tcp.4: Sort sysctl variables
ClosedPublic
Actions

Authored by • karels on Jul 18 2022, 4:53 PM.

Details

Reviewers

Group Reviewers

manpages
transport

Commits

rGa795c6e93444: tcp.4: Sort sysctl variables

Summary

In preparation for updates including missing variables, sort the
sysctl variables in the MIB variables section alphabetically.
Add a new "hostcache" entry for the hostcache node, containing the
intro text that was previously in hostcache.enable.

Test Plan

tested with mandoc

Diff Detail

Repository

rG FreeBSD src repository

Lint

Lint Not Applicable

Unit

Tests Not Applicable

Event Timeline

• karels created this revision.Jul 18 2022, 4:53 PM

Herald added a subscriber: imp. · View Herald TranscriptJul 18 2022, 4:53 PM

• karels requested review of this revision.Jul 18 2022, 4:53 PM

Harbormaster completed remote builds in B46465: Diff 108287.Jul 18 2022, 4:53 PM

I realize this is a large review. Most of the diffs result from moving blocks around. I have attempted to add inline comments for all the changes I made other than moving blocks.

share/man/man4/tcp.4
587	Parenthetical fast_finwait2_recycle was added since the two are no longer together.
598	This item is new, using mostly text from hostcache.enable (which was first). "See hostcache.enable" was added.
616	Markup added to "hashsize" and "bucketlimit".
619	The next line was added; the list looked odd without any leading text.
645	Spelling of "purge" was corrected from "pruge".
690	Added the word "keepalive" to previous line.
932	"See pmtud_blackhole_detection" was added.

Big change, but good to have. I've added a few bits of my own, mostly minor fixes.

share/man/man4/tcp.4
594	s/tcp/TCP/ and below (same sentence).
649	Maybe move the "(default)" just before the closing sentence stop.
653	I think we don't need the comma here.
667	s/on fly/on the fly/
695	maybe "... before sending keepalive probes (if enabled)." (Active voice)
701	Maybe provide a human-readable number here as well, just like above for the 7200000 msec?
714	s/That/A value of/
725	Format this occurance of TIME_WAIT here just like above?
747	Another instance of markup for TIME_WAIT here (see above comment)?
747	s/creating/creation/
818	Strange line breaks here, maybe it's the rendering of the diff, but could also be unintentional edits.
898	s/may be/is/

Updates per review comments.

Harbormaster completed remote builds in B46479: Diff 108332.Jul 19 2022, 5:24 PM

Comments should all be addressed.

share/man/man4/tcp.4
701	Also did the same for keepidle, just below.
717	I changed "unlisted" to "not listed."
818	The line breaks were this way in the original; maybe an artifact of a previous change.

That looks even better with your additions.
The man page part looks fine now. Let's wait for feedback from the transport group in case we need to adjust the wording.

tuexen added a subscriber: tuexen.Jul 19 2022, 6:51 PM

tuexen added inline comments.

share/man/man4/tcp.4
696	I find 7.2M msec confusing (what is the multiplier: a power of 2 or a power of 10, for example). So I would prefer `7200000 msec (2 hours)`.
701	Same applies here. I would prefer: `75000 msec (75 sec).`
707	Same applies here. I would prefer: `75000 msec (75 sec).`

• karels marked an inline comment as done.Jul 19 2022, 7:17 PM

• karels added inline comments.

share/man/man4/tcp.4
696	That was the original text, before bcr's request :). It's hard to imagine that anyone would think that time would be expressed as a power of 2, but it's not exactly SI (where M would be powers of 10). Any other opinions? I could do 7.2 million sec, or 7,200,000 msec (yes, that's US...).

tuexen accepted this revision.Jul 19 2022, 7:31 PM

tuexen added inline comments.

share/man/man4/tcp.4
696	Ahh, I see. The only important information for me is the it is 2 hours... 7,200,000 would help me to easily convert it to 7200 seconds... I leave it up to you or bcr, who knows better how to write human readable text than me, I think.