wimax/i2400m: fix bad race condition check in RX path

The i2400m->rx_roq data structure is protected against race conditions
with a reference count (i2400m->rx_roq_refcount); the pointer can be
read-referenced under the i2400m->rx_lock spinlock.

The code in i2400m_rx_edata() wasn't properly following access
protocol, performing an invalid check on i2400m->rx_roq (which is
cleared to NULL when the refcount drops to zero). As such, it was
missing to detect when the data structure is no longer valid and
oopsing with a NULL pointer dereference.

This commit fixes said check by verifying, under the rx_lock spinlock,
that i2400m->rx_roq is non-NULL and then increasing the reference
count before dropping the spinlock.

Signed-off-by: Inaky Perez-Gonzalez <inaky.perez-gonzalez@intel.com>

diff --git a/drivers/net/wimax/i2400m/rx.c b/drivers/net/wimax/i2400m/rx.c
index c835ae8b89ce6c8555589ffc5245875ededdbda8..6cd12d64fc7ac7ced57149e244e2fc4660e4b081 100644 (file)
--- a/drivers/net/wimax/i2400m/rx.c
+++ b/drivers/net/wimax/i2400m/rx.c
@@ -1033,12 +1033,12 @@ void i2400m_rx_edata(struct i2400m *i2400m, struct sk_buff *skb_rx,
                ro_sn = (reorder >> I2400M_RO_SN_SHIFT) & I2400M_RO_SN;
 
                spin_lock_irqsave(&i2400m->rx_lock, flags);
-               roq = &i2400m->rx_roq[ro_cin];
-               if (roq == NULL) {
+               if (i2400m->rx_roq == NULL) {
                        kfree_skb(skb); /* rx_roq is already destroyed */
                        spin_unlock_irqrestore(&i2400m->rx_lock, flags);
                        goto error;
                }
+               roq = &i2400m->rx_roq[ro_cin];
                kref_get(&i2400m->rx_roq_refcount);
                spin_unlock_irqrestore(&i2400m->rx_lock, flags);