grsecurity forums

Dear PaX Team & Spender,

I have encountered the "NMI watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [readlink / systemd-udevd]" messages followed by:

Code: Select all

INFO: rcu_sched detected stalls on CPU/tasks:
$1-...: (1 GPs behind) idle=327/1400000000000000/0 softirq=1209/1210 fqs=2
$(detected by 2, t=60007 jiffies, g=154, c=153, q=626)
rcu_sched kthread starved for 60004 jiffies! g154 c153 f0x2 RCU_GP_WAIT_FQS(3) ->state=0x0

please refer to this image:

I have used the following combinations:

Code: Select all

grsecurity-3.1-4.7.1-201608161813.patch over vanilla linux-4.7.1
grsecurity-3.1-4.7.2-201608211829.patch over vanilla linux-4.7.2

Please note, that both vanilla Linux kernels are working fine without the grsecurity patch applied.


Any ideas?

Thanks in advance!

Kind Regards,
arno

you should post a full dmesg and perhaps get a backtrace on each cpu via sysrq just to see who's spinning on what.

I had this fatal bug for about a month with 4.7.1 and 4.7.2 on all KVM virtual machines (debian/suse/arch on 2 different hosts/guests combo), grsec-only, I could almost not use grsec anymore because of it, only one in five boot attempts went through.

It appears to have been fixed in the .patch last week.

I had the same with 4.7.1 and 4.7.2 + grsec patches on that time. (both on my laptop and in qemu)

Now obtained the 4.7.3 + grsecurity-3.1-4.7.3-201609072139.patch and everything is working fine!