Re: [dm-devel resend] dm mpath: fix UAF in multipath_message()

Li Lingfeng <lilingfeng3@xxxxxxxxxx> · Wed, 2 Aug 2023 11:06:10 +0800

Friendly ping ...

Thanks,
Li

在 2023/5/18 20:11, lilingfeng (A) 写道:

在 2022/10/18 3:56, Mike Snitzer 写道:
On Mon, Oct 10 2022 at 10:41P -0400,
Luo Meng <luomeng@xxxxxxxxxxxxxxx> wrote:

From: Luo Meng <luomeng12@xxxxxxxxxx>

If dm_get_device() create dd in multipath_message(),
and then call table_deps() after dm_put_table_device(),
it will lead to concurrency UAF bugs.

One of the concurrency UAF can be shown as below:

          (USE)                        |    (FREE)
                                       |  target_message
                                       |    multipath_message
                                       |      dm_put_device
                                       | dm_put_table_device #
                                       |          kfree(td)  # 
table_device *td
ioctl # DM_TABLE_DEPS_CMD             |         ...
   table_deps                          |         ...
   dm_get_live_or_inactive_table       |         ...
     retrieve_dep                      |         ...
     list_for_each_entry               |         ...
       deps->dev[count++] =            |         ...
           huge_encode_dev             |         ...
           (dd->dm_dev->bdev->bd_dev)  | list_del(&dd->list)
                                       |        kfree(dd) # 
dm_dev_internal

The root cause of UAF bugs is that find_device() failed in
dm_get_device() and will create dd and refcount set 1, kfree()
in dm_put_table() is not protected. When td, which there are
still pointers point to, is released, the concurrency UAF bug
will happen.

This patch add a flag to determine whether to create a new dd.

Fixes: 8215d6ec5fee(dm table: remove unused dm_get_device range 
parameters)
Signed-off-by: Luo Meng <luomeng12@xxxxxxxxxx>
---
  drivers/md/dm-mpath.c         |  2 +-
  drivers/md/dm-table.c         | 43 
+++++++++++++++++++++--------------
  include/linux/device-mapper.h |  2 ++
  3 files changed, 29 insertions(+), 18 deletions(-)

