Comment 32 for bug 1393744

Same problem here. Some notes about what happen.
Sorry for not providing complete logs.
This is my notes which is not very readable, not even for me.. :)

The machine has been continuously updated since 13.10.
Updated from 14.04 to 15.04 by going through 14.10.
All done yesterday. Problem detected when arriving at 15.04.

Usually either a
1: ... rt2x00queue_write_tx_frame: Error - Arrived at non-free entry in the non-full queue 0.
2: ... rt2x00mmio_regbusy_read() Indirect register access failed: offset=0x0000308c, value=0x00010400
  this is followed by: ieee80211 phy0: wlan0: No probe response from AP .... after 500ms, disconnecting
3: ... rt2x00queue_flush_queue: Warning - Queue 0 failed to flush

After turning on nohwcrypt error 1 went away, and was replaced by 2 and 3.
With nohwcrypt turned off I got 2 and then 3, leading to a time out.

1 alone gives failing network

2 and 3, or 3 alone, later gives
... authenticate with ...
... direct probe to ... (try 1/3)
... direct probe to ... (try 2/3)
... direct probe to ... (try 3/3)
... authentication with ... timed out

and then network fails

$ lspci | fgrep 'Ralink'
03:07.0 Network controller: Ralink corp. RT2561/RT61 802.11.g PCI

Some additional info
  Chipset detected - rt: 2661, rf: 0003, rev: 000b
  Selected rate control algorithm 'minstrel_ht'
  Firmware .. 0.8

It ia an AMD triple core proc

$ uname -a
Linux hydra 3.19.0-17-generic #17-Ubuntu SMP Wed May 6 16:46:12 UTC 2015 x86-64 x86_64 x86_64 GNU/Linux

Just to verify that the module is in fact the latest one. I've compiled them at one point so just to make sure they are provided by the distro.

$ cat /sys/module/rt2x00*/version
2.3.0 (reported three times)