diff options
author | Lv Yunlong <lyl2019@mail.ustc.edu.cn> | 2021-04-26 19:06:25 +0300 |
---|---|---|
committer | David S. Miller <davem@davemloft.net> | 2021-04-26 23:07:30 +0300 |
commit | 6d72e7c767acbbdd44ebc7d89c6690b405b32b57 (patch) | |
tree | 7aaccf01bd1be481d9662bee62901264654a91fc /drivers/net/ethernet/qualcomm/emac | |
parent | f77bd544a6bbe69aa50d9ed09f13494cf36ff806 (diff) | |
download | linux-6d72e7c767acbbdd44ebc7d89c6690b405b32b57.tar.xz |
net:emac/emac-mac: Fix a use after free in emac_mac_tx_buf_send
In emac_mac_tx_buf_send, it calls emac_tx_fill_tpd(..,skb,..).
If some error happens in emac_tx_fill_tpd(), the skb will be freed via
dev_kfree_skb(skb) in error branch of emac_tx_fill_tpd().
But the freed skb is still used via skb->len by netdev_sent_queue(,skb->len).
As i observed that emac_tx_fill_tpd() haven't modified the value of skb->len,
thus my patch assigns skb->len to 'len' before the possible free and
use 'len' instead of skb->len later.
Fixes: b9b17debc69d2 ("net: emac: emac gigabit ethernet controller driver")
Signed-off-by: Lv Yunlong <lyl2019@mail.ustc.edu.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'drivers/net/ethernet/qualcomm/emac')
-rw-r--r-- | drivers/net/ethernet/qualcomm/emac/emac-mac.c | 4 |
1 files changed, 3 insertions, 1 deletions
diff --git a/drivers/net/ethernet/qualcomm/emac/emac-mac.c b/drivers/net/ethernet/qualcomm/emac/emac-mac.c index 117188e3c7de..87b8c032195d 100644 --- a/drivers/net/ethernet/qualcomm/emac/emac-mac.c +++ b/drivers/net/ethernet/qualcomm/emac/emac-mac.c @@ -1437,6 +1437,7 @@ netdev_tx_t emac_mac_tx_buf_send(struct emac_adapter *adpt, { struct emac_tpd tpd; u32 prod_idx; + int len; memset(&tpd, 0, sizeof(tpd)); @@ -1456,9 +1457,10 @@ netdev_tx_t emac_mac_tx_buf_send(struct emac_adapter *adpt, if (skb_network_offset(skb) != ETH_HLEN) TPD_TYP_SET(&tpd, 1); + len = skb->len; emac_tx_fill_tpd(adpt, tx_q, skb, &tpd); - netdev_sent_queue(adpt->netdev, skb->len); + netdev_sent_queue(adpt->netdev, len); /* Make sure the are enough free descriptors to hold one * maximum-sized SKB. We need one desc for each fragment, |