Obtaining unparsed IEE80211 elements.

[Sh3Rm4n]

I have a usecase, where I'd like to obtain the raw Elements.

As the Nl80211Message is eagerly parsed and Nl80211Element is part of Nl80211BssInfo is part of Nl80211Attr is part of Nl80211Message, there is no easy way to obtain the elements unparsed.

I currently have two options to obtain the "unparsed" form of the elements.

1. Use emit again, which converts the IEE80211 elements back to the serialized form. This is where Fix emit inconsitencies and add tests #30 is needed because of some incosistency issues for emit implementations.
2. Use a custom Nl80211Message variant, which does only partially parse the payload (or doesn't at all), where the Nl80211Message::attributes field type is just Vec<u8> so that nothing is parsed by GenetlinkHandle::request.

What do you think about adding a switch somewhere (maybe in the Nl80211AttrsBuilder or even the concrete Nl80211AttrsBuilder<Nl80211Scan>), which only parses the message partially (if this is even technically possible).
E.g. maybe adding a flag/option cto force all IEs to be parsed as Other(Vec<u8>)?

Otherwise I could imagine contributing a little more convenience around a RawNl80211Message . WDYT?

[Sh3Rm4n]

I have another solution drafted in #38

This will replace Nl80211Elements from BssInfo and only provide a RawNl80211Elements type, which is a Vec<u8> wrapper.

Providing the IEEE Elements unparsed has some benefits:

• IEs are untrusted data directly comming from IEs and only passed on via the netlink API
• IEs are defined separately from the rest of the NL80211 API and splitting the parsing into another step reflects that
  • Renaming the type to IEEE80211Elements (see Nl80211Element should be IEEE80211Element #24) would also make this more obvious
• Separating the parsing into another step make it forward compatible to newer IEEE 802.11 standards.
• Obtaining the IEs as Vec<u8> could enable the possibility to parse them later on with a newer parser (another crate or even a future version of wl-nl80211)

[donnaskiez]

Hi,

A small nit, the term "Information Elements" isn't really used by the standards anymore, rather the term "elements" is used instead (for a while now).

[Sh3Rm4n]

Thanks for the heads up, I've updated the description accordingly.

[cathay4t]

How about we introduce Nl80211BssInfo::RawInformationElements(Vec<u8>) storing both parsed and unparsed IE there?

Both my project don't require parsed object. So if IE data is untrusted, then showing raw data only is better.

[Sh3Rm4n]

This is essentially what I did in https://github.com/rust-netlink/wl-nl80211/pull/38/files, WDYT?