netdev - Re: [PATCHv3 net-next 2/5] {IPv4,xfrm} Add ESN support for AH ingress part

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: Windows password security audit tool. GUI, reports in PDF.

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

Message-ID: <52D634CE.8020501@windriver.com>
Date:	Wed, 15 Jan 2014 15:12:14 +0800
From:	Fan Du <fan.du@...driver.com>
To:	Steffen Klassert <steffen.klassert@...unet.com>
CC:	<davem@...emloft.net>, <netdev@...r.kernel.org>
Subject: Re: [PATCHv3 net-next 2/5] {IPv4,xfrm} Add ESN support for AH ingress
 part



On 2014年01月14日 18:34, Steffen Klassert wrote:
> On Tue, Jan 14, 2014 at 06:17:26PM +0800, Fan Du wrote:
>>
>>
>> On 2014年01月14日 18:09, Steffen Klassert wrote:
>>> On Tue, Jan 14, 2014 at 06:01:32PM +0800, Fan Du wrote:
>>>>
>>>>
>>>> On 2014年01月14日 17:54, Steffen Klassert wrote:
>>>>> On Tue, Jan 14, 2014 at 09:39:09AM +0800, Fan Du wrote:
>>>>>> @@ -381,7 +393,14 @@ static int ah_input(struct xfrm_state *x, struct sk_buff *skb)
>>>>>>   	sg_init_table(sg, nfrags);
>>>>>>   	skb_to_sgvec(skb, sg, 0, skb->len);
>>>>>>
>>>>>> -	ahash_request_set_crypt(req, sg, icv, skb->len);
>>>>>> +	if (x->props.flags&    XFRM_STATE_ESN) {
>>>>>> +		sg_unmark_end(&sg[nfrags - 1]);
>>>>>> +		/* Attach seqhi sg right after packet payload */
>>>>>> +		*seqhi = htonl(XFRM_SKB_CB(skb)->seq.output.hi);
>>>>>
>>>>> This is ah_input(), so you should use the high bits of the input
>>>>> sequence number here. The ipv6 patch has the same problem.
>>>>
>>>> ok, I will fix this.
>>>>
>>>>>
>>>>>> +		sg_init_table(seqhisg, sglists);
>>>>>
>>>>> Why do you add a separate SG table for this?
>>>>
>>>> It just initialize a single seqhisg, which is actually followed behind packet payload sg table.
>>>> initialized seqhisg actually mark itself as the end of sg list.
>>>>
>>>
>>> Why don't you just add this entry to the existing SG table?
>>>
>>
>> Do you mean scatterwalk_crypto_chain ?
>
> No, I mean something like:
>
> sg_init_table(sg, nfrags + sglists)
>
> if (x->props.flags&  XFRM_STATE_ESN) {
> 	*seqhi = XFRM_SKB_CB(skb)->seq.input.hi;
> 	sg_set_buf(sg + nfrags, seqhi, seqhi_len);
> }
>

Calling skb_to_sgvec to map first part of payload into global sg list will mark the sg
which contains last data of payload as new end, that's side effect of skb_to_sgvec.
This doesn't work well unless calling sg_unmark_end again to revert such effect.

Here we need another method to map payload to sg list without changing the end sg.

-- 
浮沉随浪只记今朝笑

--fan
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html