Amit, I noticed following bug in qlnic driver.

Also, skb->truesize should not be changed by drivers, unless dealing
with fragments.

When you have :
	skb->truesize = skb->len + sizeof(struct sk_buff);

you are lying to stack that can queue many "small" UDP packets,
accouting for small packets in socket rcvbuf, while the truesize was
really 1532 + sizeof(struct sk_buff)

Could you take a look ?

Thanks


[PATCH] qlcnic: dont assume NET_IP_ALIGN is 2

qlcnic driver allocates rx skbs and gives to hardware too bytes of extra
storage, allowing for corruption of kernel data.

NET_IP_ALIGN being 0 on some platforms (including x86), drivers should
not assume it's 2.

rds_ring->skb_size = rds_ring->dma_size + NET_IP_ALIGN;
...
skb = dev_alloc_skb(rds_ring->skb_size);
skb_reserve(skb, 2);
pci_map_single(pdev, skb->data, rds_ring->dma_size, PCI_DMA_FROMDEVICE);

(and rds_ring->skb_size == rds_ring->dma_size) -> bug


Because of extra alignment (1500 + 32) -> four extra bytes are available
before the struct skb_shared_info, so corruption is not noticed.

Note: this driver could use netdev_alloc_skb_ip_align()

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
---
diff --git a/drivers/net/qlcnic/qlcnic_init.c b/drivers/net/qlcnic/qlcnic_init.c
index 75ba744..60ab753 100644
--- a/drivers/net/qlcnic/qlcnic_init.c
+++ b/drivers/net/qlcnic/qlcnic_init.c
@@ -1316,7 +1316,7 @@ qlcnic_alloc_rx_skb(struct qlcnic_adapter *adapter,
 		return -ENOMEM;
 	}
 
-	skb_reserve(skb, 2);
+	skb_reserve(skb, NET_IP_ALIGN);
 
 	dma = pci_map_single(pdev, skb->data,
 			rds_ring->dma_size, PCI_DMA_FROMDEVICE);


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html