diff options
| author | Frediano Ziglio <frediano.ziglio@huawei.com> | 2014-10-09 11:45:39 +0100 |
|---|---|---|
| committer | Frediano Ziglio <frediano.ziglio@huawei.com> | 2014-10-09 11:46:08 +0100 |
| commit | d0909cd184404d9a78e96225f69e109c1eafa256 (patch) | |
| tree | aa18e4cbc8fe848c71c5e31ecf4e578cd3f37440 | |
| parent | 55dac0371cd75c25063ba91cb12595927679b00b (diff) | |
ARM: hisi: fix wrong unmapping in hip04 eth driver
rx_phys array is initially initialized with zeroes. However it could be
that in hip04_rx_poll addresses are still zero and we try to unmap them.
For dma addresses 0 is considered perfectly fine so on some condition
(for instance under Xen) the unmapping can cause a core.
Signed-off-by: Frediano Ziglio <frediano.ziglio@huawei.com>
| -rw-r--r-- | drivers/net/ethernet/hisilicon/hip04_eth.c | 8 |
1 files changed, 5 insertions, 3 deletions
diff --git a/drivers/net/ethernet/hisilicon/hip04_eth.c b/drivers/net/ethernet/hisilicon/hip04_eth.c index 7f01b910349e..7e2b8f38ba7e 100644 --- a/drivers/net/ethernet/hisilicon/hip04_eth.c +++ b/drivers/net/ethernet/hisilicon/hip04_eth.c @@ -420,9 +420,11 @@ static int hip04_rx_poll(struct napi_struct *napi, int budget) if (unlikely(!skb)) net_dbg_ratelimited("build_skb failed\n"); - dma_unmap_single(&ndev->dev, priv->rx_phys[priv->rx_head], - RX_BUF_SIZE, DMA_FROM_DEVICE); - priv->rx_phys[priv->rx_head] = 0; + if (priv->rx_phys[priv->rx_head]) { + dma_unmap_single(&ndev->dev, priv->rx_phys[priv->rx_head], + RX_BUF_SIZE, DMA_FROM_DEVICE); + priv->rx_phys[priv->rx_head] = 0; + } desc = (struct rx_desc *)skb->data; len = be16_to_cpu(desc->pkt_len); |