Re: Headerless packets hitting ethernet?

Paul Rusty Russell (Paul.Russell@rustcorp.com.au)
Mon, 27 Sep 1999 09:46:35 +0930


In message <199909261613.UAA09105@ms2.inr.ac.ru> you write:
> You found where sk_buff is located eventually, did not you?
> Please, do tell me that it was difficult task. 8)8)

Alexey, it's a simple optimization problem. The reader has a finite
cache; every time she's forced to look up a function which has a wierd
name, grep for a definition, or backtrack because she chose the wrong
code branch, there's a chance that some other definition gets pushed
out of the cache. If it's bad enough, the reader's mind will thrash
and no progress will be made (at this point, human livelock avoidance
algorithms usually make them install FreeBSD).

Strategy-style function pointers in structs make the code hard to
follow, hence the VFS convention of calling the `foo' strategy
xxx_foo, so you can easily follow it (eg. grep works).

Would you look at cleanup patches to do this against the latest
softnet? I'm at least thinking dst_output and struct packet_type's
`func', maybe others I find.

It'd be a start,
Rusty.

--
Hacking time.

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/