Third patchset in the set of patches to bring BBRv1 into the FreeBSD tree

[rrs]

Summary

This is the third patchset to work on getting BBR into the tree (and the latest rack we use as well). This brings
in the tcp_ratelimit.c file so that TCP transports (and possibly others) have a mechanism that can adequately
use the ratelimit code from the transport. The original ratelimit code was designed so the user at the
socket option level could "use the ratelimit" code. But this is really not practical and should be the job
of TCP (i.e. the transport).

Test Plan

This set has been tested at NF with both BBR and Rack. There are limitations
of cards that support it but we have only tested two of the cards that are
known to have ratelimit support. Other updates will be needed to expand
that to include future cards (some already out there).

[rrs]

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).

[rrs]

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

[• hselasky]

Changes look good to me.

conf/files

4258

Can you double check this syntax? I thought "config" was very simple and did not understand multiple items after "|" .

[rrs]

conf/files

4258

I think it used to be that simple, but the fact that we have tcp_pcap (line below) setup the same way makes
me think it was fixed.

I can try to configure without inet and inet6 and see what happens :)

I wonder though if it should not imitate the tcp_pcap line below that has inet and inet6 first.

[rrs]

conf/files

4258

Ok a quick check shows that it blows up nicely on compile when I
take away INET and INET6.

But this will be true for any of the ones with multiple entries. I suspect
that this is the hint, the compile blows up, you go look at the RATELIMIT line
and see you dropped both INET and INET6 and go aahh...

Until someone fixes config to be a bit smarter of course.

Note there are 100's of lines that have multiple inet foo | inet6 foo
so it is not alone ... and I guess just gives us the "human" hint in the file