tcp: make IsLost() related accounting variables signed
ClosedPublic
Actions

Authored by rscheff on Oct 12 2023, 8:42 PM.

Details

Reviewers

cc
gallatin
tuexen
rrs
glebius

Group Reviewers

transport

Commits

rG22dc8609c565: tcp: use signed IsLost() related accounting variables

Summary

Coverity found that one safety check (kassert) was not
functional, as possible incorrect subtractions during
the accounting wouldn't show up as (invalid) negative
values.

Also reshuffle the zeroing of these variables to align
with the struct (ascending addresses).

Reported by: gallatin

Diff Detail

Repository

rG FreeBSD src repository

Lint

Lint Not Applicable

Unit

Tests Not Applicable

Event Timeline

rscheff created this revision.Oct 12 2023, 8:42 PM

Herald added a reviewer: transport. · View Herald TranscriptOct 12 2023, 8:42 PM

Herald added subscribers: melifaro, imp. · View Herald Transcript

rscheff requested review of this revision.Oct 12 2023, 8:42 PM

Harbormaster completed remote builds in B53953: Diff 128689.Oct 12 2023, 8:42 PM

cc accepted this revision.Oct 12 2023, 9:01 PM

cc added inline comments.

sys/netinet/tcp_var.h
131–132	Looks need to adjust the tabs here if that's the case, then good to go. for example: "int32_t hole_bytes;"

This revision is now accepted and ready to land.Oct 12 2023, 9:01 PM

rscheff added inline comments.Oct 17 2023, 11:58 AM

sys/netinet/tcp_var.h
131–132	Not quite sure what you mean; the line is <tab>int32_t<tab><tab>hole_byte;<tab> /* comment Identical to the sack_bytes_rexmit above. Maybe the tab width is set different somehow? But I could change all the "int32_t<tab><tab>" to "int32_t<space><tab>" in this struct to prevent single-char tabs...

whitespace change (int32_t<tab><tab> to int32_t<space><tab>)

This revision now requires review to proceed.Oct 17 2023, 12:00 PM

Harbormaster completed remote builds in B54031: Diff 128904.Oct 17 2023, 12:00 PM

cc added inline comments.Oct 17 2023, 12:55 PM

sys/netinet/tcp_var.h
131–132	Not quite sure what you mean; the line is <tab>int32_t<tab><tab>hole_byte;<tab> /* comment Identical to the sack_bytes_rexmit above. Maybe the tab width is set different somehow? But I could change all the "int32_t<tab><tab>" to "int32_t<space><tab>" in this struct to prevent single-char tabs... Maybe that's a display problem. I checked the source code file. For uint32_t, there is a single tab after it. For int32_t, there are double tabs after it. So double tabs should be the format after int32_t.