[LEDE-DEV] stall/hang in netifd on LEDE r1318 on Linksys WRT1900AC V1
pat
pat at patfruth.com
Wed Aug 17 12:16:12 PDT 2016
Hello Josua,
To my great satisfaction, your suggestion works beautifully !!
Thank you very much for taking time to pass this suggestion along.
Pat
> On Aug 17, 2016, at 9:08 AM, Josua Mayer <josua.mayer97 at gmail.com> wrote:
>
> Hi Syrone, Pat,
>
> I ran into the same issue on the Clearfog Pro, and my colleague figured
> out a way to make it disappear.
> Please try out this patch and let me know if it helps on your boards:
> https://github.com/Artox/lede-project/commit/db724f8ff1ed4c77668f691ed4d066a8e0f2693e
>
> From db724f8ff1ed4c77668f691ed4d066a8e0f2693e Mon Sep 17 00:00:00 2001
> From: Josua Mayer <josua.mayer97 at gmail.com>
> Date: Wed, 17 Aug 2016 16:42:07 +0200
> Subject: [PATCH] mvebu: enable cpu hotplug support in kernel
>
> This option prevents the rcu stalls in mvneta on armada-38x.
>
> Signed-off-by: Josua Mayer <josua.mayer97 at gmail.com>
> ---
> target/linux/mvebu/config-4.4 | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/target/linux/mvebu/config-4.4 b/target/linux/mvebu/config-4.4
> index d0f042e..6c4ff70 100644
> --- a/target/linux/mvebu/config-4.4
> +++ b/target/linux/mvebu/config-4.4
> @@ -209,6 +209,7 @@ CONFIG_HAVE_UID16=y
> CONFIG_HAVE_VIRT_CPU_ACCOUNTING_GEN=y
> CONFIG_HIGHMEM=y
> # CONFIG_HIGHPTE is not set
> +CONFIG_HOTPLUG_CPU=y
> CONFIG_HWBM=y
> CONFIG_HWMON=y
> CONFIG_HZ_FIXED=0
> --
> 2.6.6
>
>
> Am 15.08.2016 um 08:57 schrieb Syrone Wong:
>> I have the same issue. Everything works well on
>> https://github.com/lede-project/source/commit/22ef1c83b35cd5633b0c58c9c38a43494a906a6a,
>> boot hang when compiling
>> https://github.com/lede-project/source/commit/b9b665ae49469a73d254b1a219a4a7c4e22f27c0
>> last night.
>>
>> I'm too lazy to attach TTL cable, then I revert to the older version.
>>
>> I hope my information help.
>>
>> Best Regards,
>> Syrone Wong
>>
>>
>> On Mon, Aug 15, 2016 at 2:27 PM, pat <pat at patfruth.com> wrote:
>>> Dear LEDE devs,
>>>
>>> There doesn’t appear to be an LEDE forum yet, else I’d post it on the forum. So I’m hoping someone on the mail list has a suggestion here.
>>>
>>> I’ve been running a build of OpenWRT DD R49195 since earlier this year.
>>> I thought I’d try to move to LEDE.
>>>
>>> To that end, I’ve just built an image based on LEDE r1318 for a Linksys WRT1900AC V1 (aka mamba).
>>> The build completed successfully, and the image appears to have flashed successfully.
>>> Upon booting, the boot process stalls/hangs. I see the following in a serial/tty console;
>>>
>>> .
>>> …
>>> ….
