#11862: ath: Failed to stop TX DMA, queues=0x004
------------------------------+-----------------------------------
  Reporter:  igor@…           |      Owner:  nbd
      Type:  defect           |     Status:  assigned
  Priority:  response-needed  |  Milestone:  Barrier Breaker 14.07
 Component:  kernel           |    Version:  Trunk
Resolution:                   |   Keywords:
------------------------------+-----------------------------------

Comment (by syzop@…):

 I agree with [comment:283 ericwong]'s comment that it appears channel
 related. I have about 35 access points (WNDR3700v2, WNDR3700v4 and
 WNDR3800), and about 80-90% that had the problem were all on channel 11,
 even though I had channel 1, 6 and 11 usage split almost perfectly
 33%/33%/33% between all the AP's (and no two ap's with same channel next
 to each other, the usual kind of stuff).

 When I changed all the access points that were using channel 11 to channel
 1 or 6 it had a BIG positive impact on the number of TX DMA errors... they
 went from about 800 per day to around 250 (total errors of all AP's
 combined). Actually that's still with a few access points on channel 11,
 it will probably get near 100 per day if I change those last ones too.

 When I say channel related, it doesn't necessarily mean that channel 11 is
 "faulty" or something, more likely that channel is just saturated or has
 some interference and that's what triggering this error.
 Which, by the way, brings me to another feeling I have: that this TX DMA
 error we are all seeing may not be the SOURCE of the problem but rather a
 RESULT of a problem, such as over-crowded channel and/or interference.
 Anyway, I have no proof to support this statement. The Channel Saturation
 statistics I'm getting are unreliable for some reason.
 Makes you wonder if this error simply means that the band is so crowded at
 the time that the send request (TX DMA) times out because it was unable to
 send (at all) in time because the AP is waiting for the band to clear
 before transmit.

 On a side note, the real solution for us.. with so much crowded traffic..
 is to move as many clients to 5GHz as possible. On the 5GHz band I rarely
 see any TX DMA errors, talking about 1% of the error rate I see on the
 2.4GHz band, often even zero.
 Unfortunately that's easier said than done, most equipment is not from us.
 Fortunately many NEW devices are 5GHz capable nowadays, but even just a
 year ago dual band was not too common for laptops.. and the fact that you
 can't just simply buy and put a dual band card in every laptop due to
 manufacturers maintaining a silly WiFi whitelist in the BIOS is not
 exactly helping. Ending up patching BIOS's etc... Anyway, that's a
 completely different subject...

--
Ticket URL: <https://dev.openwrt.org/ticket/11862#comment:290>
OpenWrt <http://openwrt.org>
Opensource Wireless Router Technology
_______________________________________________
openwrt-tickets mailing list
[email protected]
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-tickets

Reply via email to