Firmware Version: DD-WRT v3.0-r32735M kongac (07/18/17)
Kernel Version: Linux 4.4.77 #366 SMP Tue Jul 18 01:06:33 CEST 2017 armv7l
Upgraded: from 32700M via ddup --flash-latest (that version crashed)
Reset: Yes, Erase NVRAM
Status: Up and running for 1 hour, basic setup as AP, 2,4GHz and 5Ghz working. SFE is off
Will test later with OpenVPN, USB, DLNA and SFE on and will report back
Errors: non so far
Temp: 64C
Big Thanks to Kong and BrainSlayer!!
Uptime 24h with SFE off
Tested USB/NAS and miniDLNA all working.
Tested OpenVPN client on PIA also working
With SFE on I have had a lock up, but ludicrous speed when it was working
Some minor GUI glitches, sometimes the GUI hangs, closing the browser and login again get things working again, GUI is much more responsive though
Do you have serial attached? I just prepared a build with some extra debug info. I've seen the rcu_stall only once and don't know if I can reproduce it. Thus I would need someone with serial that can easily reproduce the hang.
I have a serial on the WZR-1750DHPD, what do you need?
Joined: 18 Mar 2014 Posts: 12889 Location: Netherlands
Posted: Fri Jul 21, 2017 9:00 Post subject:
<Kong> wrote:
It is possible, that the crash with enabled SFE only happens if you have the webif open.
I have configured my router further with NAS/DLNA OpenVPN etc. Enabled SFE again and copied more then 100 GB without a problem.
So I erased nvram and started over again, and after the first copy of 2 GB with SFE lockup!, nothing in the logs
And per your instructions I have had the WEB gui open.
So reboot copied again without the WEBGui and transferred 10 times 2 GB without a problem.
Opened the WEB gui and expected a lockup but no, transferred again 10 times 2 GB without a problem.
Reboot, opende WEB gui transferred 2 GB and lockup!
So it seems that to provoke a lockup you have to reboot then immediately open the WEB gui and then start the transfer.
One observation the transfer always finishes and after the transfer there is a lockup.
It is possible, that the crash with enabled SFE only happens if you have the webif open.
I have configured my router further with NAS/DLNA OpenVPN etc. Enabled SFE again and copied more then 100 GB without a problem.
So I erased nvram and started over again, and after the first copy of 2 GB with SFE lockup!, nothing in the logs
And per your instructions I have had the WEB gui open.
So reboot copied again without the WEBGui and transferred 10 times 2 GB without a problem.
Opened the WEB gui and expected a lockup but no, transferred again 10 times 2 GB without a problem.
Reboot, opende WEB gui transferred 2 GB and lockup!
So it seems that to provoke a lockup you have to reboot then immediately open the WEB gui and then start the transfer.
One observation the transfer always finishes and after the transfer there is a lockup.
Furthermore what is your take on the fact that this build is slower without SFE then previous builds?
This build without SFE 30 MByte/s, previous builds 40 MByte/s
Problem is, that sfe needs to updates interface stats in the bridge code in order to accel data that travels through the bridge, there was a bug in locking the data structure. Once a user space app reads the interface stats, e.g. the webif it can stall. It is possible, that writing the stats slows down nat a bit, even if sfe module is not loaded as this code is also used when sfe is not enabled. See BS recent commits. _________________ KONG PB's: http://www.desipro.de/ddwrt/
KONG Info: http://tips.desipro.de/
Posted: Fri Jul 21, 2017 11:45 Post subject: update on my r6300
After wiping, the router seemed fine for the first couple of days. Later, I enabled the NAS portion for something and sure enough, I started experiencing lockup. Turned off sfe mode for now until I hear more word about stability. If lockups still occur, Ill probably roll back to June builds as they were stable.
Joined: 26 Dec 2013 Posts: 37 Location: Indiana, USA
Posted: Fri Jul 21, 2017 13:35 Post subject:
<Kong> wrote:
It is possible, that the crash with enabled SFE only happens if you have the webif open.
In general, could having the webif open 24/7 cause issues? I've never thought of that, but I've probably had random hangups for a year or so now (always using the latest build), and I always have the webif's open in a tab to monitor bandwidth. It's always been sporadic but any of my 3 linked kong routers are prone. Normally a router will last at least a week before I see any hang ups, but I do count on it happening. Could still be something in my config. also I'm sure. _________________ R8000~Kong R7000~Kong R6300v2~Kong WNDR3700v4~BS
It is possible, that the crash with enabled SFE only happens if you have the webif open.
In general, could having the webif open 24/7 cause issues? I've never thought of that, but I've probably had random hangups for a year or so now (always using the latest build), and I always have the webif's open in a tab to monitor bandwidth. It's always been sporadic but any of my 3 linked kong routers are prone. Normally a router will last at least a week before I see any hang ups, but I do count on it happening. Could still be something in my config. also I'm sure.
With all the mem leaks I fixed in the last couple of weeks, yes. It is possible, that after a few weeks ram is gone. _________________ KONG PB's: http://www.desipro.de/ddwrt/
KONG Info: http://tips.desipro.de/
