Re: [v4 PATCH 1/2] NETFILTER module xt_hmark, new target for HASH based fwmark

[Patrick McHardy <kaber@trash.net>]

On 12/01/2011 12:05 PM, Hans Schillstrom wrote:
>>>> And final question, why not simply use ipv6_skip_exthdr()?
>>> problems with fragments...
>> So the probem is that it will return the transport layer protocol
>> header for fragments with frag_off == 0? We also have ipv6_find_hdr()
>> which we could modify to indicate this in the frag_off pointer.
> ipv6_find_hdr() will do the trick with a light modification
> What about a wrapper like:
>
> int __ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset,
> 		  int target, unsigned short *fragoff,  int *fragflg)
> {
> ...
> 		if (nexthdr == NEXTHDR_FRAGMENT) {
> 			unsigned short _frag_off;
> 			__be16 *fp;
>
> 			if (fragflg)
> 			        fragflg = 1;
> 			fp = skb_header_pointer(skb,
> 						start+offsetof(struct frag_hdr,
> 							       frag_off),
> 						sizeof(_frag_off),
> 						&_frag_off);
>
> ...
> }
>
> int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset,
> 		  int target, unsigned short *fragoff)
> {
>          return __ipv6_find_hdr(skb, offset, terget, fragoff, NULL);
> }

Hmm that would require to change all current callers. I was more
thinking of unconditionally setting *frag_off in case of
fragments, then you can initialize it to some impossible value
like 0xffff and determine the presence of a fragment header
based on its value after calling ipv6_find_hdr().