[LEDE-DEV] [PATCH] [17.01] Kernel: bump to 4.4.51

Sun Feb 26 01:54:50 PST 2017

Op za, 25 feb 2017 om 9:02 , schreef Stijn Segers 
<francesco.borromini at inventati.org>:
> So...
> 
> 
> I have been running on 4.4.51 on a few days now but it looks like 
> there's some weird stuff going on on my DIR-860L... I don't know if 
> this is merely a timeline coincidence or if it's something linked to 
> .51 (or .50).
> 
> I checked its uptime and it reported < 2h (flashed it two days ago), 
> wifi connection periodically craps out, when I check if my desktop 
> still has a wired connection, it shows the router doesn't respond to 
> DHCP requests anymore either... DNS is broken as well. Restarting 
> both dnsmasq (DHCP) and Unbound (DNS) services doesn't work, they 
> keep hanging for some reason.
> 
> I have a few traces like these in logread:
> 
> Sat Feb 25 20:10:09 2017 kern.err kernel: [ 3040.980000] INFO: 
> rcu_sched detected stalls on CPUs/tasks:
> Sat Feb 25 20:10:09 2017 kern.err kernel: [ 3040.990000] 	3-...: (0 
> ticks this GP) idle=4b8/0/0 softirq=1265/1265 fqs=0
> Sat Feb 25 20:10:09 2017 kern.err kernel: [ 3041.000000] 	(detected 
> by 1, t=6004 jiffies, g=4116, c=4115, q=2999)
> Sat Feb 25 20:10:09 2017 kern.info kernel: [ 3041.010000] Task dump 
> for CPU 3:
> Sat Feb 25 20:10:09 2017 kern.info kernel: [ 3041.020000] swapper/3   
>    R running      0     0      1 0x00100000
> Sat Feb 25 20:10:09 2017 kern.warn kernel: [ 3041.030000] Stack : 
> 00000000 00000001 7fabc3c8 00000000 00000000 780022c0 804572a4 
> 80400000
> Sat Feb 25 20:10:09 2017 kern.warn kernel: [ 3041.030000] 	  8040675c 
> 00000001 00000001 80406680 80406724 80400000 00000000 8001346c
> Sat Feb 25 20:10:09 2017 kern.warn kernel: [ 3041.030000] 	  1100fc03 
> 00000003 87c74000 87c75ec0 80400000 8005dbf8 1100fc03 00000003
> Sat Feb 25 20:10:09 2017 kern.warn kernel: [ 3041.030000] 	  00000000 
> 80400000 804572a4 8005dbf0 80400000 8001abd8 1100fc03 00000000
> Sat Feb 25 20:10:09 2017 kern.warn kernel: [ 3041.030000] 	  00000004 
> 804064a0 000000a0 8001abe0 a0080040 50204041 0353017a 04405c16
> Sat Feb 25 20:10:09 2017 kern.warn kernel: [ 3041.030000] 	  ...
> Sat Feb 25 20:10:09 2017 kern.warn kernel: [ 3041.100000] Call 
> Trace:[<8001346c>] 0x8001346c
> Sat Feb 25 20:10:09 2017 kern.warn kernel: [ 3041.110000] 
> [<8005dbf8>] 0x8005dbf8
> Sat Feb 25 20:10:09 2017 kern.warn kernel: [ 3041.120000] 
> [<8005dbf0>] 0x8005dbf0
> Sat Feb 25 20:10:09 2017 kern.warn kernel: [ 3041.120000] 
> [<8001abd8>] 0x8001abd8
> Sat Feb 25 20:10:09 2017 kern.warn kernel: [ 3041.130000] 
> [<8001abe0>] 0x8001abe0
> Sat Feb 25 20:10:09 2017 kern.warn kernel: [ 3041.140000] 
> [<80214444>] 0x80214444
> Sat Feb 25 20:10:09 2017 kern.warn kernel: [ 3041.150000]
> Sat Feb 25 20:10:09 2017 kern.err kernel: [ 3041.150000] rcu_sched 
> kthread starved for 6019 jiffies! g4116 c4115 f0x0 s3 ->state=0x1
> 
> Any ideas what I should be looking for?
> 
> Other stuff (ar71xx and x86/64 namely) are still up (uninterrupted 
> since their last flash to a firmware with a 4.4.51 kernel) and I 
> haven't gotten any complaints from their users.
> 
> Stijn

As a follow-up, I reverted to a .49 build and that exhibits similar 
symptoms...

Sun Feb 26 10:46:18 2017 kern.err kernel: [ 1292.420000] INFO: 
rcu_sched detected stalls on CPUs/tasks:
Sun Feb 26 10:46:18 2017 kern.err kernel: [ 1292.430000] 1-...: (0 
ticks this GP) idle=0be/0/0 softirq=101979/101979 fqs=0
Sun Feb 26 10:46:18 2017 kern.err kernel: [ 1292.440000] (detected by 
2, t=6004 jiffies, g=4716, c=4715, q=758)
Sun Feb 26 10:46:18 2017 kern.info kernel: [ 1292.450000] Task dump for 
CPU 1:
Sun Feb 26 10:46:18 2017 kern.info kernel: [ 1292.460000] swapper/1 R 
running 0 0 1 0x00100000
Sun Feb 26 10:46:18 2017 kern.warn kernel: [ 1292.470000] Stack : 
00000000 00001400 00000005 00000000 00000000 00000000 804572a4 80400000
Sun Feb 26 10:46:18 2017 kern.warn kernel: [ 1292.470000] 8040675c 
00000001 00000001 80406540 80406724 80400000 00000000 8001346c
Sun Feb 26 10:46:18 2017 kern.warn kernel: [ 1292.470000] 00000000 
00000001 87c70000 87c71ec0 80400000 8005dbf8 1100fc03 00000001
Sun Feb 26 10:46:18 2017 kern.warn kernel: [ 1292.470000] 00000000 
80400000 804572a4 8005dbf0 80400000 8001abd8 1100fc03 00000000
Sun Feb 26 10:46:18 2017 kern.warn kernel: [ 1292.470000] 00000004 
804064a0 000000a0 8001abe0 a0080240 50205861 03530172 04405c16
Sun Feb 26 10:46:18 2017 kern.warn kernel: [ 1292.470000] ...
Sun Feb 26 10:46:18 2017 kern.warn kernel: [ 1292.540000] Call 
Trace:[<8001346c>] 0x8001346c
Sun Feb 26 10:46:18 2017 kern.warn kernel: [ 1292.550000] [<8005dbf8>] 
0x8005dbf8
Sun Feb 26 10:46:18 2017 kern.warn kernel: [ 1292.560000] [<8005dbf0>] 
0x8005dbf0
Sun Feb 26 10:46:18 2017 kern.warn kernel: [ 1292.570000] [<8001abd8>] 
0x8001abd8
Sun Feb 26 10:46:18 2017 kern.warn kernel: [ 1292.570000] [<8001abe0>] 
0x8001abe0
Sun Feb 26 10:46:18 2017 kern.warn kernel: [ 1292.580000]
Sun Feb 26 10:46:18 2017 kern.err kernel: [ 1292.580000] rcu_sched 
kthread starved for 6018 jiffies! g4716 c4715 f0x0 s3 ->state=0x1

So it's not the .50/.51 bumps at least. Wondering if SQM (Cake) has 
something to do with it.