 
 		rcu_read_unlock();
mm/backing-dev.c +0 -0
index ad011308cebe..43b48750b491 100644

mm/backing-dev.c static LIST_HEAD(offline_cgwbs); · line 513

Quoted 7 lines
mm/backing-dev.c +4 -0
@@ -507,6 +507,15 @@ static LIST_HEAD(offline_cgwbs);
 static void cleanup_offline_cgwbs_workfn(struct work_struct *work);
 static DECLARE_WORK(cleanup_offline_cgwbs_work, cleanup_offline_cgwbs_workfn);
 
+static void cgwb_free_rcu(struct rcu_head *rcu_head)
+{
+	struct bdi_writeback *wb = container_of(rcu_head,
+			struct bdi_writeback, rcu);

nit:

	struct bdi_writeback *wb;

	wb = container_of(rcu_head, struct bdi_writeback, rcu);

looks nicer, no?

Quoted 20 lines
mm/backing-dev.c +5 -2
+	percpu_ref_exit(&wb->refcnt);
+	kfree(wb);
+}
+
 static void cgwb_release_workfn(struct work_struct *work)
 {
 	struct bdi_writeback *wb = container_of(work, struct bdi_writeback,
@@ -529,11 +538,10 @@ static void cgwb_release_workfn(struct work_struct *work)
 	list_del(&wb->offline_node);
 	spin_unlock_irq(&cgwb_lock);
 
-	percpu_ref_exit(&wb->refcnt);
 	wb_exit(wb);
 	bdi_put(bdi);
 	WARN_ON_ONCE(!list_empty(&wb->b_attached));
-	kfree_rcu(wb, rcu);
+	call_rcu(&wb->rcu, cgwb_free_rcu);
 }
 
 static void cgwb_release(struct percpu_ref *refcnt)
#6 lore raw
Quoted 80 lines · Andrew Morton
Quoted 78 lines · On Mon, 10 Apr 2023 21:08:26 +0800 Baokun Li <[email protected]> wrote:

KASAN report null-ptr-deref:
==================================================================

BUG: KASAN: null-ptr-deref in bdi_split_work_to_wbs+0x5c5/0x7b0
Write of size 8 at addr 0000000000000000 by task sync/943
CPU: 5 PID: 943 Comm: sync Tainted: 6.3.0-rc5-next-20230406-dirty #461
Call Trace:
 <TASK>
 dump_stack_lvl+0x7f/0xc0
 print_report+0x2ba/0x340
 kasan_report+0xc4/0x120
 kasan_check_range+0x1b7/0x2e0
 __kasan_check_write+0x24/0x40
 bdi_split_work_to_wbs+0x5c5/0x7b0
 sync_inodes_sb+0x195/0x630
 sync_inodes_one_sb+0x3a/0x50
 iterate_supers+0x106/0x1b0
 ksys_sync+0x98/0x160

[...]
==================================================================

The race that causes the above issue is as follows:

cpu1 cpu2
-------------------------|------------------------- inode_switch_wbs
INIT_WORK(&isw->work, inode_switch_wbs_work_fn) queue_rcu_work(isw_wq, &isw->work)
// queue_work async
inode_switch_wbs_work_fn
wb_put_many(old_wb, nr_switched)

    percpu_ref_put_many
     ref->data->release(ref)
     cgwb_release
      queue_work(cgwb_release_wq, &wb->release_work)
      // queue_work async
       &wb->release_work
       cgwb_release_workfn
                            ksys_sync
                             iterate_supers
                              sync_inodes_one_sb
                               sync_inodes_sb
                                bdi_split_work_to_wbs
                                 kmalloc(sizeof(*work), GFP_ATOMIC)
                                 // alloc memory failed
        percpu_ref_exit
         ref->data = NULL
         kfree(data)
                                 wb_get(wb)
                                  percpu_ref_get(&wb->refcnt)
                                   percpu_ref_get_many(ref, 1)
                                    atomic_long_add(nr, &ref->data->count)
                                     atomic64_add(i, v)
                                     // trigger null-ptr-deref

bdi_split_work_to_wbs() traverses &bdi->wb_list to split work into all wbs. If the allocation of new work fails, the on-stack fallback will be used and the reference count of the current wb is increased afterwards. If cgroup writeback membership switches occur before getting the reference count and the current wb is released as old_wd, then calling wb_get() or wb_put() will trigger the null pointer dereference above.

