View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0002652 | GNUnet | transport service | public | 2012-11-13 22:25 | 2012-12-21 15:49 |
| Reporter | cy | Assigned To | Christian Grothoff | Due Date | |
| Priority | normal | Severity | crash | Reproducibility | always |
| Status | closed | Resolution | fixed | ||
| Product Version | Git master | ||||
| Target Version | 0.9.5 | Fixed in Version | 0.9.5 | ||
| Summary | 0002652: timed out task crashes transport | ||||
| Description | Because dead tasks are treated as fatal errors, a task that timed out is crashing gnunet-service-transport. Likely the task ID is being initialized invalidly, so the attempt to remove it considers it a dead task. Though it might be a good idea not to fatally exit when a dead task is found, instead just assuming it must either have already been executed or just be a bad task ID. The only evidence that a task is dead is the presence of a task ID integer and lack of a corresponding task, which does not need cleanup in any fashion. As near as I can tell this is what happens: 1) session->nat_connection_timeout is set to a valid task 2) A timeout occurs 3) that task is run and removed from the queue 4) session->nat_connection_timeout is not set to GNUNET_SCHEDULER_NO_TASK 5) disconnect_session is called 6) it tries to cancel by session->nat_connection_timeout 7) there's no task by that ID on the queue! 8) blanket assert(0) causes gnunet-service-transport to crash. simply skipping step 8 would cause no problems, as the task was run and removed properly after the appropriate delay. IMO cancelling an already run task should at the very worst result in a warning, not an abort. | ||||
| Steps To Reproduce | Run gnunet with my peculiar nat issues. Cause a NAT connection timeout Observe coredump. Observe gnunet flopping around uselessly like a fish out of water. | ||||
| Additional Information | I have the ports gnunet is set to use already forwarded manually over my router, so there should be no NAT problems. | ||||
| Tags | No tags attached. | ||||
| has duplicate | 0002654 | closed | Matthias Wachs | TCP uses invalid session when behind NAT |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2012-11-13 22:25 | cy | New Issue | |
| 2012-11-13 22:25 | cy | Status | new => assigned |
| 2012-11-13 22:25 | cy | Assigned To | => Matthias Wachs |
| 2012-11-14 08:17 | Christian Grothoff | Note Added: 0006587 | |
| 2012-11-14 08:17 | Christian Grothoff | Assigned To | Matthias Wachs => Christian Grothoff |
| 2012-11-14 08:17 | Christian Grothoff | Target Version | => 0.9.5 |
| 2012-11-14 08:18 | Christian Grothoff | Status | assigned => resolved |
| 2012-11-14 08:18 | Christian Grothoff | Fixed in Version | => 0.9.5 |
| 2012-11-14 08:18 | Christian Grothoff | Resolution | open => fixed |
| 2012-11-19 08:32 | Matthias Wachs | Relationship added | has duplicate 0002654 |
| 2012-12-21 15:49 | Christian Grothoff | Status | resolved => closed |