Page MenuHomeFreeBSD

netinet6: allow disabling excess log messages
ClosedPublic

Authored by kaktus on Feb 16 2023, 6:39 PM.
Tags
None
Referenced Files
F81611896: D38644.id117767.diff
Thu, Apr 18, 11:05 PM
Unknown Object (File)
Mon, Apr 8, 5:46 PM
Unknown Object (File)
Mon, Apr 8, 2:16 PM
Unknown Object (File)
Sun, Apr 7, 11:27 PM
Unknown Object (File)
Sun, Apr 7, 5:24 PM
Unknown Object (File)
Sun, Apr 7, 3:54 PM
Unknown Object (File)
Sun, Apr 7, 3:26 PM
Unknown Object (File)
Jan 16 2024, 4:25 PM

Details

Summary

RFC 4443 specifies cases where certain packets, like those originating from local-scope addresses destined outside of the scope shouldn't be forwarded. The current practice is to drop them, send ICMPv6 message where appropriate, and log the message:

cannot forward src fe80:10::426:82ff:fe36:1d8, dst 2001:db8:db8::10, nxt 58, rcvif vlan5, outif vlan2

At times the volume of such messages cat get very high. Let's allow local admins to disable such messages on per vnet basis, keeping the current default (log).

Reported by: zarychtam@plan-b.pwste.edu.pl

Diff Detail

Repository
rG FreeBSD src repository
Lint
Lint Not Applicable
Unit
Tests Not Applicable

Event Timeline

zlei added a subscriber: zlei.

Looks good to me

sys/netinet6/in6_proto.c
409

I think for the reasons outside of firewall rules can be omitted.

People utilize firewalls should be aware that firewalls have their own logging facilities for the dropped packets (those can't be forwarded).

sys/netinet6/ip6_forward.c
117โ€“118

As V_ip6_log_discarded is mostly constant then I'd prefer to check it first.

This revision is now accepted and ready to land.Feb 17 2023, 2:17 AM

Thanks for taking care of it, looks to me like another positive change.

sys/netinet6/in6_proto.c
408

Perhaps a sentence like "Log packets originating from link-local source addresses that weren't forwarded" will be more accurate (if a precise description is really required), but I am not the best at phrasing it.

kaktus edited the summary of this revision. (Show Details)
kaktus added a reviewer: manpages.
This revision now requires review to proceed.Feb 22 2023, 3:54 PM
sys/netinet6/in6_proto.c
409

I believe they do but the extra 41 bytes of description could save some time for those who may be in a hurry to find a solution for a local firewall issue :-)

Rename to match log statement, stats variable and counter description presented by netstat -s.

Discussed with: zarychtam@plan-b.pwste.edu.pl

cy requested changes to this revision.Feb 22 2023, 6:01 PM
cy added a subscriber: cy.
cy added inline comments.
sys/netinet6/in6_proto.c
409

There are other examples (MIBs) which enable dropping packets. They do not mention firewalls. I agree with zlei. The extra verbiage should be removed.

This revision now requires changes to proceed.Feb 22 2023, 6:01 PM
pauamma_gundo.com added inline comments.
share/man/man4/inet6.4
32

Remember to bump.

348
459

Add RFC 4443?

460โ€“468

Out of scope for this change, but noting while here that a 23 year old draft isn't the best reference.

Address comments:

  • by popular demand don't mention firewall
  • update man page wording
share/man/man4/inet6.4
459

Other protocols use the SEE ALSO section for that, for example RFC4443 is mentioned like that in icmp6(4) and I don't know which way is preferred by the docs team. This manual page as a whole could use some love as there are many other RFCs to be mentioned in the STANDARDS / SEE ALSO sections.

460โ€“468

It seems that this extension hasn't been ratified to official standard[1] but our implementation supports it so maybe as a future rework of this whole manpage at least the "work in progress material" should be removed.

  1. https://datatracker.ietf.org/doc/draft-ietf-ipngwg-scopedaddr-format/
share/man/man4/inet6.4
459

"SEE ALSO" works for me. I have no marked preference, but it would be puzzling, when the ICMP6 RFC is there in icmp6(4), to have it as a standard in this page.

as requested mention RFC4443 in man page

tidy up

the standard approach is to rate limit with ppsratecheck

i would say this should be 2 patches: 1. do the rate check 2. add the tunable

sys/netinet6/in6_proto.c
349

Just a nit but cannot is a little more formal than can't. Seems to flow a little better too.

In D38644#881703, @mjg wrote:

Please refrain from making further requests if you're unwilling to review them later.

This revision was not accepted when it landed; it landed in state Needs Review.Mar 13 2023, 4:50 PM
This revision was automatically updated to reflect the committed changes.