I thought a little bit more about the part of reporting the broken and missing attributes, and I'm not sure I'm entirely happy with that part. For reporting broken attributes, I think the pointer is really the best thing we can do - adding the attribute number in addition to that would seem to be easy to misuse in userspace, and there's no information about nesting at that point and the "malformed attribute" ID will not make any sense without knowing the nesting level. So I think we shouldn't add the attribute ID there. The userspace app can disentangle the nesting, or it can know that it never used nesting, for example. Similarly, the "missing attribute" ID - which I did add an attribute for but didn't use it yet - can result in similar interpretation problems. I'm not really sure what to do with this one - technically one could report back the entire nesting chain, but having that kind of state might be really difficult. Did anyone already have any plans for the "missing attribute" functionality? And another thought regarding the strict checking: we might also want to extend that to strictly check the *size* of attributes as well. For example, right now, if the kernel wants a u8 attribute but userspace uses a u16 or u32, this will work on little-endian and be accepted but wrongly be 0 on big endian. johannes
