Warner> : Makes you wonder why everyone bases their stuff on the
Warner> berkeley stack, : it sucks rocks from a performance
Warner> standpoint, and the speed problems : are inherent in it's
Warner> overall design and structure (read this as: mbufs : are
Warner> fundamentally flawed)... but we have some buglets left in
Warner> our : stuff so maybe I should cut myself short right here
Warner> ;)
Warner> While there are bugs in the now current linux code, there
Warner> are some big names that trashed mbufs. Van Jacobson has a
Warner> BSD based stack that he's heavily hacked to *NOT* use
Warner> mbufs, but a scheme that escapes me at the moment.
He called them pbufs. They are continuous memory buffers that are
alocated by devices on input... now, this model seams familiar doesn't,
it ? ;-)
Warner> He also did the hyperoptimizations of the checksum
Warner> calculations for the m68k (I believe).
Warner> Why do people like them? mbufs are good enough to get
Warner> decent performance on most available hardware and they are
The big reason for mbufs is that they were the best choice for Vax
systems with very little memory... Today we can afford to use
contiguous buffers and the traffic patterns and network speeds have
increased greatly. Mbufs are a dead end...
Warner> in a stack that is relatively solid. The linux crowd
Warner> knows well the pitfalls of trying to reinvent the TCP
Warner> wheel :-).
As knows well the pitfalls in using BSD TCP... BSD 4.4 is a solid TCP
implementation but as known pitfalls and it will need replacement some
time in the future. The Linux crowd is well positioned in that race...
Don't take this as the Not Ivented Here Syndrome...
We do take a lot of concepts invented on BSD but we do try to build
our network stack because there are things we think we should do
different. This is a big win to both free software and networking.
regards,
Pedro.