net: fix a race in gro_cell_poll()

Dmitry Kravkov reported packet drops for GRE packets since GRO support
was added.

There is a race in gro_cell_poll() because we call napi_complete()
without any synchronization with a concurrent gro_cells_receive()

Once bug was triggered, we queued packets but did not schedule NAPI
poll.

We can fix this issue using the spinlock protected the napi_skbs queue,
as we have to hold it to perform skb dequeue anyway.

As we open-code skb_dequeue(), we no longer need to mask IRQS, as both
producer and consumer run under BH context.

Bug added in commit c9e6bc644e (net: add gro_cells infrastructure)

Reported-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Tested-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/include/net/gro_cells.h b/include/net/gro_cells.h
index 4fd8a4b4b7eeee98198c2e999d2a72951d95f8fa..e5062c955ea6594c3d7bd7850b76bce6a3640b5a 100644 (file)
--- a/include/net/gro_cells.h
+++ b/include/net/gro_cells.h
@@ -17,7 +17,6 @@ struct gro_cells {
 
 static inline void gro_cells_receive(struct gro_cells *gcells, struct sk_buff *skb)
 {
-       unsigned long flags;
        struct gro_cell *cell = gcells->cells;
        struct net_device *dev = skb->dev;
 
@@ -35,32 +34,37 @@ static inline void gro_cells_receive(struct gro_cells *gcells, struct sk_buff *s
                return;
        }
 
-       spin_lock_irqsave(&cell->napi_skbs.lock, flags);
+       /* We run in BH context */
+       spin_lock(&cell->napi_skbs.lock);
 
        __skb_queue_tail(&cell->napi_skbs, skb);
        if (skb_queue_len(&cell->napi_skbs) == 1)
                napi_schedule(&cell->napi);
 
-       spin_unlock_irqrestore(&cell->napi_skbs.lock, flags);
+       spin_unlock(&cell->napi_skbs.lock);
 }
 
+/* called unser BH context */
 static inline int gro_cell_poll(struct napi_struct *napi, int budget)
 {
        struct gro_cell *cell = container_of(napi, struct gro_cell, napi);
        struct sk_buff *skb;
        int work_done = 0;
 
+       spin_lock(&cell->napi_skbs.lock);
        while (work_done < budget) {
-               skb = skb_dequeue(&cell->napi_skbs);
+               skb = __skb_dequeue(&cell->napi_skbs);
                if (!skb)
                        break;
-
+               spin_unlock(&cell->napi_skbs.lock);
                napi_gro_receive(napi, skb);
                work_done++;
+               spin_lock(&cell->napi_skbs.lock);
        }
 
        if (work_done < budget)
                napi_complete(napi);
+       spin_unlock(&cell->napi_skbs.lock);
        return work_done;
 }