netdev - Re: [PATCH v3] myri10ge: fix an incorrect free for skb in myri10ge_sw

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <857f3f9c-abc3-779e-d03b-76c23f6e13af@suse.de>
Date:   Wed, 6 Apr 2022 15:33:49 +0300
From:   Denis Kirjanov <dkirjanov@...e.de>
To:     Xiaomeng Tong <xiam0nd.tong@...il.com>, christopher.lee@...i.com,
        davem@...emloft.net, kuba@...nel.org, pabeni@...hat.com
Cc:     netdev@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v3] myri10ge: fix an incorrect free for skb in
 myri10ge_sw_tso



4/6/22 06:55, Xiaomeng Tong пишет:
> All remaining skbs should be released when myri10ge_xmit fails to
> transmit a packet. Fix it within another skb_list_walk_safe.
> 
> Signed-off-by: Xiaomeng Tong <xiam0nd.tong@...il.com>
> ---
> 
> changes since v2:
>   - free all remaining skbs. (Xiaomeng Tong)
> 
> changes since v1:
>   - remove the unneeded assignmnets. (Xiaomeng Tong)
> 
> v2:https://lore.kernel.org/lkml/20220405000553.21856-1-xiam0nd.tong@gmail.com/
> v1:https://lore.kernel.org/lkml/20220319052350.26535-1-xiam0nd.tong@gmail.com/
> 
> ---
>   drivers/net/ethernet/myricom/myri10ge/myri10ge.c | 6 ++----
>   1 file changed, 2 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/net/ethernet/myricom/myri10ge/myri10ge.c b/drivers/net/ethernet/myricom/myri10ge/myri10ge.c
> index 50ac3ee2577a..21d2645885ce 100644
> --- a/drivers/net/ethernet/myricom/myri10ge/myri10ge.c
> +++ b/drivers/net/ethernet/myricom/myri10ge/myri10ge.c
> @@ -2903,11 +2903,9 @@ static netdev_tx_t myri10ge_sw_tso(struct sk_buff *skb,
>   		status = myri10ge_xmit(curr, dev);
>   		if (status != 0) {
>   			dev_kfree_skb_any(curr);
> -			if (segs != NULL) {
> -				curr = segs;
> -				segs = next;
> +			skb_list_walk_safe(next, curr, next) {
>   				curr->next = NULL;
> -				dev_kfree_skb_any(segs);
> +				dev_kfree_skb_any(curr);

why can't we just do the following?
         skb_list_walk_safe(segs, skb, nskb) {
                 status = myri10ge_xmit(curr, dev);
                 if (err)
                         break;

         }

         /* Free all of the segments. */
         skb_list_walk_safe(segs, skb, nskb) {
                 if (err)
                         kfree_skb(skb);
                 else
                         consume_skb(skb);
         }
         return err;


>   			}
>   			goto drop;
>   		}