D-Link DWA-140 rev D1 makes pfsense crash

petran

Hello everyone,

I am trying to setup pfsense on an old PC of mine so it can act as an AP & Captive Portal providing controlled wireless internet access through the existing LAN. (It is intended for donation to a local school where it can be of use)

After unsuccessful attempts to use existing USB WiFi dongles (atheros & realtek chips), I consulted the compatibility list and purchased a Ralink USB dongle, the D-Link DWA-140 rev D1 which appears to be using the RT5372 chip.

pfsense recognizes the device as

ugen0.3: <ralink> at usbus0
run0: <1.0> on usbus0
run0: MAC/BBP RT5392 (rev 0x0223), RF RT5372 (MIMO 2T2R), address xx:xx:xx:xx:xx:xx</ralink>

However, when I assign it to an interface and apply the changes, pfsense crashes: it dumps a lot of information on the screen and reboots. Then it reboots again when reaching the part where it loads the specific interface. It also reports something like "panic reboot due to double fault"

I have tried using it on either WAN on LAN interface without any different results. At some point, I managed to get a glimpse of the following message before it rebooted:

Driver RT3071 firmware 0.33 loaded

I've read elsewhere in the forum that this dongle has been used as an AP, however there are 5 revisions of it.

If anyone could provide ideas/suggestions as to what I could try to make it work, I would appreciate it.
I would not want to purchase yet another piece of h/w for this, but if it's a waste of time trying to make it work I'd rather not waste any more on it.

Thanks for your time, if you've read this far! :)

petran

I investigated this some more:
This usb NIC is supposed to be supported in freebsd 10.1-RELEASE by the "run(4)" driver. However, enabling it even via the ifconfig command causes it to crash the same way, right after displaying "run0: firmware RT3071 ver. 0.33 loaded".
I made a fresh pfsense install with the nic connected to the usb port and it also crashed before completing the installation. Finally, I checked the NIC in a windows system to make sure it is not faulty.

The crash dump info file contains the following lines:

Dump header from device /dev/ada0s1b
  Architecture: i386
  Architecture Version: 1
  Dump Length: 71168B (0 MB)
  Blocksize: 512
  Dumptime: Thu Feb 19 00:58:50 2015
  Hostname: pfSense.localdomain
  Magic: FreeBSD Text Dump
  Version String: FreeBSD 10.1-RELEASE-p4 #0 36d7dec(releng/10.1)-dirty: Thu Jan 22 15:12:38 CST 2015
    root@pfsense-22-i386-builder:/usr/obj.i386/usr/pfSensesrc/src/sys/pfSense_SMP.10
  Panic String: double fault
  Dump Parity: 1244061018
  Bounds: 17
  Dump Status: good

and the bottom part of the dump file is:

<118>pfSense (pfSense) 2.2-RELEASE i386 Thu Jan 22 14:04:25 CST 2015
<118>Bootup complete
...
ugen0.3: <ralink> at usbus0
run0: <1.0> on usbus0
run0: MAC/BBP RT5392 (rev 0x0223), RF RT5372 (MIMO 2T2R), address ec:22:80:0b:41:6d
run0: firmware RT3071 ver. 0.33 loaded

Fatal double fault:
eip = 0xc0ad7162
esp = 0xc47f3e50
ebp = 0xc47f42f0
cpuid = 0; apic id = 00
panic: double fault
cpuid = 0
KDB: enter: panic</ralink>

I set up pfsense in a VM on another pc and when I connected the usb nic it crashed as above.
Then I set up freebsd 10.1-RELEASE in another VM, connected the nic and started it. It did not crash.

What gives?!

stephenw10

I would guess some quirk in that particular chip is giving the driver a hard time. Have you found any reports of anyone using it successfully with pfSense or FreeBSD?

Edit: Just tested a run(4) adapter I have here and it did the exact same thing. Different Ralink chipset.

Steve

petran

I have verified it works ok with freebsd 10.1-RELEASE, however I've also found a relevant bug #4175 report in the pfsense bugtracker. It appears that the same also happens with other similar ralink chips in pfsense 2.2

I needed a quick solution, so I used two ethernet nics and an external AP instead. I had to give this USB nic back…

stephenw10

Yes, I added my info there. I tested my hardware with PC-BSD 10.1 and it ran fine.
Feel free to add your hardware info to that bug report.

Steve