This issue was introduced in v4.3-rc7 (see fix tag1). Both sync_inodes_sb() and __writeback_inodes_sb_nr() calls to bdi_split_work_to_wbs() can trigger this issue. For scenarios called via sync_inodes_sb(), originally commit 7fc5854f8c6e ("writeback: synchronize sync(2) against cgroup writeback membership switches") reduced the possibility of the issue by adding wb_switch_rwsem, but in v5.14-rc1 (see fix tag2) removed the "inode_io_list_del_locked(inode, old_wb)" from inode_switch_wbs_work_fn() so that wb->state contains WB_has_dirty_io, thus old_wb is not skipped when traversing wbs in bdi_split_work_to_wbs(), and the issue becomes easily reproducible again.

To solve this problem, percpu_ref_exit() is called under RCU protection to avoid race between cgwb_release_workfn() and bdi_split_work_to_wbs(). Moreover, replace wb_get() with wb_tryget() in bdi_split_work_to_wbs(), and skip the current wb if wb_tryget() fails because the wb has already been shutdown.

Quoted 4 lines
Quoted 2 lines
Fixes b817525a4a80 ("writeback: bdi_writeback iteration must not skip dying ones")
Fixes f3b6a6df38aa ("writeback, cgroup: keep list of inodes attached to bdi_writeback")

Cc Roman for this second commit.

Thanks for the heads up!

The patch looks good to me.

Thanks!

author
#7 lore raw
In reply to Andrew Morton
Quoted 80 lines · Andrew Morton
Quoted 78 lines · On Mon, 10 Apr 2023 21:08:26 +0800 Baokun Li <[email protected]> wrote:

KASAN report null-ptr-deref:
==================================================================

BUG: KASAN: null-ptr-deref in bdi_split_work_to_wbs+0x5c5/0x7b0
Write of size 8 at addr 0000000000000000 by task sync/943
CPU: 5 PID: 943 Comm: sync Tainted: 6.3.0-rc5-next-20230406-dirty #461
Call Trace:
  <TASK>
  dump_stack_lvl+0x7f/0xc0
  print_report+0x2ba/0x340
  kasan_report+0xc4/0x120
  kasan_check_range+0x1b7/0x2e0
  __kasan_check_write+0x24/0x40
  bdi_split_work_to_wbs+0x5c5/0x7b0
  sync_inodes_sb+0x195/0x630
  sync_inodes_one_sb+0x3a/0x50
  iterate_supers+0x106/0x1b0
  ksys_sync+0x98/0x160

[...]
==================================================================

The race that causes the above issue is as follows:

cpu1 cpu2
-------------------------|------------------------- inode_switch_wbs
INIT_WORK(&isw->work, inode_switch_wbs_work_fn) queue_rcu_work(isw_wq, &isw->work)
// queue_work async
inode_switch_wbs_work_fn

    wb_put_many(old_wb, nr_switched)
     percpu_ref_put_many
      ref->data->release(ref)
      cgwb_release
       queue_work(cgwb_release_wq, &wb->release_work)
       // queue_work async
        &wb->release_work
        cgwb_release_workfn
                             ksys_sync
                              iterate_supers
                               sync_inodes_one_sb
                                sync_inodes_sb
                                 bdi_split_work_to_wbs
                                  kmalloc(sizeof(*work), GFP_ATOMIC)
                                  // alloc memory failed
         percpu_ref_exit
          ref->data = NULL
          kfree(data)
                                  wb_get(wb)
                                   percpu_ref_get(&wb->refcnt)
                                    percpu_ref_get_many(ref, 1)
                                     atomic_long_add(nr, &ref->data->count)
                                      atomic64_add(i, v)
                                      // trigger null-ptr-deref

bdi_split_work_to_wbs() traverses &bdi->wb_list to split work into all wbs. If the allocation of new work fails, the on-stack fallback will be used and the reference count of the current wb is increased afterwards. If cgroup writeback membership switches occur before getting the reference count and the current wb is released as old_wd, then calling wb_get() or wb_put() will trigger the null pointer dereference above.

This issue was introduced in v4.3-rc7 (see fix tag1). Both sync_inodes_sb() and __writeback_inodes_sb_nr() calls to bdi_split_work_to_wbs() can trigger this issue. For scenarios called via sync_inodes_sb(), originally commit 7fc5854f8c6e ("writeback: synchronize sync(2) against cgroup writeback membership switches") reduced the possibility of the issue by adding wb_switch_rwsem, but in v5.14-rc1 (see fix tag2) removed the "inode_io_list_del_locked(inode, old_wb)" from inode_switch_wbs_work_fn() so that wb->state contains WB_has_dirty_io, thus old_wb is not skipped when traversing wbs in bdi_split_work_to_wbs(), and the issue becomes easily reproducible again.

