Re: Shutting down a VM with Kernel 4.14 will sometime hang and a reboot is the only way to recover.

From: David Hill <dhill@redhat.com>
To: Harald Moeller <h.moeller@hakimo.net>, kvm@vger.kernel.org
Subject: Re: Shutting down a VM with Kernel 4.14 will sometime hang and a reboot is the only way to recover.
Date: Wed, 6 Dec 2017 21:44:46 -0500	[thread overview]
Message-ID: <f657502a-e102-ddcc-6ca0-9d52ad573018@redhat.com> (raw)
In-Reply-To: <08dddfe6-97c0-95a3-8b26-327c14d618a8@hakimo.net>

Have you tried adding this:

cat<<EOF>/etc/modprobe.d/vhost-net.conf
options vhost_net  experimental_zcopytx=0
EOF

reboot

Other than this, you can try bisecting but in my case, the system wont 
boot when reaching a given commit.

On 2017-12-02 11:37 AM, Harald Moeller wrote:
> Hello, my name is Harry and this is my first post here, hope I'm doing 
> this the right way, sorry if not ...
>
> I'm not a subscriber to the full list yet so I understand I shall ask 
> you to be personally CCed.
>
> I am following this as I do experience the same (or sort-a same) issue 
> with 4.14.2.
>
> My setup is more simple, just an oVirt host shutting down some VMs. 
> Doesn't happen all the time but I'd say around 3 from 10.
>
> This is what I see (slightly different from David):
>
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: INFO: task qemu-kvm:1173 
> blocked for more than 120 seconds.
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel:       Tainted: G          
> I     4.14.2-1.el7.hakimo.x86_64 #4
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: "echo 0 > 
> /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: qemu-kvm        D 0 
> 1173      1 0x00000084
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: Call Trace:
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: __schedule+0x28d/0x880
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel:  schedule+0x36/0x80
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: 
> vhost_net_ubuf_put_and_wait+0x61/0x90 [vhost_net]
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel:  ? 
> remove_wait_queue+0x60/0x60
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: 
> vhost_net_ioctl+0x317/0x8e0 [vhost_net]
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: do_vfs_ioctl+0xa7/0x5f0
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel:  SyS_ioctl+0x79/0x90
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: do_syscall_64+0x67/0x1b0
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: 
> entry_SYSCALL64_slow_path+0x25/0x25
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: RIP: 0033:0x7fb8862d1107
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: RSP: 002b:00007fff4acd7e58 
> EFLAGS: 00000246 ORIG_RAX: 0000000000000010
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: RAX: ffffffffffffffda RBX: 
> 000055abaa2d29c0 RCX: 00007fb8862d1107
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: RDX: 00007fff4acd7e60 RSI: 
> 000000004008af30 RDI: 0000000000000028
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: RBP: 00007fff4acd7e60 R08: 
> 000055aba805e10f R09: 00000000ffffffff
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: R10: 0000000000000004 R11: 
> 0000000000000246 R12: 000055ababf32510
> Dec 01 16:11:53 oVirtHost01.xyz.net kernel: R13: 0000000000000001 R14: 
> 000055ababf32498 R15: 000055abaa2a0b40
>
> This is still happening after reverting the three suggested commits
>
> 1f8b977ab32dc5d148f103326e80d9097f1cefb5 ("sock: enable MSG_ZEROCOPY")
>
> c1d1b437816f0afa99202be3cb650c9d174667bc ("net: convert (struct 
> ubuf_info)->refcnt to refcount_t")
>
> 581fe0ea61584d88072527ae9fb9dcb9d1f2783e {"net: orphan frags on 
> stand-alone ptype in dev_queue_xmit_nit"}
>
> Anything I could be helpful with trying to solve this? Any more info I 
> could provide?
>
> Harry
>