[lkml]   [2015]   [Mar]   [13]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [PATCH net-next 1/2] bpf: allow extended BPF programs access skb fields
On 03/13/2015 05:22 PM, Alexei Starovoitov wrote:
> On 3/13/15 2:57 AM, Daniel Borkmann wrote:
>> On 03/13/2015 03:21 AM, Alexei Starovoitov wrote:
>>> introduce user accessible mirror of in-kernel 'struct sk_buff':
>> For each member, I'd also add BUILD_BUG_ON()s similarly as we have in
>> convert_bpf_extensions(). That way, people won't forget to adjust the
>> code.
> I thought about it, but didn't add it, since we already have them
> in the same file several lines above this spot.
> I think one build error per .c file should be enough to attract
> attention.
> Though I'll add a comment to convert_bpf_extensions() that build_bug_on
> errors should be addressed in two places.

My concern would be that in case the code gets changed, this spot
could easily be overlooked by someone possibly unfamiliar with the
code, since no build error triggers there.

So I guess it wouldn't hurt or cost us much to also adding the
BUILD_BUG_ON()s there instead of a comment.

>> The remaining fields we export in classic BPF would be skb->hash,
>> skb->protocol, skb->vlan_tci, are we adding them as well to match
>> up functionality with classic BPF? For example, I can see hash being
>> useful as a key to be used with eBPF maps, etc.
> I want to do skb->protocol and skb->vlan_tci differently and hopefully
> more generically than classic did them, so they will come in the
> follow on patches. skb->hash also should be done differently.
> So yes. All of them are subjects of future patches/discussions.

Ok, sounds good.


 \ /
  Last update: 2015-03-13 18:01    [W:0.038 / U:6.040 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site