net.inet.tcp.tolerate_missing_ts=1

[tuexen]

Summary

As of June 2021, several TCP stacks violate RFC 7323 from September 2014:

• Omniswitch 63xx and 64xx models (negotiating TS, but not sending them).
• HP LaserJet 500 color M551dn (negotiating TS, but not sending them).
• IPFW TCP keep-alive injection (not sending TS when injecting packets).
• Mac OS 11.4 (negotiating TS, but not sending TS on keep-alive packets).

Therefore tolerating violations for now...

[tuexen]

Change man-page.

[rgrimes]

share/man/man4/tcp.4

385–387

Perhaps make the same explination here as is in the comment of the code so that people wont turn this off without knowing what it is that is non compliant?

[tuexen]

share/man/man4/tcp.4

385–387

I was thinking about this. I wanted to provide some explicit informations about the current situation,
in case someone considers to change the default in the future. That is why I put the information in
the code, not just in the commit message.

However, I don't want to blame anybody. That is the reason why I did not put it in the more
prominent place in the man page. On the other hand, your arugerment is also reasonable.
I'll bring it up at the next transport call...

[jtl]

share/man/man4/tcp.4

385–387

How about a compromise?

As of June 2021, several implementations in common use are known to be non-compliant. Disabling this sysctl may impair the host's ability to communicate with these non-compliant implementations.

[rrs]

I wound not advise putting names in the man page.

[tuexen]

share/man/man4/tcp.4

385–387

Used a similar text.