Project

General

Profile

Actions

Bug #4722

closed

Ralink USB driver yields a double fault panic on pfSense, works on FreeBSD with equivalent config

Added by Jim Pingle over 9 years ago. Updated about 9 years ago.

Status:
Needs Patch
Priority:
Very Low
Assignee:
-
Category:
Wireless
Target version:
-
Start date:
05/21/2015
Due date:
% Done:

0%

Estimated time:
Plus Target Version:
Release Notes:
Affected Version:
Affected Architecture:
All

Description

I've got a Ralink USB wireless adapter (Buffalo WLI-UC-GNM) that works perfectly on stock FreeBSD (10.1-STABLE) but when used on pfSense, crashes/panics. Tried on 2.2.2, 2.2.3, and 2.3 snapshots.

Adapter info from dmesg (address redacted):

run0: <1.0> on usbus4
run0: MAC/BBP RT3070 (rev 0x0201), RF RT3020 (MIMO 1T1R), address cc:e1:d5:xx:xx:xx
run0: firmware RT2870 ver. 0.33 loaded

From usbconfig:

ugen4.2: <802.11 n WLAN Ralink> at usbus4, cfg=0 md=HOST spd=HIGH (480Mbps) pwr=ON (450mA)

  bLength = 0x0012 
  bDescriptorType = 0x0001 
  bcdUSB = 0x0200 
  bDeviceClass = 0x0000  <Probed by interface class>
  bDeviceSubClass = 0x0000 
  bDeviceProtocol = 0x0000 
  bMaxPacketSize0 = 0x0040 
  idVendor = 0x0411 
  idProduct = 0x01a2 
  bcdDevice = 0x0101 
  iManufacturer = 0x0001  <Ralink>
  iProduct = 0x0002  <802.11 n WLAN>
  iSerialNumber = 0x0003  <1.0>
  bNumConfigurations = 0x0001

The same crash happens with either hostap mode or BSS mode. Full crash dump is attached. Here is the backtrace:

Tracing pid 0 tid 100179 td 0xc75b5310
kdb_enter(c1448f13,c1448f13,c16052cf,c207b1b4,0,...) at kdb_enter+0x3d/frame 0xc207b168
panic(c16052cf,0,0,0,ef39e418,...) at panic+0x144/frame 0xc207b1a8
dblfault_handler() at dblfault_handler+0xab/frame 0xc207b1a8
--- trap 0x17, eip = 0xc0af43c2, esp = 0xef39df78, ebp = 0xef39e418 ---
run_select_chan_group(c78b3800,c783d2d4,ef39ea38,0,0,...) at run_select_chan_group+0x12/frame 0xef39e418
run_set_chan(c78b3800,c783d2d4,ef39f878,0,0,...) at 0xc0af2964/frame 0xef39ee08
run_init_locked(ef39fbd8,c12ad2eb,c75b5310,0,1,...) at 0xc0b12d8d/frame 0xef39fbb0
run_ioctl(c7889000,80206910,0) at run_ioctl+0x2d7/frame 0xef39fc14
parent_updown(c7889000,1,0,0,0,...) at parent_updown+0x22/frame 0xef39fc28
taskqueue_run_locked(c76fc880,c76fc898,0,c143554f,0,...) at taskqueue_run_locked+0xee/frame 0xef39fc6c
taskqueue_thread_loop(c6d900a4,ef39fce8,0,0,0,...) at taskqueue_thread_loop+0xc7/frame 0xef39fca4
fork_exit(c0d57fc0,c6d900a4,ef39fce8) at fork_exit+0xa3/frame 0xef39fcd4
fork_trampoline() at fork_trampoline+0x8/frame 0xef39fcd4

Looks similar to a crash someone else encountered on a custom NanoBSD (non-pfSense) build vs. stock FreeBSD:
https://www.mail-archive.com/freebsd-wireless@freebsd.org/msg03525.html


Files

run_crash.txt (157 KB) run_crash.txt Jim Pingle, 05/21/2015 08:27 AM
Actions #1

Updated by Chris Buechler over 9 years ago

  • Status changed from New to Confirmed
  • Affected Version changed from 2.2.3 to 2.2.x

guessing this is probably all 2.2.x versions.

Does the workaround in #4740 also work around this?

Actions #2

Updated by Jim Pingle over 9 years ago

Apparently so. Moving the sleep down below the other line allows it to function. Occasionally drops an error on the console but I have a client behind it connected and working.

run0: firmware RT2870 ver. 0.33 loaded
run0_wlan0: ieee80211_new_state_locked: pending RUN -> SCAN transition lost

Doesn't seem to have a negative impact otherwise. Seems like a safe change also.

Actions #3

Updated by Jim Pingle over 9 years ago

Spoke too soon, I went back and tried it on the original hardware that was used to replicate the problem and it still crashes there with the sleep before or after.

Moving the USB dongle to an APU I can't make it crash there at all even with a stock 2.2.3. May just have to chalk this up to some other hardware combination issue.

Actions #4

Updated by Chris Buechler about 9 years ago

  • Status changed from Confirmed to Needs Patch
  • Target version deleted (2.3)
  • Affected Version deleted (2.2.x)

seems specific to certain combinations of hardware, and doesn't appear to be in anything we're doing.

Actions

Also available in: Atom PDF