You cannot use tcp_compute_pipe() with rack. It does not use
the same variables as the default stack. Instead you must use the
ctf_flight_size() function to get whats in flight.

Well I am going to voice a different view. I do *not* see this as a nice cleanup. I see it
has a lessening of information.

This updates the patch set, the ktls.h information was incorrect. I also add the fixes
that Michael is doing on the Rack side to BBR as well (check for invalid length in non-tso case).

I would like to see the Rack version of the code be identical to the tcp_input version.. i.e. it should check

First version was missing a (void) in the registration function.. opps

This adds a simple registration mechanism that stacks can use as they successfully register/deregister to
make the LRO code aware if any stack wants to use mbuf-queueing per Drew's suggestion.

One more update, mainly comments but we also fix a couple of things

1. The timestamp provided via the ctf common functions should be the real time not something from the mbuf, we let the transports look at those flags (which both BBR and the latest Rack do).
2. We also for LRO add a gating of the number of acks (currently set to infinity) that can also cause a wakeup.

The LRO patch was missing data and length limits for input processing via
MBUF_QUEUE methods. This update adds that and an elaborate set of
comments to the rack_bbr_common.c module describing in detail what
a transport designer using MBUF_QUEUEING needs to contemplate.

Please validate the rack_bbr_common.c has (or does not have) the extra bit you added in the early part.

I plan on commiting this August 1st unless I hear screams.. its been pending since July 15th....

Update to address Michael's comments about copyright (missing new magic declarations at the top) and
also address an issue where if we have an interface that marks it to "not supporting" we properly
disable the rate limiting for that interface and do not allow the user to enable it (since it does
not support it).

Add in the hopeless exit to tcp_hpts and also get the corrected (with DSACK patch) rack_18q22p2

Update to include some fixes Michael and I have been working on in a slow slow machine as well
as bring the Rack stack up to the latest production (with some small changes to make it compile).

And we can't loose the commented out opt_kern_tls until the kern tls hits the tree

It helps if one remembers to bring the changes of the .h file too ... opps

Address phabricator nits

In my detailed testing of hpts I found we are *not* doing the right thing as to calculating
the time for sleep. The code was starting at where the prev_slot was and moving
through counting the slots until it found a non-empty slot.

So interestingly those includes are not needed.. not even sure the tls one is need (hwtls should be in the tree soon).. however more
interesting is why my freebsd head build works.. not sure why since it does not have hwtls in the mix.. need to go check whats going
on in my conf.

sorry wrong revision.. the trouble is in the DSACK commit which is r347382... thats what we are backing out of our sync... (off by
one error.. must be a programmer :-) )

There are for more problems then this. And its not explicitly with HPTS but it becomes the canary in the coal mine.
The basic problem is in r347381. This cause a huge slow down to the system at least with INVARIANTs. Hpts in
the form you are dealing with does handle wheel wrap well. We are backing out that change in our upstream
merge. And there will be changes I will be committing to fix wheel_wrap in hpts.