>>> [ 16.641909] device eth0 entered promiscuous mode
>>> [ 16.649345] IPv6: ADDRCONF(NETDEV_UP): br-lan: link is not ready
>>> [ 76.692269] INFO: rcu_sched self-detected stall on CPU
>>> [ 76.697461] 1-...: (6000 ticks this GP) idle=4b7/140000000000001/0 softirq=902/919 fqs=5980
>>> [ 76.702321] INFO: rcu_sched detected stalls on CPUs/tasks:
>>> [ 76.702341] 1-...: (6000 ticks this GP) idle=4b7/140000000000001/0 softirq=902/919 fqs=5980
>>> [ 76.702352] (detected by 0, t=6002 jiffies, g=111, c=110, q=990)
>>> [ 76.702356] Task dump for CPU 1:
>>> [ 76.702368] netifd R running 0 1106 1 0x00000002
>>> [ 76.702401] [<c00101ec>] (__schedule) from [<c00da48c>] (SyS_ioctl+0x34/0x5c)
>>> [ 76.702418] [<c00da48c>] (SyS_ioctl) from [<c0009c80>] (ret_fast_syscall+0x0/0x3c)
>>> [ 76.750523] (t=6006 jiffies g=111 c=110 q=990)
>>> [ 76.755178] Task dump for CPU 1:
>>> [ 76.758420] netifd R running 0 1106 1 0x00000002
>>> [ 76.764838] [<c001fa3c>] (unwind_backtrace) from [<c001c3a4>] (show_stack+0x10/0x14)
>>> [ 76.772617] [<c001c3a4>] (show_stack) from [<c006b3b8>] (rcu_dump_cpu_stacks+0x78/0xb0)
>>> [ 76.780648] [<c006b3b8>] (rcu_dump_cpu_stacks) from [<c006e7c0>] (rcu_check_callbacks+0x28c/0x754)
>>> [ 76.789637] [<c006e7c0>] (rcu_check_callbacks) from [<c00708dc>] (update_process_times+0x38/0x64)
>>> [ 76.798543] [<c00708dc>] (update_process_times) from [<c007f738>] (tick_sched_timer+0x21c/0x260)
>>> [ 76.807358] [<c007f738>] (tick_sched_timer) from [<c0071694>] (__hrtimer_run_queues+0xf8/0x1b8)
>>> [ 76.816084] [<c0071694>] (__hrtimer_run_queues) from [<c00718ac>] (hrtimer_interrupt+0xac/0x200)
>>> [ 76.824898] [<c00718ac>] (hrtimer_interrupt) from [<c02edac0>] (armada_370_xp_timer_interrupt+0x30/0x38)
>>> [ 76.834407] [<c02edac0>] (armada_370_xp_timer_interrupt) from [<c00664f0>] (handle_percpu_devid_irq+0x6c/0x84)
>>> [ 76.844447] [<c00664f0>] (handle_percpu_devid_irq) from [<c00623c0>] (generic_handle_irq+0x24/0x34)
>>> [ 76.853521] [<c00623c0>] (generic_handle_irq) from [<c0062698>] (__handle_domain_irq+0x98/0xac)
>>> [ 76.862247] [<c0062698>] (__handle_domain_irq) from [<c0009428>] (armada_370_xp_handle_irq+0x50/0xb0)
>>> [ 76.871496] [<c0009428>] (armada_370_xp_handle_irq) from [<c000a5f4>] (__irq_svc+0x54/0x70)
>>> [ 76.879869] Exception stack(0xce1b5de8 to 0xce1b5e30)
>>> [ 76.884938] 5de0: 00000000 cf83fca4 00000000 cf83eca4 c05c6614 cf83fca4
>>> [ 76.893141] 5e00: cf83fc80 cf83f830 00000000 00000000 00000000 00000000 cf83eca8 ce1b5e38
>>> [ 76.901341] 5e20: c0028e20 c0042198 a0000013 ffffffff
>>> [ 76.906417] [<c000a5f4>] (__irq_svc) from [<c0042198>] (raw_notifier_chain_register+0x10/0x40)
>>> [ 76.915064] [<c0042198>] (raw_notifier_chain_register) from [<c0028e20>] (register_cpu_notifier+0x28/0x3c)
>>> [ 76.924759] [<c0028e20>] (register_cpu_notifier) from [<c028d29c>] (mvneta_open+0xb8/0x170)
>>> [ 76.933147] [<c028d29c>] (mvneta_open) from [<c03164a0>] (__dev_open+0x8c/0x108)
>>> [ 76.940569] [<c03164a0>] (__dev_open) from [<c0316754>] (__dev_change_flags+0xb0/0x140)
>>> [ 76.948597] [<c0316754>] (__dev_change_flags) from [<c03167fc>] (dev_change_flags+0x18/0x48)
>>> [ 76.957072] [<c03167fc>] (dev_change_flags) from [<c032b4a8>] (dev_ifsioc+0xd0/0x320)
>>> [ 76.964928] [<c032b4a8>] (dev_ifsioc) from [<c032bf60>] (dev_ioctl+0x7f4/0x8c0)
>>> [ 76.972266] [<c032bf60>] (dev_ioctl) from [<c00da408>] (do_vfs_ioctl+0x6a4/0x6f4)
>>> [ 76.979772] [<c00da408>] (do_vfs_ioctl) from [<c00da48c>] (SyS_ioctl+0x34/0x5c)
>>> [ 76.987106] [<c00da48c>] (SyS_ioctl) from [<c0009c80>] (ret_fast_syscall+0x0/0x3c)
>>>
>>> It seems netifd is hung (maybe in the mvneta driver)????
>>> Has anyone else already seen this?
>>> What is causing this?
>>> How do I fix it?
>>>
>>> Thanks
>>>
>>>
>>> _______________________________________________
>>> Lede-dev mailing list
>>> Lede-dev at lists.infradead.org
>>> http://lists.infradead.org/mailman/listinfo/lede-dev
>>
>> _______________________________________________
>> Lede-dev mailing list
>> Lede-dev at lists.infradead.org
>> http://lists.infradead.org/mailman/listinfo/lede-dev
>>
More information about the Lede-dev
mailing list