Garrett D'Amore writes: > Actually, quite apart from bit-rot, there is one compelling reason to > remove the code out _now_ if nobody is using. That is that this code > inserts somethings on a few hotpaths that have a measurable negative > impact on performance. Removing this code will improve the performance > of IP forwarding in certain pathological cases (64-byte packets).
There's more than one such reason. The code also adds quite a few never-tested edge cases to the kernel. If we care about stability or security, that's not a good thing, as complexity is the enemy here. Also, every new project that touches those paths must (uselessly) consider how that functionality should evolve to support the new feature. Of course, there are other such invasive bits in the stack (CGTP, IPMP, TX, and Zones are good examples), but it's not as though we need more. I find those arguments more compelling than just how fast we can forward TCP ACKs and interactive TELNET. ;-} -- James Carlson, Solaris Networking <james.d.carlson at sun.com> Sun Microsystems / 1 Network Drive 71.232W Vox +1 781 442 2084 MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677
