Ok, I did some register dumping on the working version and a later non-working version. Specifically, that last is trunk before the latest merge of last night. Attached is diff -y of the dump files (from good to bad).
If anyone sees anything interesting, please share it. I'll also post my thoughts if I happen to be able to make sense of it
Could you try attached patch?
Sorry to report, but it just seemed to loop on configuring TX queues 0 and 1.
It kept repeating lines like
Jun 14 214834 jekaterina kernel [ 40.390677] phy0 -> rt2x00mac_conf_tx Info - Configured TX queue 1 - CWmin 3, CWmax 4, Aifs 2.
Jun 14 214834 jekaterina kernel [ 40.390693] phy0 -> rt2x00mac_conf_tx Info - Configured TX queue 0 - CWmin 2, CWmax 3, Aifs 2.
...until I hit crtl-alt-del and booted up the old working one.
Judging from the patch, though, this is most probably caused by something in the merge instead of the antenna force thingy...
Please try latest rt2x00.git, problem should now be fixed.