Firmware panic, 'Unable to remove station entry...'

[Oleksij Rempel]

Am 25.04.2014 00:46, schrieb Jason Conti:
> On Thu, Apr 24, 2014 at 4:16 PM, Oleksij Rempel <linux at rempel-privat.de> wrote:
>> Looks like ot filed to allocate tx buffer:
>>
>> static struct ath_tx_buf* ath_tx_buf_alloc(struct ath_softc_tgt *sc)
>> {
>>         struct ath_tx_buf *bf = NULL;
>>
>>         bf = asf_tailq_first(&sc->sc_txbuf);
>>         if (bf !=  NULL) {
>>                 adf_os_mem_set(&bf->bf_state, 0, sizeof(struct
>> ath_buf_state));
>>                 asf_tailq_remove(&sc->sc_txbuf, bf, bf_list);
>>         } else {
>>                 adf_os_assert(0);
>>                 ^^^^^^^^^^^^^^
>>         }
>>
>>         return  bf;
>> }
>>
> 
> That's very interesting. I thought from the 0x12345678 it was hitting
> one of those asserts but I wasn't sure how to figure out the location
> in the firmware.
> 
>> What is your reproduction case? WHat is your AP, wireless mode G/N? How
>> stable is the connection? Stable transfer rate? How about host usb
>> controller?
>>
>> I will need more data to track it down...
> 
> The connection is pretty stable in general. It usually only happens
> when downloading a large file, wget/zsync an iso. However recently I
> have been hitting it on the upload portion of speedtest.comcast.net.
> It isn't always reproducible, I was surprised I hit it so quickly
> after updating the firmware. Sometimes I can go days without seeing a
> firmware panic.
> 
> AP is also a TP-LINK, TL-WR941N, it is in 11bgn mixed mode, I guess it
> is using N with the TL-WN821N wireless adapter, but I'm not sure how
> to check.

There are at least 5 version of "TL-WR941N", some of them are draft-n.
Isually speed test can confirm it. On AP with Gbit LAN i can about 25
MByte/s.

Till now i was able to reproduce some crashes only with bad connection.
Probably there are lots of retries and packets are pending it the ram of
adapter. Hmm.... maybe we went out of memory..? It would be interesting
to disable debug model in firmware to get more RAM, or recheck what
happen with te TX queue. Ideas are welcome.

> I average about 20mbps download and 5.5mbps upload on the test above,
> but I don't usually get that in general. If I hit it while downloading
> an iso I noticed the download is about 2.5mbps or higher, but I'm not
> always watching when it happens so it could be a coincidence. Between
> two machines on my network (one with ath9k and then this machine with
> ath9k-htc) I can get a sustained 10mbps both ways, but I haven't been
> able to trigger the bug that way.

I you talking about mbits, then it is really bad performance for this
devices.

> From lsusb -t:
> /:  Bus 01.Port 1: Dev 1, Class=root_hub, Driver=ehci-pci/8p, 480M
>     |__ Port 2: Dev 7, If 0, Class=Vendor Specific Class, Driver=ath9k_htc, 480M
> 
> Which is:
> Bus 001 Device 007: ID 0cf3:7015 Atheros Communications, Inc. TP-Link
> TL-WN821N v3 802.11n [Atheros AR7010+AR9287]
> 
> Thanks for taking a look, I appreciate it. Let me know if you would
> like any more information.
> 
> Jason Conti
> 
> _______________________________________________
> ath9k_htc_fw mailing list
> ath9k_htc_fw at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/ath9k_htc_fw
> 


-- 
Regards,
Oleksij

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 278 bytes
Desc: OpenPGP digital signature
URL: <http://lists.infradead.org/pipermail/ath9k_htc_fw/attachments/20140425/3baffa7d/attachment.sig>