Re: iwlagn is getting very shaky

["Guy, Wey-Yi" <wey-yi.w.guy@intel.com>]

hi all,

On Tue, 2011-10-18 at 23:25 -0700, Norbert Preining wrote:
> Hi David, hi all
> 
> On Di, 18 Okt 2011, David Rientjes wrote:
> > There have been recent issues in 3.1-rc9 reported with iwlagn, see the 
> > thread at https://lkml.org/lkml/2011/10/15/107 even though you have 
> 
> Interesting. I read through the thread and activated the debugfs 
> option.
> 
> I could get my hardware back by
> 	echo 1 > /sys/kernel/debug/ieee80211/phy0/iwlagn/debug/force_reset
> 
> [ 2761.352629] ieee80211 phy0: Hardware restart was requested
> [ 2761.352714] iwlagn 0000:06:00.0: L1 Enabled; Disabling L0S
> [ 2761.355763] iwlagn 0000:06:00.0: Radio type=0x1-0x2-0x0
> [ 2779.484308] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 1/3)
> [ 2779.684128] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 2/3)
> [ 2779.884087] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 3/3)
> [ 2780.084079] wlan0: direct probe to 00:24:c4:ab:bd:e0 timed out
> [ 2788.051381] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 1/3)
> [ 2788.248079] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 2/3)
> [ 2788.448083] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 3/3)
> [ 2788.648140] wlan0: direct probe to 00:24:c4:ab:bd:e0 timed out
> [ 2796.614710] wlan0: authenticate with 00:24:c4:ab:bd:ef (try 1)
> [ 2796.615623] wlan0: authenticated
> [ 2796.618046] wlan0: associate with 00:24:c4:ab:bd:ef (try 1)
> [ 2796.622748] wlan0: RX AssocResp from 00:24:c4:ab:bd:ef (capab=0x1 status=0 aid=1)
> [ 2796.622751] wlan0: associated
> [ 2871.224192] e1000e: eth0 NIC Link is Down
> 
> I unplugged the cable and could ping the world still, nice....
> 
> After a short time I got:
> [ 2895.575964] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 1/3)
> [ 2895.772067] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 2/3)
> [ 2895.972101] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 3/3)
> [ 2896.172054] wlan0: direct probe to 00:24:c4:ab:bd:e0 timed out
> [ 2905.316968] wlan0: deauthenticating from 00:24:c4:ab:bd:ef by local choice (reason=2)
> [ 2905.356316] cfg80211: Calling CRDA to update world regulatory domain
> [ 2905.361965] wlan0: authenticate with 00:24:c4:ab:bd:e0 (try 1)
> [ 2905.560063] wlan0: authenticate with 00:24:c4:ab:bd:e0 (try 2)
> [ 2905.760091] wlan0: authenticate with 00:24:c4:ab:bd:e0 (try 3)
> [ 2905.960077] wlan0: authentication with 00:24:c4:ab:bd:e0 timed out
> [ 2913.908984] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 1/3)
> [ 2914.108116] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 2/3)
> [ 2914.308116] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 3/3)
> [ 2914.508103] wlan0: direct probe to 00:24:c4:ab:bd:e0 timed out
> [ 2922.473062] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 1/3)
> [ 2922.672109] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 2/3)
> [ 2922.872106] wlan0: direct probe to 00:24:c4:ab:bd:e0 (try 3/3)
> [ 2923.072103] wlan0: direct probe to 00:24:c4:ab:bd:e0 timed out
> 
> And at this time the tx_queue showed me:
> -----------------------------------------------------------
> hwq 00: read=91 write=91 stop=1 swq_id=0x00 (ac 0/hwq 0)s.

>         stop-count: 1

it is very interesting, for sure there is a bug here which cause NIC
stop working, if you look at the tx queue, hwq 0 is stop, which mean
nothing go out. I am not sure how we get into this? yes, most likely
force_reset will fix it by reload the firmware and reset all the queues

Could you help me how to repro this problem?

Thanks
Wey


> hwq 01: read=0 write=0 stop=0 swq_id=0x05 (ac 1/hwq 1)
>         stop-count: 0
> hwq 02: read=127 write=127 stop=0 swq_id=0x0a (ac 2/hwq 2)
>         stop-count: 0
> hwq 03: read=0 write=0 stop=0 swq_id=0x0f (ac 3/hwq 3)
>         stop-count: 0
> hwq 04: read=13 write=13 stop=0 swq_id=0x00 (ac 0/hwq 0)
> hwq 05: read=0 write=0 stop=0 swq_id=0x00 (ac 0/hwq 0)
> hwq 06: read=0 write=0 stop=0 swq_id=0x00 (ac 0/hwq 0)
> hwq 07: read=0 write=0 stop=0 swq_id=0x00 (ac 0/hwq 0)
> hwq 08: read=0 write=0 stop=0 swq_id=0x00 (ac 0/hwq 0)
> hwq 09: read=0 write=0 stop=0 swq_id=0x00 (ac 0/hwq 0)
> hwq 10: read=0 write=0 stop=0 swq_id=0x2a (ac 2/hwq 10)
> hwq 11: read=0 write=0 stop=0 swq_id=0x2c (ac 0/hwq 11)
> hwq 12: read=0 write=0 stop=0 swq_id=0x00 (ac 0/hwq 0)
> hwq 13: read=0 write=0 stop=0 swq_id=0x00 (ac 0/hwq 0)
> hwq 14: read=0 write=0 stop=0 swq_id=0x00 (ac 0/hwq 0)
> hwq 15: read=0 write=0 stop=0 swq_id=0x00 (ac 0/hwq 0)
> hwq 16: read=0 write=0 stop=0 swq_id=0x00 (ac 0/hwq 0)
> hwq 17: read=0 write=0 stop=0 swq_id=0x00 (ac 0/hwq 0)
> hwq 18: read=0 write=0 stop=0 swq_id=0x00 (ac 0/hwq 0)
> hwq 19: read=0 write=0 stop=0 swq_id=0x00 (ac 0/hwq 0)
> -------------------------------------------------
> 
> Hope that helps. Anyone let me know if you need more testing.
> 
> Once more, be reminded the the firmware of the iwlagn is from
> an experimental build that should solve the AGN stopped working
> problem.
> 
> Best wishes
> 
> Norbert
> ------------------------------------------------------------------------
> Norbert Preining            preining@{jaist.ac.jp, logic.at, debian.org}
> JAIST, Japan                                 TeX Live & Debian Developer
> DSA: 0x09C5B094   fp: 14DF 2E6C 0307 BE6D AD76  A9C0 D2BF 4AA3 09C5 B094
> ------------------------------------------------------------------------
> SCRAMOGE (vb.)
> To cut oneself whilst licking envelopes.
> 			--- Douglas Adams, The Meaning of Liff