Frequent disconnects under certain network conditions
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
wpasupplicant (Ubuntu) |
Invalid
|
Medium
|
Unassigned | ||
Precise |
Fix Released
|
Medium
|
Unassigned |
Bug Description
[Impact]
Under WPA2-Enterprise secured networks, roaming between the secure APs need to cache the keys used, since although the AP is different, the same authentication backend (a server of some sort, possibly RADIUS) is most likely the same. This only impacts specific setups using WPA Enterprise security on wireless networks in environments where lots of APs are installed and roaming happens frequently.
[Test Case]
Run the new WPA supplicant.
Notice whether there are fewer wireless disconnects events.
[Regression Potential]
Low, this only affects WPA Enterprise security, and only in specific setups where roaming between one AP to the next is both frequent and expected. The included patches are backports from WPA 1.0, which have been in Ubuntu since Quantal, with no adverse effects.
However, there are quite a few patches at cause, which introduce complexity in the testing and the actual functional changes.
See http://
---
There are several upstream patches that help reduce network disconnects and excessive roaming on some networks.
Mainly they are related to Opportunistic Key Caching, authorization and association timing fixes.
Related branches
tags: | added: precise |
Changed in wpasupplicant (Ubuntu): | |
status: | New → Incomplete |
Changed in wpasupplicant (Ubuntu): | |
status: | Incomplete → Invalid |
importance: | Undecided → Medium |
Changed in wpasupplicant (Ubuntu Precise): | |
status: | New → Triaged |
importance: | Undecided → Medium |
description: | updated |
description: | updated |
I think it is inaccurate to say "this only affects enterprise setups". The bug may only affect enterprise setups, but the code may be shared with non-enterprise setups. The point of "Regression Potential" is to shine light on what parts of the package are affected by the change to aid testers in looking for regressions.
This is a big patch, so I'd like to see this get wide exposure on multiple setups while still in -proposed.