Thread (3 messages) 3 messages, 3 authors, 2020-01-22

Re: [v2] nbd: fix potential NULL pointer fault in nbd_genl_disconnect

From: Josef Bacik <josef@toxicpanda.com>
Date: 2020-01-21 14:09:31
Also in: lkml

On 1/20/20 7:45 AM, Sun Ke wrote:
quoted hunk ↗ jump to hunk
Open /dev/nbdX first, the config_refs will be 1 and
the pointers in nbd_device are still null. Disconnect
/dev/nbdX, then reference a null recv_workq. The
protection by config_refs in nbd_genl_disconnect is useless.

To fix it, just add a check for a non null task_recv in
nbd_genl_disconnect.

Signed-off-by: Sun Ke <redacted>
---
v1 -> v2:

add an omitted mutex_unlock.
---
  drivers/block/nbd.c | 4 ++++
  1 file changed, 4 insertions(+)
diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c
index b4607dd96185..668bc9cb92ed 100644
--- a/drivers/block/nbd.c
+++ b/drivers/block/nbd.c
@@ -2008,6 +2008,10 @@ static int nbd_genl_disconnect(struct sk_buff *skb, struct genl_info *info)
  		       index);
  		return -EINVAL;
  	}
+	if (!nbd->task_recv) {
+		mutex_unlock(&nbd_index_mutex);
+		return -EINVAL;
+	}
  	if (!refcount_inc_not_zero(&nbd->refs)) {
  		mutex_unlock(&nbd_index_mutex);
  		printk(KERN_ERR "nbd: device at index %d is going down\n",
This doesn't even really protect us, we need to have the nbd->config_lock held 
here to make sure it's ok.  The IOCTL path is safe because it creates the device 
on open so it's sure to exist by the time we get to the disconnect, we don't 
have that for genl_disconnect.  So I'd add the config_mutex before getting the 
config_ref, and then do the check, something like

mutex_lock(&nbd->config_lock);
if (!refcount_inc_not_zero(&nbd->refs)) {
}
if (!nbd->recv_workq) {
}
mutex_unlock(&nbd->config_lock);

Thanks,

Josef
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help