• TransportProtocols
• 16October2025

Attendees

• Nick Banks
• Peter Lei
• cc@
• gallatin@
• glebius@
• melifaro@
• rscheff@
• tuexen@

Ongoing Short Term Projects

• At least for testing purposes (packetdrill) it would be good to have LRO support for tun interfaces. tuexen@ suggested to generalize LRO to not depend on an Ethernet header being present. gallatin@ prefers to not make the LRO code more complex, since it most likely will impact the performance. tuexen@ has implemented LRO support for the tap interfaces. packetdrill support will be committed soon.
• BBLog is missing a man page. tuexen@ will write one.
• tuexen@ will add tests for handling RST ACK segments in all states and test the base, RACK and BBR stack.

• D46056 fixes a bug when sending UDP packets. tuexen@ committed it and reverted it, since it broke two tests related to IPv6 socket options. tuexen@ will look into it. This issue is found again by syzkaller.

• D46425 needs some review. rscheff@ and tuexen@ will review it.

• glebius@ is looking at the potential deadlock due to locking two inpcbs in different order. Still on the To-Do list.

• The maximum sizes for socket buffers, the TCP send and recv buffers were discussed. It is agreed that the current values are too small. rgrimes@ will survey what is used by other OSes and a decision will be based on that. @thj agrees with adjusting the maximum buffer sizes. @rscheff will ping rgrimes@ to try to make this change into 15.0, with either glebius@ or rscheff@ making the necessary adjustments before the deadline. tuexen@ proposed D52871, D52872, and D52873. All people on the call agree that the value should be bumped to from 2 to 8, which is considered conservative. This should be fine for FreeBSD 15. Bumping it again to a larger value will be reconsidered in combination with some methods to mitigate potential attacks related to the exhaustion of mbufs.

• glebius@ wants feedback on D51792. The problem space was discussed. glebius@ will try to invite gallatin@ to the next meeting for discussion of the patch. This patch can be split up, but really should go in together with another patch which only exists as prototype by gallatin@, which addresses suboptimal port selection.

• Make use of vtnet checksum offloading in bhyve: D51289, D51291 and D51688. Reviews appreciated. pouria reviewed D51289, but some advice needed on specific circumstances. glebius@ suggested to have ae@ and melifaro@ brought in to give advice how to handle the specific case.

• TCP related panic in FreeBSD 14.3: PR 288904. tuexen@ suggested to enable BBLog to collect more information.

• It was discussed that deprecating the support for the old TCP SACK RFC would be good. rscheff@ will add a deprecation notice to go into FreeBSD 15 and the code will be removed in FreeBSD 16. Change was put in for 15.0; rscheff@ will clean up the conditionals on newsack and clean up the codes in the coming weeks.
• cc@ reported a very large congestion window when using the RACK stack with CUBIC congestion control. He will look into that.

• glebius@ wants review D52771. tuexen@ and thj@ will review it.

• gallatin@ suggests to test the new simple_tx feature of the iflib driver.

• Generic improvements of NIC drivers were discussed.
• Nick Banks triggered a discussion on how to deal with constants like number of milli seconds in a second and so on.

• tuexen@ wants feedback on D52948.

• It was discussed how to deal with dependencies of kernel modules used in the ktest infrastructure. glebius@ will provide a proof of concept for the solution discussed: don't autoload modules being tested by the test modules and allow test modules to fail loading in case some conditions are not met.

Ongoing Longer Term Projects

• Now the RACK and the BBR stack are compiled by default in head. The BBR and the RACK stack will continue to coexists. We discussed if the RACK stack could be made the default and what should drive this decision. tuexen@ will reach out to the mailing list to ask people to test the RACK stack and provide feedback. In January, we will look again at the decision process based on the feedback we got. Feedback received:
  • You need to compile in the TCPHPTS. It is a loadable module now, thanks to glebius@.
  • RACK is on the loopback interface slower than the base stack (CPU limited). olivier@ has a script to test combinations of stacks and CC modules.

  • A problem with RACK and pf, which couldn't be reproduced by olivier@. tuexen@ has an idea what is going wrong and will provide a patch to the reporter for testing.D43769 fixes this, but adds some instructions to the code path.

  • When using mbuf queueing, some dtrace static probes are missed. Fixed.
  • There is a performance regression by just loading the HPTS module. This is fixed.
• It was discussed that it would be a good idea to move the freebsd stack from ticks to micro seconds.

• rscheff@ is improving AccECN code: D36303. rrs@ will review it and implement it for the rack stack. Also the interaction between LRO and AccECN++ has to be taken into account. D42563 further adds the support of ACKofACK.

• D23230 is implementing ECN++. rrs@ will review it.

• tuexen@ wants to write a tool which dumps the BBLog information of a TCP endpoint from a life system or a core. Using kvm_read() has drawbacks as glebius@ pointed out, using a python kgdb script may be an alternative. An alternative is to write a python kgdb script which pumps data into tcplog_dumper. This would allow to minimize code duplication. glebius@ will try to work on this soon. thj@ has an initial implementation at tcplog_dumper.

• The handling on ECN with TSO by NICs was discussed. tuexen@ will add a configuration option for various Intel NICs to make the TSO behavior configurable, rscheff@ will make in the TCP stack configurable wether classical ECN or AccECN will use TSO. tuexen@ wants feedback on D44258 and D44259.

• rgrimes@ reports that FreeBSD using the default parameters can not saturate a Gigabit pipe with a delay used by recent access technologies. He is proposing to bump up some default values.

• glebius@ will re-activate D35199 which speeds up finding bugs.

• tuexen@ will add fast open tests to the FreeBSD testsuite. This might require some additional counters.
• gallatin@ will look into improving performance by moving things around in structures. This will reduce cache misses...

• cc@ wants review of D47130. Possible ways of breaking up tcp_do_segment() were discussed. No agreement was reached. cc@ will try several ways and bring the alternatives back for further discussion.

• pouria was asking how to go about getting Geneve implemented in fbsd as an alternate L2 overlay encapsulation protocol; thj@ recommended to go the fastest route to bring something up, and check if vxlan api can be reuses or deviations are necessary. pouria mentioned that much of the plumbing from vxlan could be reused or copied. rscheff@ supports this kind of work to have alternatives, but raised concern that the heavy lifting is in the control plane when SDN is to be deployed.

Next Meeting

30 October 2025 @ 1500 UTC using FreeBSDTCPCall