Re: [net-next PATCH v2 1/5] introduce IFE action

[Cong Wang <xiyou.wangcong@gmail.com>]

On Wed, Feb 24, 2016 at 5:09 AM, Jamal Hadi Salim <jhs@mojatatu.com> wrote:
> On 16-02-23 04:44 PM, Cong Wang wrote:
>>
>> On Tue, Feb 23, 2016 at 4:49 AM, Jamal Hadi Salim <jhs@mojatatu.com>
>> wrote:
>
>
>>> +#define MODULE_ALIAS_IFE_META(metan)   MODULE_ALIAS("ifemeta"
>>> __stringify_1(metan))
>>> +
>>> +int get_meta_u32(struct sk_buff *skb, struct tcf_meta_info *mi);
>>> +int get_meta_u16(struct sk_buff *skb, struct tcf_meta_info *mi);
>>> +int tlv_meta_encode(void *skbdata, u16 attrtype, u16 dlen, const void
>>> *dval);
>>> +int alloc_meta_u32(struct tcf_meta_info *mi, void *metaval);
>>> +int alloc_meta_u16(struct tcf_meta_info *mi, void *metaval);
>>> +int check_meta_u32(u32 metaval, struct tcf_meta_info *mi);
>>> +int encode_meta_u32(u32 metaval, void *skbdata, struct tcf_meta_info
>>> *mi);
>>> +int validate_meta_u32(void *val, int len);
>>> +int validate_meta_u16(void *val, int len);
>>> +void release_meta_gen(struct tcf_meta_info *mi);
>>> +int register_ife_op(struct tcf_meta_ops *mops);
>>> +int unregister_ife_op(struct tcf_meta_ops *mops);
>>
>>
>>
>> These are exported to other modules, please add some prefix to protect
>> them.
>>
>
> suggestion? I thought they seemed unique enough.
>

I would pick "ife_", for example, ife_get_meta_u32() ...


>
>>> + * copyright   Jamal Hadi Salim (2015)
>>
>>
>>
>> 2016? ;)
>>
>
> Yes. This code has been around since then. Actually earlier than that
> running. But i formatted it for kernel inclusion in about first week
> of January. Dave asked me to wait for the ethertype to get it in.
> The IETF still hasnt come through a year later and so i re-submitted
> with no default ethertype. I think 2015 is deserving here, no?
> I hope i dont spend another year discussing on the list ;-> /me runs

I thought it should be always the time when you submit the code, but
you are the author you have the right to make it whatever you want... ;)

[...]

>>> +       if ((parm->flags & IFE_ENCODE) && (parm->flags & IFE_DECODE)) {
>>> +               pr_info("Ambigous: Cant have both encode and decode\n");
>>> +               return -EINVAL;
>>> +       }
>>
>>
>>
>> Is it possible to fold them into one bit?
>
>
> As in the check you mean?
>

I meant to suggest to make IFE_ENCODE and IFE_DECODE share
one bit.