: Sync-Flood Fixes
Get series with b4
b4 am 200706201055.17078@strip-the-willowVersions newest first
- v1 Latest4 of 5 patches 0 repliesNew
4 of 5 patches are in this archive. The cover letter announces 5.
Patches in v1 No applies recorded yet
| # | Subject and files | Size | Replies | Applied |
|---|---|---|---|---|
| 1/5 | : Shorten variable names in dccp_check_seqnonet/ | +17 | 0 replies | |
| 2/5 | : Fix Reset/Sync-Flood Bugnet/ | +11 | 0 replies | |
| 3/5 | : Parameter renamingnet/ | +6 | 0 replies | |
| 5/5 | : Send Reset upon Sync in state REQUESTnet/ | +3 | 0 replies |
Reviews trailers per patch
No Reviewed-by, Acked-by or Tested-by trailers for v1 yet.
Cover letter [PATCH 0/5][BUG-fix]: Sync-Flood Fixes
Arnaldo asked me to send these, they have been made independent of the patches in the test tree (but will be made available for review).
Patch #1: Cosmetics, but required by other patches - shortens many occurrences of
very long variable names.
Patch #2: Reset/Sync-Flood Bug fix.
The bug has been known for a long while (I recently found an email from Ian who mentioned it as early as December), a previous discussion with Eddie helped to track down where the bug originates. Patch has detailed description of how bug is triggered, and fixes the described symptom.
Patch #3: Renames parameter for clarity (was confusing).
Patch #4: Rate-Limit DCCP Syncs. This is recommended in RFC 4340 and from own experience
I found that rate-limiting alone can terminate oscillating Sync message exchanges, i.e. it increases resilience against malfunction and attacks.
Patch #5: Second bug - in REQUEST state, a Reset is sent in response to Sync, not SyncAck.