diff --git a/drivers/md/dm-mpath.c b/drivers/md/dm-mpath.c
index 0e325469a252..1f51059520ac 100644
--- a/drivers/md/dm-mpath.c
+++ b/drivers/md/dm-mpath.c
@@ -2001,7 +2001,7 @@ static int multipath_message(struct dm_target 
*ti, unsigned argc, char **argv,
          goto out;
      }
  -    r = dm_get_device(ti, argv[1], dm_table_get_mode(ti->table), 
&dev);
+    r = __dm_get_device(ti, argv[1], dm_table_get_mode(ti->table), 
&dev, false);
      if (r) {
          DMWARN("message: error getting device %s",
                 argv[1]);
diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index d8034ff0cb24..ad18a41f1608 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -338,12 +338,8 @@ dev_t dm_get_dev_t(const char *path)
  }
  EXPORT_SYMBOL_GPL(dm_get_dev_t);
  -/*
- * Add a device to the list, or just increment the usage count if
- * it's already present.
- */
-int dm_get_device(struct dm_target *ti, const char *path, fmode_t 
mode,
-          struct dm_dev **result)
+int __dm_get_device(struct dm_target *ti, const char *path, fmode_t 
mode,
+            struct dm_dev **result, bool create_dd)
  {
      int r;
      dev_t dev;
@@ -367,19 +363,21 @@ int dm_get_device(struct dm_target *ti, const 
char *path, fmode_t mode,
        dd = find_device(&t->devices, dev);
      if (!dd) {
-        dd = kmalloc(sizeof(*dd), GFP_KERNEL);
-        if (!dd)
-            return -ENOMEM;
-
-        if ((r = dm_get_table_device(t->md, dev, mode, 
&dd->dm_dev))) {
-            kfree(dd);
-            return r;
-        }
+        if (create_dd) {
+            dd = kmalloc(sizeof(*dd), GFP_KERNEL);
+            if (!dd)
+                return -ENOMEM;
  -        refcount_set(&dd->count, 1);
-        list_add(&dd->list, &t->devices);
-        goto out;
+            if ((r = dm_get_table_device(t->md, dev, mode, 
&dd->dm_dev))) {
+                kfree(dd);
+                return r;
+            }
  +            refcount_set(&dd->count, 1);
+            list_add(&dd->list, &t->devices);
+            goto out;
+        } else
+            return -ENODEV;
      } else if (dd->dm_dev->mode != (mode | dd->dm_dev->mode)) {
          r = upgrade_mode(dd, mode, t->md);
          if (r)
@@ -390,6 +388,17 @@ int dm_get_device(struct dm_target *ti, const 
char *path, fmode_t mode,
      *result = dd->dm_dev;
      return 0;
  }
+EXPORT_SYMBOL(__dm_get_device);
+
+/*
+ * Add a device to the list, or just increment the usage count if
+ * it's already present.
+ */
+int dm_get_device(struct dm_target *ti, const char *path, fmode_t 
mode,
+          struct dm_dev **result)
+{
+    return __dm_get_device(ti, path, mode, result, true);
+}
  EXPORT_SYMBOL(dm_get_device);
    static int dm_set_device_limits(struct dm_target *ti, struct 
dm_dev *dev,
diff --git a/include/linux/device-mapper.h 
b/include/linux/device-mapper.h
index 04c6acf7faaa..ef4031a844b6 100644
--- a/include/linux/device-mapper.h
+++ b/include/linux/device-mapper.h
@@ -178,6 +178,8 @@ dev_t dm_get_dev_t(const char *path);
  int dm_get_device(struct dm_target *ti, const char *path, fmode_t 
mode,
            struct dm_dev **result);
  void dm_put_device(struct dm_target *ti, struct dm_dev *d);
+int __dm_get_device(struct dm_target *ti, const char *path, fmode_t 
mode,
+            struct dm_dev **result, bool create_dd);
    /*
   * Information about a target type
--
2.31.1

This patch is treating the one multipath case like it is only consumer
of dm_get_device() that might have this issue. It feels too focused on
an instance where dm_get_device()'s side-effect of creating the device
is unwelcome. Feels like you're treating the symptom rather than the
problem (e.g. that the "kfree() in dm_put_table() is not protected"?)

Mike

In other cases, kfree() in dm_put_table() is protected by srcu.
For example:
          USE                              FREE
table_deps                            dev_remove
 dm_get_live_or_inactive_table         dm_sync_table
  // lock
  srcu_read_lock(&md->io_barrier)
                                        // wait for unlock
synchronize_srcu(&md->io_barrier)
  retrieve_deps
 dm_put_live_table
  // unlock
  srcu_read_unlock(&md->io_barrier...)
                                       dm_table_destroy
                                        linear_dtr
                                         dm_put_device
                                          dm_put_table_device

However, in multipath_message(), the dm_dev is created and destroyed
under srcu_read_lock, which means destroying dm_dev in this process
and using dm_dev in other process will happen at same time since both
of them hold srcu_read_lock.

target_message
 dm_get_live_table // lock
  multipath_message
   dm_get_device // created
    // may be got by other processes
   dm_put_device // destroyed
    // may be used by other processes
 dm_put_live_table // unlock

We figured out some other solutions:
1) Judge refcount of dd under some lock before using dm_dev;
2) Get dd from list and use dm_dev under rcu;
3) Implement dm_get_device_xxx() with reference to dm_get_device()
for dm-mpath to avoid creating dd when find_device() failed.

Compared to solutions above, Luo Meng's patch may be more appropriate.
Any suggestions will be appreciated.

Li

--
dm-devel mailing list
dm-devel@xxxxxxxxxx
https://listman.redhat.com/mailman/listinfo/dm-devel



--
dm-devel mailing list
dm-devel@xxxxxxxxxx
https://listman.redhat.com/mailman/listinfo/dm-devel







