[solved] Unstable RT73 + Asus WL-167g

Live forum: http://rt2x00.serialmonkey.com/viewtopic.php?t=4699

onamatic

27-03-2008 17:01:23

I use
Fedora FC6 2.6.20-1.2962
Source from rt73-cvs-2008032218
Asus WL-167g usb adapter.

The whole thing installed like a charm and worked first time. It is extremely unstable though and falls to pieces in a non-deterministic fashion (ie. it doesn't seem to depend on traffic volume) with the following errors

[coder12aqw3i]Mar 27 14:28:55 localhost kernel: usb 5-6: new high speed USB device using ehci_hcd and address 4
Mar 27 14:28:56 localhost kernel: usb 5-6: configuration #1 chosen from 1 choice
Mar 27 14:28:56 localhost kernel: rt73: idVendor = 0xb05, idProduct = 0x1723
Mar 27 14:28:56 localhost kernel: rt73: Interface up for first time, activating permanent MAC
Mar 27 14:28:56 localhost kernel: rt73: Active MAC is: 00:1a:92:59:47:ba.
Mar 27 14:28:56 localhost kernel: rt73: Local MAC = 00:1a:92:59:47:ba
Mar 27 14:28:56 localhost kernel: rt73: driver version - 1.0.3.6 CVS
...snipped standard ntpd messages out...
Mar 27 15:37:06 localhost kernel: BUG: sleeping function called from invalid context at mm/slab.c:3035
Mar 27 15:37:06 localhost kernel: in_atomic():0, irqs_disabled():1
Mar 27 15:37:06 localhost kernel: [<c046ebf8>] kmem_cache_alloc+0x1b/0x56
Mar 27 15:37:06 localhost kernel: [<c0588fee>] usb_control_msg+0x40/0xde
etc.....
[/coder12aqw3i]

It's then impossible to resurrect the device without physically removing the USB stick or ( reloading/modprobing the rt73 module).

I know this isn't an rt73 error as such but if anyone could point me in the right direction I'd be very grateful.

hariseldon99

04-05-2008 10:38:19

er, bump. I pretty much have the exact same issue. Asus wl-167g on Mandriva 2008.0 kernel2.6.22.18 rt73usb driver rt73 driver version - 1.0.3.6 CVS
. Couldn;t patch wpa_suplicant to get wpa2psk to work so am using iwpriv instead. The connection tends to get clobbered from time to time (not a wap problem as other machines connect to the same wap just fine).

Vern

04-05-2008 19:08:49

Hi,

The problem you're describing may be fixed in cvs20080402xx and later. Could you try the latest CVS tarball and see if you're still "scheduling atomic"?

If you are, please compile with debug enabled, and attach a gzipped copy of /var/log/kern.log to a post in this thread, if possible.

Thanks,

hariseldon99

05-05-2008 06:35:11

Hi,

The problem you're describing may be fixed in cvs20080402xx and later. Could you try the latest CVS tarball and see if you're still "scheduling atomic"?

If you are, please compile with debug enabled, and attach a gzipped copy of /var/log/kern.log to a post in this thread, if possible.

Thanks,[/quotevekippwn]

Thanks for your response. I downloaded and compiled the cvs tarball. Everything worked fine until my machine started hanging while watching mythtv over the wireless. dmesg and /var/log/kernel/errors.log show "scheduling atomic" errors

Output of dmesg
[urlvekippwn]http://pastebin.com/f3248037b[/urlvekippwn]

Output of /var/log/kernel/errors.log

[urlvekippwn]http://pastebin.com/f64b143a7[/urlvekippwn]

I compiled it with debugging and modprobed it with debug=31. Here are the /var/log/kernel/errors.log

[urlvekippwn]http://pastebin.com/m1e9d438e[/urlvekippwn]

and /var/log/syslog (too big for pastebin or attachment, private link below) files immediately afterwards

[urlvekippwn]http://www.ph.utexas.edu/~daneel/syslog.gz[/urlvekippwn]


update

and now the cvs driver causes my whole system to hang, even after I recomile without debug (

fatah

05-05-2008 10:04:52

Last time I have encountered this http//pastebin.com/f3248037b as above. I don't know how it is related. But it disappear when I change the last line of #define RT73_USB_DEVICES { ... to {}} to make it null in the end of array. Sorry if I am wrong.

Vern

05-05-2008 20:54:25

Hi hariseldon99,

You seem to have found an interesting corner case. Can you try the attached patch? It should (OK, may) help at least one of the "scheduling while atomic" messages you're getting.

Thanks,

(edit) fatah Glad to see you're looking at code. I checked just to be sure, and that particular fix is already in. Thanks.

hariseldon99

06-05-2008 00:26:05

Hi hariseldon99,

You seem to have found an interesting corner case. Can you try the attached patch? It should (OK, may) help at least one of the "scheduling while atomic" messages you're getting.

Thanks,

(edit) fatah Glad to see you're looking at code. I checked just to be sure, and that particular fix is already in. Thanks.[/quote1qqknv6k]

Hi.

Thanks for the patch. I gunzipped it in the /usr/src/rt73-<version>/Modules directory and applied

[code1qqknv6k]
patch -p0<atom5.patch
[/code1qqknv6k]

and ran

[code1qqknv6k]
make && make install
[/code1qqknv6k]

and reloaded the module. Unfortunately, my system locked up completely a few minutes later (had to hard reset, same as before) and /var/log/kernel/errors.log showed

[code1qqknv6k]
May 5 19:03:08 landau kernel: BUG: scheduling while atomic: emerald/0x10000200/8233
May 5 19:03:08 landau kernel: BUG: scheduling while atomic: X/0x10000200/6470
[/code1qqknv6k]

Update
Note that this was patched on yesterday's cvs build.


What's goin' on? ? ? ? ? ? ? ? ? ? ?

Vern

06-05-2008 18:23:32

Hi hariseldon99,

Do you have a syslog for that effort? FWIW you can trim that to a more manageable size by grepping on the timestamp. For example, using the syslog from your previous post,[code3ep3wqyk]egrep 'May 5 01:4' /var/log/syslog|gzip >syslog.gz[/code3ep3wqyk]would capture all the information needed.

Thanks,

hariseldon99

07-05-2008 01:22:15

Hi hariseldon99,

Do you have a syslog for that effort? FWIW you can trim that to a more manageable size by grepping on the timestamp. For example, using the syslog from your previous post,[code233rrwis]egrep 'May 5 01:4' /var/log/syslog|gzip >syslog.gz[/code233rrwis]would capture all the information needed.

Thanks,[/quote233rrwis]

Well I loaded the driver again a couple of hours ago and it seems to be working without errors all of a sudden. This would be the 1.0.3.6 CVS 2008050416 version with your patch above.

The syslog output so far is as follows
[url233rrwis]http://pastebin.com/f32ad1e92[/url233rrwis]

No messages in /var/log/kernel/errors.log yet... /var/log/kernel/info.log says

[url233rrwis]http://pastebin.com/m1a9c5b0b[/url233rrwis]

Should it cause a system freeze again I'll post the syslog. For now, it seems to be working. Thanks again. )

Vern

07-05-2008 01:50:33

Well I loaded the driver again a couple of hours ago and it seems to be working without errors all of a sudden.[/quote145qfqjt]Not reassuring. Are you watching mythtv over the wireless?

BTW, thanks for the quick turnaround.

Vern

09-05-2008 17:02:28

No complaints? I've put the patch into CVS.