[BUG] ethernet:enic: A use after free bug in enic_hard_start_xmit

[BUG] ethernet:enic: A use after free bug in enic_hard_start_xmit

[lyl2019]

Hi, maintainers.
    Our code analyzer reported a uaf bug, but it is a little
difficult for me to fix it directly.

File: drivers/net/ethernet/cisco/enic/enic_main.c
In enic_hard_start_xmit, it calls enic_queue_wq_skb(). Inside
enic_queue_wq_skb, if some error happens, the skb will be freed
by dev_kfree_skb(skb). But the freed skb is still used in 
skb_tx_timestamp(skb).

```
	enic_queue_wq_skb(enic, wq, skb);// skb could freed here

	if (vnic_wq_desc_avail(wq) < MAX_SKB_FRAGS + ENIC_DESC_MAX_SPLITS)
		netif_tx_stop_queue(txq);
	skb_tx_timestamp(skb); // freed skb is used here.
```
Bug introduced by fb7516d42478e ("enic: add sw timestamp support").

Thanks for your time, looking forwarding to your reply.
Lv Yunlong <lyl2019@mail.ustc.edu.cn>

Re: [BUG] ethernet:enic: A use after free bug in enic_hard_start_xmit

[Govindarajulu Varadarajan]

On Tue, 2021-04-27 at 10:55 +0800, lyl2019@mail.ustc.edu.cn wrote:
> Hi, maintainers.
>     Our code analyzer reported a uaf bug, but it is a little
> difficult for me to fix it directly.
> 
> File: drivers/net/ethernet/cisco/enic/enic_main.c
> In enic_hard_start_xmit, it calls enic_queue_wq_skb(). Inside
> enic_queue_wq_skb, if some error happens, the skb will be freed
> by dev_kfree_skb(skb). But the freed skb is still used in 
> skb_tx_timestamp(skb).
> 
> ```
>         enic_queue_wq_skb(enic, wq, skb);// skb could freed here
> 
>         if (vnic_wq_desc_avail(wq) < MAX_SKB_FRAGS + ENIC_DESC_MAX_SPLITS)
>                 netif_tx_stop_queue(txq);
>         skb_tx_timestamp(skb); // freed skb is used here.
> ```
> Bug introduced by fb7516d42478e ("enic: add sw timestamp support").

Thank you for reporting this.

One solution is to make enic_queue_wq_skb() return error and goto spin_unlock()
incase of error.

Would you like to send the fix for this?

--
Govind

Re: Re: [BUG] ethernet:enic: A use after free bug in enic_hard_start_xmit

[lyl2019]

> -----原始邮件-----
> 发件人: "Govindarajulu Varadarajan" <govind.varadar@gmail.com>
> 发送时间: 2021-05-01 08:10:02 (星期六)
> 收件人: lyl2019@mail.ustc.edu.cn, benve@cisco.com, davem@davemloft.net, kuba@kernel.org
> 抄送: netdev@vger.kernel.org, linux-kernel@vger.kernel.org
> 主题: Re: [BUG] ethernet:enic: A use after free bug in enic_hard_start_xmit
> 
> On Tue, 2021-04-27 at 10:55 +0800, lyl2019@mail.ustc.edu.cn wrote:
> > Hi, maintainers.
> >     Our code analyzer reported a uaf bug, but it is a little
> > difficult for me to fix it directly.
> > 
> > File: drivers/net/ethernet/cisco/enic/enic_main.c
> > In enic_hard_start_xmit, it calls enic_queue_wq_skb(). Inside
> > enic_queue_wq_skb, if some error happens, the skb will be freed
> > by dev_kfree_skb(skb). But the freed skb is still used in 
> > skb_tx_timestamp(skb).
> > 
> > ```
> >         enic_queue_wq_skb(enic, wq, skb);// skb could freed here
> > 
> >         if (vnic_wq_desc_avail(wq) < MAX_SKB_FRAGS + ENIC_DESC_MAX_SPLITS)
> >                 netif_tx_stop_queue(txq);
> >         skb_tx_timestamp(skb); // freed skb is used here.
> > ```
> > Bug introduced by fb7516d42478e ("enic: add sw timestamp support").
> 
> Thank you for reporting this.
> 
> One solution is to make enic_queue_wq_skb() return error and goto spin_unlock()
> incase of error.
> 
> Would you like to send the fix for this?
> 
> --
> Govind
> 

It is my pleasure, i have submitted this fix.
See https://lore.kernel.org/patchwork/patch/1420692/.

Thanks.