To solve this problem, percpu_ref_exit() is called under RCU protection to avoid race between cgwb_release_workfn() and bdi_split_work_to_wbs(). Moreover, replace wb_get() with wb_tryget() in bdi_split_work_to_wbs(), and skip the current wb if wb_tryget() fails because the wb has already been shutdown.

Quoted 3 lines
Quoted 2 lines
Fixes b817525a4a80 ("writeback: bdi_writeback iteration must not skip dying ones")
Fixes f3b6a6df38aa ("writeback, cgroup: keep list of inodes attached to bdi_writeback")

Cc Roman for this second commit.

Thanks! I forgot to cc to Roman!

Quoted 4 lines
Quoted 2 lines

Having two Fixes: is awkward. These serve as a guide to tell -stable maintainers which kernels need the fix. Can we be more precise?

Yes! I'm very sorry about that. As I replied earlier, this issue was introduced in v4.3-rc7.

Quoted 47 lines
Quoted 38 lines
fs/fs-writeback.c +10 -7
@@ -978,6 +978,16 @@ static void bdi_split_work_to_wbs(struct backing_dev_info *bdi,
  			continue;
  		}
 
+		/*
+		 * If wb_tryget fails, the wb has been shutdown, skip it.
+		 *
+		 * Pin @wb so that it stays on @bdi->wb_list.  This allows
+		 * continuing iteration from @wb after dropping and
+		 * regrabbing rcu read lock.
+		 */
+		if (!wb_tryget(wb))
+			continue;
+
  		/* alloc failed, execute synchronously using on-stack fallback */
  		work = &fallback_work;
  		*work = *base_work;
@@ -986,13 +996,6 @@ static void bdi_split_work_to_wbs(struct backing_dev_info *bdi,
  		work->done = &fallback_work_done;
 
  		wb_queue_work(wb, work);
-
-		/*
-		 * Pin @wb so that it stays on @bdi->wb_list.  This allows
-		 * continuing iteration from @wb after dropping and
-		 * regrabbing rcu read lock.
-		 */
-		wb_get(wb);
  		last_wb = wb;
 
  		rcu_read_unlock();
mm/backing-dev.c +0 -0
index ad011308cebe..43b48750b491 100644

mm/backing-dev.c static LIST_HEAD(offline_cgwbs); · line 513

Quoted 7 lines
mm/backing-dev.c +4 -0
@@ -507,6 +507,15 @@ static LIST_HEAD(offline_cgwbs);
  static void cleanup_offline_cgwbs_workfn(struct work_struct *work);
  static DECLARE_WORK(cleanup_offline_cgwbs_work, cleanup_offline_cgwbs_workfn);
 
+static void cgwb_free_rcu(struct rcu_head *rcu_head)
+{
+	struct bdi_writeback *wb = container_of(rcu_head,
+			struct bdi_writeback, rcu);

nit:

Quoted 5 lines
	struct bdi_writeback *wb;

	wb = container_of(rcu_head, struct bdi_writeback, rcu);

looks nicer, no?

Yes, it looks neater.

Quoted 20 lines
Quoted 20 lines
mm/backing-dev.c +5 -2
+	percpu_ref_exit(&wb->refcnt);
+	kfree(wb);
+}
+
  static void cgwb_release_workfn(struct work_struct *work)
  {
  	struct bdi_writeback *wb = container_of(work, struct bdi_writeback,
@@ -529,11 +538,10 @@ static void cgwb_release_workfn(struct work_struct *work)
  	list_del(&wb->offline_node);
  	spin_unlock_irq(&cgwb_lock);
 
-	percpu_ref_exit(&wb->refcnt);
  	wb_exit(wb);
  	bdi_put(bdi);
  	WARN_ON_ONCE(!list_empty(&wb->b_attached));
-	kfree_rcu(wb, rcu);
+	call_rcu(&wb->rcu, cgwb_free_rcu);
  }
 
Quoted 1 line

mm/backing-dev.c static void cgwb_release_workfn(struct work_struct *work) · line 547

Quoted 1 line
mm/backing-dev.c +0 -0
  static void cgwb_release(struct percpu_ref *refcnt)

Thanks!

Signature · 3 lines
With Best Regards,
Baokun Li
.