[meta-freescale] linux-fslc 4.1.7: rcu_preempt detected stalls

[Lars Michael]

Dear all,
Running kernel linux-fslc 4.1.7 (fido release) on an i.MX6Q board is pretty stable but sometimes we get a kernel crash, in one incident this trace was logged:

2018-08-16 18:21:54 cpb539 kern.err kernel:INFO: rcu_preempt detected stalls on CPUs/tasks: {} (detected by 2, t=2102 jiffies, g=71882781, c=71882780, q=7659)2018-08-16 18:21:54 cpb539 kern.err kernel:All QSes seen, last rcu_preempt kthread activity 2101 (103484804-103482703), jiffies_till_next_fqs=1, root ->qsmask 0x02018-08-16 18:21:54 cpb539 kern.info kernel:swapper/2 R running 0 0 1 0x000000002018-08-16 18:21:54 cpb539 kern.warning kernel:[<80016914>] (unwind_backtrace) from [<80013308>] (show_stack+0x10/0x14)2018-08-16 18:21:54 cpb539 kern.warning kernel:[<80013308>] (show_stack) from [<8006e590>] (rcu_check_callbacks+0x880/0x88c)2018-08-16 18:21:54 cpb539 kern.warning kernel:[<8006e590>] (rcu_check_callbacks) from [<800708d8>] (update_process_times+0x38/0x64)2018-08-16 18:21:54 cpb539 kern.warning kernel:[<800708d8>] (update_process_times) from [<80080c58>] (tick_sched_timer+0x58/0x9c)2018-08-16 18:21:54 cpb539 kern.warning kernel:[<80080c58>] (tick_sched_timer) from [<80071608>] (__run_hrtimer.isra.5+0x44/0xd4)2018-08-16 18:21:54 cpb539 kern.warning kernel:[<80071608>] (__run_hrtimer.isra.5) from [<8007194c>] (hrtimer_interrupt+0x100/0x2fc)2018-08-16 18:21:54 cpb539 kern.warning kernel:[<8007194c>] (hrtimer_interrupt) from [<80016054>] (twd_handler+0x30/0x38)2018-08-16 18:21:54 cpb539 kern.warning kernel:[<80016054>] (twd_handler) from [<800657e8>] (handle_percpu_devid_irq+0x88/0xa0)2018-08-16 18:21:54 cpb539 kern.warning kernel:[<800657e8>] (handle_percpu_devid_irq) from [<800618ec>] (generic_handle_irq+0x2c/0x3c)2018-08-16 18:21:54 cpb539 kern.warning kernel:[<800618ec>] (generic_handle_irq) from [<80061ba4>] (__handle_domain_irq+0x7c/0xec)2018-08-16 18:21:54 cpb539 kern.warning kernel:[<80061ba4>] (__handle_domain_irq) from [<800093c0>] (gic_handle_irq+0x24/0x5c)2018-08-16 18:21:54 cpb539 kern.warning kernel:[<800093c0>] (gic_handle_irq) from [<80013e00>] (__irq_svc+0x40/0x74)2018-08-16 18:21:54 cpb539 kern.warning kernel:Exception stack(0xbf099fa0 to 0xbf099fe8)2018-08-16 18:21:54 cpb539 kern.warning kernel:9fa0: 00000002 00000000 b67c4540 8001fb00 bf098000 8087f59c 805ec408 808773042018-08-16 18:21:54 cpb539 kern.warning kernel:9fc0: bf099ff0 808c1b90 00000001 00000000 01000000 bf099fe8 80010b48 80010b4c2018-08-16 18:21:54 cpb539 kern.warning kernel:9fe0: 600f0013 ffffffff2018-08-16 18:21:54 cpb539 kern.warning kernel:[<80013e00>] (__irq_svc) from [<80010b4c>] (arch_cpu_idle+0x38/0x3c)2018-08-16 18:21:54 cpb539 kern.warning kernel:[<80010b4c>] (arch_cpu_idle) from [<800594a4>] (cpu_startup_entry+0x12c/0x1c4)2018-08-16 18:21:54 cpb539 kern.warning kernel:[<800594a4>] (cpu_startup_entry) from [<1000948c>] (0x1000948c)2018-08-16 18:21:54 cpb539 kern.err kernel:rcu_preempt kthread starved for 2101 jiffies!
A similiar stall got successfully fixed by applying this patch:0001-inet-fix-potential-deadlock-in-reqsk_queue_unlink.patch
I know we are on a dated kernel version, but are there any suggestions on patches that could possibly fix this?
The plan is of course to move to a more recent stable kernel, but a short term fix is really needed.
Thanks and regards,Lars
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.yoctoproject.org/pipermail/meta-freescale/attachments/20180920/74b73390/attachment.html>