Comment 14 for bug 568090

Oh yes, my bad, it wasn't a duplicate. However I think that most of the respondents probably suffered from this bug and not the one you described. If you think your bug still needs to be fixed on 2.6.32 feel free to unmark it as a duplicate.

It's very interesting that you don't have the problem with an AR2413, though. I don't have other tests aside from trying other packet sizes, but if you use this computer daily you would quickly notice sluggish performance if you were affected by the problem.

Perhaps the "Subsystem:" line of lspci -vnn could be an indicator:

(scottl, unaffected by the bug)
06:01.0 Ethernet controller [0200]: Atheros Communications Inc. AR2413 802.11bg NIC [168c:001a] (rev 01)
 Subsystem: Atheros Communications Inc. Device [168c:2052]

(Jan-Olof Lindqvist and exogenetic, likely affected by the bug)
09:04.0 Ethernet controller [0200]: Atheros Communications Inc. AR2413 802.11bg NIC [168c:001a] (rev 01)
 Subsystem: Askey Computer Corp. Device [144f:7094]

That's a small sample size, it would be nice to see if other AR2413 users have the same pattern.