Re: Turn off passive scanning?

[Derek Atkins <warlord MIT EDU>]

Dan Williams <dcbw redhat com> writes:

>> Unfortunately the crash is very low level.  Even when siting on
>> the console the system locks up hard.  I couldn't even page-up/page-down
>> to see the full stack trace (and I didn't spend the time to leave
>> my meeting, find a piece of paper, and copy it all down by hand).
>> 
>> So...  Is there some way I can turn off passive scanning in NM?
>
> Not really; you should be able to achieve the same result by
> periodically running 'iwlist wlan0 scan' without NM running; there
> clearly needs to be some fixes to the driver here...

Man, is this a hard crash to capture!  When it happens it occurs
at such a low level that the machine hangs completely.  Even
the console hangs.  But you're absolutely right here.  I was able
to reproduce the crash somewhat reliably using this script in my
current environment here at the IETF:

  watch -n 1 iwlist wlan0 scan

Of course it's hard to reliably reproduce WHEN it will occur, so I
missed it a couple times.  Moreover, I couldn't get the whole stack
trace because it went off the top of the console and I couldn't page
up to see it.

> It may well not have anything to do with "passive" scanning itself, but
> just scanning in general.  There's passive (card just jumps to a channel
> and listens for a split second, then to the next channel and listen,
> etc) or active (jump to a channel, send a probe request, wait short time
> for response).

Honestly, I don't know.  This trace is somewhat different than the last
one.  This one doesn't appear to have any mac80211 APIs.  Hopefully
this (partial) stack trace is useful, but I have no idea to whom I
should send it.  Note that I'll be in this environment through the
end of the week so I can test fixes for three more days.

If this stack trace isn't usable enough I can try to change my
console to display new lines, but I don't recall how to do that
easily.  Using a serial console would be challenging on my laptop
as I have no serial port.

Let me know what you'd like me to try.

-derek

> Dan

---- top of screen ----
[<c04281d4>] ? try_to_wake_up+0x230/0x23b
[<c0437de5>] ? signal_wake_up+0x31/0x34
[<c0437ef7>] ? complete_signal+0x10f/0x11f
[<c048eecd>] ? __slab_free+0x63/0x263
[<c048eecd>] ? __slab_free+0x63/0x263
[<c048f2f6>] ? kmem_cache_free+0x71/0xa7
[<c062e288>] ? __kfree_skb+0x63/0x66
[<c062e288>] ? __kfree_skb+0x63/0x66
[<c0631e1c>] ? net_tx_action+0x6b/0xf3
[<c04327d2>] ? __do_softirq+0xff/0x109
[<c04326d3>] ? __do_softirq_0x0/0x109
[<c0406f1c>] ? do_softirq+0x77/0xdb
[<c0420902>] ? pick_next_task+0x1d/0x34
[<c0567b66>] ? acpi_idle_enter_bm+0x2b5/0x2cc
[<c0447f99>] ? tick_nohz_request_sched_tick+0x13a/0x14d
[<c0403c8d>] ? cpu_idle+0x12d/0x134
[<c069a9b2>] ? rest_init+0x43/0x50
===================
Code: 03 00 00 00 7b 00 00 00 7b 00 00 00 d8 00 40 c0 ff ff ff ff cb 3e
88 c0 60 00 00 00 02 02 21 00 89 ab 6a c0 dc 3e 88 c0 7c 21 42 <c0> 10
c0 c2 ef dc 3e 88 c0 4b ad 6a c0 10 c0 c2 ef 08 3f 88 c0
EIP: [<c0883ecb>] softirq_stack+0xecb/0x20000 SS:ESP 0068:c0883ec0
Kernel panic - not syncing: Fatal exception in interrupt

-- 
       Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
       Member, MIT Student Information Processing Board  (SIPB)
       URL: http://web.mit.edu/warlord/    PP-ASEL-IA     N1NWH
       warlord MIT EDU                        PGP key available