[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <3d26ab3e-2a3c-4c36-b165-06a1029bb0b0@engleder-embedded.com>
Date: Fri, 11 Oct 2024 22:41:31 +0200
From: Gerhard Engleder <gerhard@...leder-embedded.com>
To: Wang Hai <wanghai38@...wei.com>
Cc: netdev@...r.kernel.org, linux-kernel@...r.kernel.org,
andreas@...sler.com, davem@...emloft.net, kuba@...nel.org,
pabeni@...hat.com, edumazet@...gle.com, kristoffer@...sler.com
Subject: Re: [PATCH net] net: ethernet: aeroflex: fix potential memory leak in
greth_start_xmit_gbit()
On 11.10.24 13:39, Wang Hai wrote:
> The greth_start_xmit_gbit() returns NETDEV_TX_OK without freeing skb
> in case of skb->len being too long, add dev_kfree_skb() to fix it.
>
> Fixes: d4c41139df6e ("net: Add Aeroflex Gaisler 10/100/1G Ethernet MAC driver")
> Signed-off-by: Wang Hai <wanghai38@...wei.com>
> ---
> drivers/net/ethernet/aeroflex/greth.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/net/ethernet/aeroflex/greth.c b/drivers/net/ethernet/aeroflex/greth.c
> index 27af7746d645..8f6835a710b9 100644
> --- a/drivers/net/ethernet/aeroflex/greth.c
> +++ b/drivers/net/ethernet/aeroflex/greth.c
> @@ -484,6 +484,7 @@ greth_start_xmit_gbit(struct sk_buff *skb, struct net_device *dev)
>
> if (unlikely(skb->len > MAX_FRAME_SIZE)) {
> dev->stats.tx_errors++;
> + dev_kfree_skb(skb);
> goto out;
dev_kfree_skb(skb) is already part of the error handling, one line above
the "out" label. Why don't you just add another label which includes
dev_kfree_skb(skb) and goto that label?
Gerhard
Powered by blists - more mailing lists