Comment 109 for bug 348204

I repeat that this bug showed up in an old version (8.04 or thereabouts), was gone on 8.10, then showed up on 9.04 and has been solidly buggy ever since. Even a Sony TZ that I thought was immune to it turned out to be prone as well. I originally identified it on a Dell D830 with a 4965 chip.

The best test I have, one that rarely fails, is to transfer an 8 GB file from the suspected machine to locally wired server (rsync, nfs, anything will do). At best, the transfer will stall at around 5GB, when you then realize the wireless connection no longer works. To recover from that situation there are several approaches (from reselecting the given connection on NetworkManager to a full restart), that you will have to resort according to the phase of the moon or the spin of the electrons around.

Reviewing my previous posts I noticed that I already tried with two other distros and the bug is ubuntu-related only. Since I only use the D830 as a wired server now (switched to a Mac, where the wireless always works and the machine suspends and always wakes up when I open the lid:-)), it's a bit hard for me to do a perfect test to really blame the ubuntu kernel patch set. The test is to compile a vanilla kernel and see what happens with wireless, even if other things break, since this is supposed to be just a test...

Maybe I find sometime to test this approach on the Sony TZ, but who knows, this semester I'm swamped.