View Issue Details

IDProjectCategoryView StatusLast Update
0002975GNUnetutil librarypublic2013-12-24 20:54
ReporterMatthias Wachs Assigned ToMatthias Wachs  
PriorityimmediateSeveritycrashReproducibilityalways
Status closedResolutionfixed 
Product VersionGit master 
Target Version0.10.0Fixed in Version0.10.0 
Summary0002975: Issues due to latest crypto changes
DescriptionAfter update of gnunet9 and fulcrum peers do not connect!

- I regenerated private keys
- removed .nseproof

Issues with core:

gnunet9@sam:~$ Aug 07 11:52:03-210278 util-606201 WARNING ECC signing failed at crypto_ecc.c:775: Invalid flag
Aug 07 11:52:03-275549 core-606201 ERROR Assertion failed at gnunet-service-core_kx.c:1500.

Additional InformationThe bugs mentioned in this report are solved... will open separate bugs for other issues
TagsNo tags attached.

Activities

Matthias Wachs

2013-08-07 12:17

reporter   ~0007300

Last edited: 2013-08-07 12:20

- Crypto does not work on transport level and imho core level:

Incoming connections on gnunet9:

gnunet9@sam:~/svn/gnunet$ Aug 07 12:11:33-544252 util-617904 WARNING ECC signature verification failed at crypto_ecc.c:855: Invalid flag
Aug 07 12:11:33-544374 transport-617904 WARNING Failed to verify: invalid signature on address tcp:tcp.0.[2001:4ca0:2001:11:226:b9ff:fe7d:84ed]:2086 from peer `SLOS'
Aug 07 12:11:33-544948 util-617904 WARNING ECC signature verification failed at crypto_ecc.c:855: Invalid flag
Aug 07 12:11:33-544988 transport-617904 WARNING Failed to verify: invalid signature on address tcp:tcp.0.131.159.20.52:2086 from peer `SLOS'

- When a peer tries to connect:
Aug 07 12:11:00-526887 transport-617635 ERROR Failed to create PONG signature for peer `SLOS'
Aug 07 11:53:14-060156 transport-606200 ERROR Assertion failed at gnunet-service-transport_validation.c:1069.

Core issues

Aug 07 11:53:14-060156 transport-606200 ERROR Assertion failed at gnunet-service-transport_validation.c:1069.
Aug 07 11:53:47-746328 util-606440 WARNING ECC signing failed at crypto_ecc.c:775: Invalid flag
Aug 07 11:53:47-807717 core-606440 ERROR Assertion failed at gnunet-service-core_kx.c:1500.

Matthias Wachs

2013-08-07 12:17

reporter   ~0007301

Core issues:

Aug 07 12:10:18-667574 core-617731 ERROR Assertion failed at gnunet-service-core_kx.c:1500.
Aug 07 12:10:22-094916 util-617741 WARNING ECC signing failed at crypto_ecc.c:775: Invalid flag

Christian Grothoff

2013-08-07 14:13

manager   ~0007302

Ok, now that's very interesting. A major real-world issue not covered by our tests! We need a tests that reflects this, now!

Christian Grothoff

2013-08-07 14:14

manager   ~0007303

Oh, the issues are likely an outdated libgcrypt on those peers. We need a recent libgcrypt again...

Matthias Wachs

2013-08-08 10:03

reporter   ~0007307

OK ... we assume the issues above are solved. Here the latest issues as seen on gnunet9:

- Peer tries to connect to himself on transport level
Aug 08 10:01:13-647046 transport-831466 WARNING External protocol violation detected at plugin_transport_tcp.c:2032.
Aug 08 10:01:13-647089 transport-tcp-831466 WARNING Received WELCOME message from my own identity `HIJN' on address `131.159.74.67:43103'

DNS: 67.74.159.131.in-addr.arpa name = gnunet.informatik.tu-muenchen.de.

Christian Grothoff

2013-08-08 10:13

manager   ~0007308

Ok, so somehow now transport fails to prevent connecting to itself.

Could be an old HELLO (hostlist, etc.), so it thinks that there's another peer at its own address. So I'd say that's harmless. Maybe easy to fix if we simply yet again change the message ID for HELLO (so outdated HELLOs are invalidated).

Matthias Wachs

2013-08-08 10:25

reporter   ~0007309

trying to track it down... it seems not to be a remote peer nor validation
wip

Matthias Wachs

2013-08-08 10:52

reporter   ~0007310

Was caused by invalid shipped hellos ...
Issue: this hellos spread like the pest!
Cleaned up gnunet9 and some peers gossiped the hello and it started again :-(

Matthias Wachs

2013-08-08 10:53

reporter   ~0007311

Increasing the message id is a workaround but our system has to be resilient against this realworld issues :-D
So yes... it's harmless but interesting/helpful to see such things happen!

Issue History

Date Modified Username Field Change
2013-08-07 11:54 Matthias Wachs New Issue
2013-08-07 12:17 Matthias Wachs Note Added: 0007300
2013-08-07 12:17 Matthias Wachs Note Added: 0007301
2013-08-07 12:18 Matthias Wachs Assigned To => Christian Grothoff
2013-08-07 12:18 Matthias Wachs Priority normal => immediate
2013-08-07 12:18 Matthias Wachs Severity minor => crash
2013-08-07 12:18 Matthias Wachs Status new => assigned
2013-08-07 12:19 Matthias Wachs Description Updated
2013-08-07 12:20 Matthias Wachs Description Updated
2013-08-07 12:20 Matthias Wachs Note Edited: 0007300
2013-08-07 14:13 Christian Grothoff Note Added: 0007302
2013-08-07 14:14 Christian Grothoff Note Added: 0007303
2013-08-07 14:15 Christian Grothoff Assigned To Christian Grothoff => Matthias Wachs
2013-08-07 14:15 Christian Grothoff Status assigned => feedback
2013-08-08 10:03 Matthias Wachs Note Added: 0007307
2013-08-08 10:03 Matthias Wachs Status feedback => assigned
2013-08-08 10:13 Christian Grothoff Note Added: 0007308
2013-08-08 10:25 Matthias Wachs Note Added: 0007309
2013-08-08 10:52 Matthias Wachs Note Added: 0007310
2013-08-08 10:53 Matthias Wachs Note Added: 0007311
2013-08-08 11:26 Matthias Wachs Additional Information Updated
2013-08-08 11:26 Matthias Wachs Status assigned => resolved
2013-08-08 11:26 Matthias Wachs Resolution open => fixed
2013-08-11 10:25 Christian Grothoff Product Version => Git master
2013-08-11 10:25 Christian Grothoff Fixed in Version => 0.10.0
2013-08-11 10:25 Christian Grothoff Target Version => 0.10.0
2013-12-24 20:54 Christian Grothoff Status resolved => closed