[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-Id: <BFA64A65-43C5-4896-A65D-552181DAEF7F@cmss.chinamobile.com>
Date: Sat, 15 Sep 2018 09:22:19 +0800
From: Haishuang Yan <yanhaishuang@...s.chinamobile.com>
To: Edward Cree <ecree@...arflare.com>
Cc: David Miller <davem@...emloft.net>, kuznet@....inr.ac.ru,
jbenc@...hat.com, netdev@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2,net-next 1/2] ip_gre: fix parsing gre header in
ipgre_err
> On 2018年9月14日, at 下午8:44, Edward Cree <ecree@...arflare.com> wrote:
>
> On 13/09/18 18:58, David Miller wrote:
>> From: Haishuang Yan <yanhaishuang@...s.chinamobile.com>
>> Date: Wed, 12 Sep 2018 17:21:21 +0800
>>
>>> @@ -86,7 +86,7 @@ int gre_parse_header(struct sk_buff *skb, struct tnl_ptk_info *tpi,
>>>
>>> options = (__be32 *)(greh + 1);
>>> if (greh->flags & GRE_CSUM) {
>>> - if (skb_checksum_simple_validate(skb)) {
>>> + if (csum_err && skb_checksum_simple_validate(skb)) {
>>> *csum_err = true;
>>> return -EINVAL;
>>> }
>> You want to ignore csum errors, but you do not want to elide the side
>> effects of the skb_checksum_simple_validate() call which are to set
>> skb->csum_valid and skb->csum.
>>
>> Therefore, the skb_checksum_simple_validate() call still needs to be
>> performed. We just wont return -EINVAL in the NULL csum_err case.
>
> How about just reversing the order of the AND?
>
> if (skb_checksum_simple_validate(skb) && csum_err) {
> *csum_err = true;
> return -EINVAL;
> }
>
>
It looks good to me, thanks!
But skb_checksum_try_convert only need to be called after the checksum is
validated, so I suggested a better solution as following:
89 if (!skb_checksum_simple_validate(skb)) {
90 skb_checksum_try_convert(skb, IPPROTO_GRE, 0,
91 null_compute_pseudo);
92 } else if (csum_err) {
93 *csum_err = true;
94 return -EINVAL;
95 }
Powered by blists - more mailing lists