#9654: ath: DMA failed to stop in 10 ms
------------------------+-----------------------------------
Reporter: anonymous | Owner: nbd
Type: defect | Status: reopened
Priority: normal | Milestone: Barrier Breaker 14.07
Component: kernel | Version: Trunk
Resolution: | Keywords: atheros
------------------------+-----------------------------------
Comment (by gvalkov):
Replying to [comment:538 Tronic]:
> This would be good news, but I think that is not the problem and the
fix.
> I have the same problems on a "TL-WR2543N/ND v1" running r36088. That's
a lower changeset as the one you imagine that introduces the bug.
>
> I had these issues not from the start of using r36088. It came from one
day to another.
>
> <edit>
> BTW I have running a TL-WR1043ND with r27608 without any issues!
> </edit>
Hi Tronic!
In that case we might be talking about two different bugs that both cause
similar effects. Also keep in mind that we are talking about different
hardware. Even when the issue was present on v1.8 of TL-WR1043ND, v2.1 was
happily running on the same build and was stable.
To locate the commit that introduced the bug I started with a stable
Backfire commit number and also the current number (that was unstable).
Then I performed something similar to a binary search. But I didn't cut in
half. I preferred to choose versions that are more likely to fail. This
saves time. It took on average 5-20 minutes for the issue to appear, but
in rare cases up to 90 minutes. So if it survives for 120 minutes it's
almost certainly stable. I tested 18 builds. Once I found the exact
commit, I tested many times before and after that commit. Finally I
started modifying the code to find exactly what caused the issue.
Unfortunately I don't have TL-WR2543N, so I cannot test it. But I can help
you build different version of OpenWRT and perform the tests yourself. I
also recommend that you use my tool for network performance and stability
testing. Look for '''gns''' on this ticket and there will be a download
link somewhere. Test config: wifi card uploads, 2 TCP connections.
Once you hit the issue, it is also very important to verify that the
router failed. I've seen cases where it was actually the wifi card on the
client that failed. This is a common case if you attach a USB wifi card to
a vmware guest, but I've seen this on a laptop too and I had to change the
drivers.
Some history, why I entered the OpenWRT community:
I have TL-WR1043ND v1.8, it was on official firmware and it was
disconnecting every 15 minutes, then all clients reconnect within 10
seconds. I tried all possible official firmwares and I couldn't resolve
this. So I decided to try DD-WRT. I think this was around June 2013 and
the official release was from April 2013 (note that at that time I didn't
have the tools to perform a reliable stability test). I experienced speed
drop and also wifi stop with this release, but unlike the official
firmwares, it didn't recover until the router is restarted. Then I tested
the latest trunk build of DD-WRT for a few days: speed was excellent and
no disconnects. Then I also tried OpenWRT, it seemed stable too, and I saw
it's much better organised than DD-WRT, so I decided to keep it. Around
2013-08-08 I updated to the latest OpenWRT and left the router to my room-
mate until December. He was complaining about many issues. I updated to
2013-12-19 (after r38486) and I noticed the wifi stops when under heavy
load.
This also points to the possibility for a second issue that was present
back then. I'm not sure whether it is still present? - perhaps not or at
least it is very unlikely to hit on my model.
Good luck and let me know if you need any help building or testing!
--
Ticket URL: <https://dev.openwrt.org/ticket/9654#comment:539>
OpenWrt <http://openwrt.org>
Opensource Wireless Router Technology
_______________________________________________
openwrt-tickets mailing list
[email protected]
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-tickets
