Driver wireless rt2500usb Strange behaviour in special when accessing via ssh: 1. Typing a few characters of a command only displays (echoes) part of then, generally only the first, a wait of more than 5 secs and then another part echoes eventually executing the command. The output of the command seems to come fast and at once but not always. 2. The connection breaks from time to time (I only saw this on 2.6.32 so far). Reproducible: Always Steps to Reproduce: 1. Trying to execute a command via ssh from another computer I didn't observe any problem in kernels before 2.6.32. I just left k 2.6.31-r10 because of bug #307671.
Created attachment 225235 [details] emerge --info
Please attach your kernel .config
Created attachment 225347 [details] .config
... > 2. The connection breaks from time to time (I only saw this on 2.6.32 so far). I could already see several breaks also on 2.6.33. The "intensity" of problem depends a lot on the load (emerging, for ex.) of the computer but also occurs when nothing is running.
Additional info: When typing let's say a 5 chars command and only the 1st char is displayed, the led of the router, that indicates wireless traffic, blinks a lot as when transferring data, in general for more than 5 seconds. Then either the rest of chars are displayed and the command executed or the connection breaks. This also disturbs the rest of PCs using the router.
(In reply to comment #5) > Additional info: > > When typing let's say a 5 chars command and only the 1st char is displayed, the > led of the router, that indicates wireless traffic, blinks a lot as when > transferring data, in general for more than 5 seconds. Then either the rest of > chars are displayed and the command executed or the connection breaks. > This also disturbs the rest of PCs using the router. > This does not always happen. I could verify a high rate of broken connections, even when not using ssh. In general rt2x00 drivers (at least 2500usb) seem to be broken on kernel 2.6.32 and 2.6.33.
anything in dmesg during problem times?
(In reply to comment #7) > anything in dmesg during problem times? > No. Anyway the problem seems to be being fixed upstream. I saw this in the rt2x00 mailing list "This seems to be due to powersaving issues in the driver, which got enabled by default after 2.6.31. We are working on this, and patches for this should be going upstream soon. In the meantime you can use the workaround to disable powersaving by executing: iwconfig wlan0 power off at any time. This should make your connection more stable. "
Does the workaround work for you?
(In reply to comment #9) > Does the workaround work for you? > Yes. I inserted it in my wireless script and it works now.
Ok, great, I'll close as upstream for now and we can reopen if there's ever a backport fix from the developers.
I would like to add that this workaround does not work for everybody. This problem concerns both rt2500usb and rt2500pci (see rt2x00 mailing list). I myself am using rt2500pci and switching power management off does not improve the situation. I've got the impression that with power management on it works better. I even tried the patch being discussed in the rt2x00 mailing list. Unfortunately it didn't work either. Does it work for you to set the card manually in awake (state 3) with "iwconfig wlan0 power saving 3"? Everytime I try this the following error occurs: Error for wireless request "Set Power Management" (8B2C) : invalid argument "3". Is there another possibility to set the state of the card? Tried to edit the file /sys/devices/pci0000:00/0000:00:0e.0/0000:02:06.0/net/wlan0/dormant . But it didn't work.
