Hi guys,
Starting with kernel 4.14-rc1, my CI failed to completely shutdown
one of the
VMs and it stuck in "in shutdown" while sending this kernel message:
[ 7496.552971] INFO: task qemu-system-x86:5978 blocked for more than 120
seconds.
[ 7496.552987] Tainted: G I
4.14.0-0.rc1.git3.1.fc28.x86_64 #1
[ 7496.552996] "echo 0 /proc/sys/kernel/hung_task_timeout_secs" disables
this message.
[ 7496.553006] qemu-system-x86 D12240 5978 1 0x00000004
[ 7496.553024] Call Trace:
[ 7496.553044] __schedule+0x2dc/0xbb0
[ 7496.553055] ? trace_hardirqs_on+0xd/0x10
[ 7496.553074] schedule+0x3d/0x90
[ 7496.553087] vhost_net_ubuf_put_and_wait+0x73/0xa0 [vhost_net]
[ 7496.553100] ? finish_wait+0x90/0x90
[ 7496.553115] vhost_net_ioctl+0x542/0x910 [vhost_net]
[ 7496.553144] do_vfs_ioctl+0xa6/0x6c0
[ 7496.553166] SyS_ioctl+0x79/0x90
[ 7496.553182] entry_SYSCALL_64_fastpath+0x1f/0xbe
[ 7496.553190] RIP: 0033:0x7fa1ea0e1817
[ 7496.553196] RSP: 002b:00007ffe3d854bc8 EFLAGS: 00000246 ORIG_RAX:
0000000000000010
[ 7496.553209] RAX: ffffffffffffffda RBX: 000000000000001d RCX:
00007fa1ea0e1817
[ 7496.553215] RDX: 00007ffe3d854bd0 RSI: 000000004008af30 RDI:
0000000000000021
[ 7496.553222] RBP: 000055e33352b610 R08: 000055e33024a6f0 R09:
000055e330245d92
[ 7496.553228] R10: 000055e33344e7f0 R11: 0000000000000246 R12:
000055e33351a000
[ 7496.553235] R13: 0000000000000001 R14: 0000000400000000 R15:
0000000000000000
[ 7496.553284]
Showing all locks held in the system:
[ 7496.553313] 1 lock held by khungtaskd/161:
[ 7496.553319] #0: (tasklist_lock){.+.+}, at: [<ffffffff8111740d>]
debug_show_all_locks+0x3d/0x1a0
[ 7496.553373] 1 lock held by in:imklog/1194:
[ 7496.553379] #0: (&f->f_pos_lock){+.+.}, at: [<ffffffff8130ecfc>]
__fdget_pos+0x4c/0x60
[ 7496.553541] 1 lock held by qemu-system-x86/5978:
[ 7496.553547] #0: (&dev->mutex#3){+.+.}, at: [<ffffffffc077e498>]
vhost_net_ioctl+0x358/0x910 [vhost_net]
I'm currently bisecting to figure out which commit breaks this but for
some reasons, when hitting this commit:
# bad: [46d4b68f891bee5d83a32508bfbd9778be6b1b63] Merge tag
'wireless-drivers-next-for-davem-2017-08-07' of
git://git.kernel.org/pub/scm/linux/kernel/git/kvalo/wireless-drivers-next
git bisect bad 46d4b68f891bee5d83a32508bfbd9778be6b1b63
the host will not allow SSHd to establish a new session and when
starting a KVM guest, the host will hard lock. I'm still bisecting but
I marked that commit as bad even though perhaps it would be good.
Hopefully, this commit was a bad one and my bisection will pinpoint which
commit broke the kernel. If you have an idea of which commit might break
the system, please let me know which one I should test first.
Thank you very much,
David Hill
[1] https://bugzilla.kernel.org/show_bug.cgi?id=197861
