]> git.baikalelectronics.ru Git - kernel.git/commit
Merge branch 'tcp-address-issues-with-ect0-not-being-set-in-dctcp-packets'
authorJakub Kicinski <kuba@kernel.org>
Sat, 21 Nov 2020 02:10:47 +0000 (18:10 -0800)
committerJakub Kicinski <kuba@kernel.org>
Sat, 21 Nov 2020 02:10:47 +0000 (18:10 -0800)
commitf560fca3853c46bfebbb4a18f3474e1ef5434cfb
tree6af6fa8f6e4c2259411d963bd709f35014284342
parent8dee3848e55b1b7a807580fc656a8c75be9a1903
parentd7e12df724a706b6a361da7809fbf1d1014ef774
Merge branch 'tcp-address-issues-with-ect0-not-being-set-in-dctcp-packets'

Alexander Duyck says:

====================
tcp: Address issues with ECT0 not being set in DCTCP packets

This patch set is meant to address issues seen with SYN/ACK packets not
containing the ECT0 bit when DCTCP is configured as the congestion control
algorithm for a TCP socket.

A simple test using "tcpdump" and "test_progs -t bpf_tcp_ca" makes the
issue obvious. Looking at the packets will result in the SYN/ACK packet
with an ECT0 bit that does not match the other packets for the flow when
the congestion control agorithm is switch from the default. So for example
going from non-DCTCP to a DCTCP congestion control algorithm we will see
the SYN/ACK IPV6 header will not have ECT0 set while the other packets in
the flow will. Likewise if we switch from a default of DCTCP to cubic we
will see the ECT0 bit set in the SYN/ACK while the other packets in the
flow will not.
====================

Link: https://lore.kernel.org/r/160582070138.66684.11785214534154816097.stgit@localhost.localdomain
Signed-off-by: Jakub Kicinski <kuba@kernel.org>