When I hot-plug a CPU to an openSUSE VM running on KVM, I get following kernel oops:
# uname -a Linux opensuse 3.16.6-2-default #1 SMP Mon Oct 20 13:47:22 UTC 2014 (feb42ea) x86_64 x86_64 x86_64 GNU/Linux # dmesg [ 54.927758] CPU2 has been hot-added [ 54.928610] smpboot: Booting Node 0 Processor 2 APIC 0x2 [ 0.004000] kvm-clock: cpu 2, msr 2:403de081, secondary cpu clock [ 54.940001] TSC synchronization [CPU#1 -> CPU#2]: [ 54.940001] Measured 195274448514 cycles TSC warp between CPUs, turning off TSC clock. [ 54.940001] tsc: Marking TSC unstable due to check_tsc_sync_source failed [ 54.960222] KVM setup async PF for cpu 2 [ 54.960265] kvm-stealtime: cpu 2, msr 24008d240 [ 54.960342] microcode: CPU2 sig=0x306c1, pf=0x1, revision=0x1 [ 55.012464] Will online and init hotplugged CPU: 2 [ 55.012574] kobject '(null)' (ffff88024009a3d8): tried to add an uninitialized object, something is seriously wrong. [ 55.012615] CPU: 1 PID: 892 Comm: systemd-udevd Not tainted 3.16.6-2-default #1 [ 55.012649] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.7.5-20140531_171129-lamiak 04/01/2014 [ 55.012688] ffff8802347cbd98 ffffffff815ca482 ffff88024009a3d8 ffffffff8130e153 [ 55.012729] ffffffff00000020 ffff8802347cbda8 ffff8802347cbd50 00000001334da7a8 [ 55.012766] 000000000000b0a0 0000000001ca6930 0000000000000002 ffff8802334da298 [ 55.012804] Call Trace: [ 55.012845] [<ffffffff81005096>] dump_trace+0x86/0x330 [ 55.012881] [<ffffffff810053d4>] show_stack_log_lvl+0x94/0x170 [ 55.012919] [<ffffffff81006a31>] show_stack+0x21/0x50 [ 55.012955] [<ffffffff815ca482>] dump_stack+0x41/0x51 [ 55.012990] [<ffffffff8130e153>] kobject_add+0xa3/0xb0 [ 55.013027] [<ffffffff812ed4a1>] blk_mq_register_hctx+0x91/0xb0 [ 55.013062] [<ffffffff812edd5e>] blk_mq_sysfs_register+0x3e/0x50 [ 55.013095] [<ffffffff812eb9c2>] blk_mq_queue_reinit_notify+0x72/0xb0 [ 55.013129] [<ffffffff810806bc>] notifier_call_chain+0x4c/0x70 [ 55.013162] [<ffffffff8105beef>] cpu_notify+0x1f/0x40 [ 55.013195] [<ffffffff8105c0d7>] _cpu_up+0x157/0x170 [ 55.013226] [<ffffffff8105c169>] cpu_up+0x79/0xa0 [ 55.013260] [<ffffffff815bf555>] cpu_subsys_online+0x35/0x80 [ 55.013294] [<ffffffff8140c64d>] device_online+0x5d/0xa0 [ 55.013328] [<ffffffff8140c705>] online_store+0x75/0x80 [ 55.013360] [<ffffffff8122127a>] kernfs_fop_write+0xda/0x150 [ 55.013394] [<ffffffff811b1aa2>] vfs_write+0xb2/0x1f0 [ 55.013426] [<ffffffff811b25e2>] SyS_write+0x42/0xa0 [ 55.013462] [<ffffffff815d0c2d>] system_call_fastpath+0x1a/0x1f [ 55.013517] [<00007f9fbf3804e0>] 0x7f9fbf3804df
I can reproduce it on a Ubuntu 14.10 host with stock KVM and on Debian host with a custom KVM. The oops sometimes causes the virtual machine to freeze.
When I hot-plug a CPU to an openSUSE VM running on KVM, I get following kernel oops:
# uname -a sync_source failed 171129- lamiak 04/01/2014 096>] dump_trace+ 0x86/0x330 3d4>] show_stack_ log_lvl+ 0x94/0x170 a31>] show_stack+ 0x21/0x50 482>] dump_stack+ 0x41/0x51 153>] kobject_ add+0xa3/ 0xb0 4a1>] blk_mq_ register_ hctx+0x91/ 0xb0 d5e>] blk_mq_ sysfs_register+ 0x3e/0x50 9c2>] blk_mq_ queue_reinit_ notify+ 0x72/0xb0 6bc>] notifier_ call_chain+ 0x4c/0x70 eef>] cpu_notify+ 0x1f/0x40 0d7>] _cpu_up+0x157/0x170 169>] cpu_up+0x79/0xa0 555>] cpu_subsys_ online+ 0x35/0x80 64d>] device_ online+ 0x5d/0xa0 705>] online_ store+0x75/ 0x80 27a>] kernfs_ fop_write+ 0xda/0x150 aa2>] vfs_write+ 0xb2/0x1f0 5e2>] SyS_write+0x42/0xa0 c2d>] system_ call_fastpath+ 0x1a/0x1f 4e0>] 0x7f9fbf3804df
Linux opensuse 3.16.6-2-default #1 SMP Mon Oct 20 13:47:22 UTC 2014 (feb42ea) x86_64 x86_64 x86_64 GNU/Linux
# dmesg
[ 54.927758] CPU2 has been hot-added
[ 54.928610] smpboot: Booting Node 0 Processor 2 APIC 0x2
[ 0.004000] kvm-clock: cpu 2, msr 2:403de081, secondary cpu clock
[ 54.940001] TSC synchronization [CPU#1 -> CPU#2]:
[ 54.940001] Measured 195274448514 cycles TSC warp between CPUs, turning off TSC clock.
[ 54.940001] tsc: Marking TSC unstable due to check_tsc_
[ 54.960222] KVM setup async PF for cpu 2
[ 54.960265] kvm-stealtime: cpu 2, msr 24008d240
[ 54.960342] microcode: CPU2 sig=0x306c1, pf=0x1, revision=0x1
[ 55.012464] Will online and init hotplugged CPU: 2
[ 55.012574] kobject '(null)' (ffff88024009a3d8): tried to add an uninitialized object, something is seriously wrong.
[ 55.012615] CPU: 1 PID: 892 Comm: systemd-udevd Not tainted 3.16.6-2-default #1
[ 55.012649] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.7.5-20140531_
[ 55.012688] ffff8802347cbd98 ffffffff815ca482 ffff88024009a3d8 ffffffff8130e153
[ 55.012729] ffffffff00000020 ffff8802347cbda8 ffff8802347cbd50 00000001334da7a8
[ 55.012766] 000000000000b0a0 0000000001ca6930 0000000000000002 ffff8802334da298
[ 55.012804] Call Trace:
[ 55.012845] [<ffffffff81005
[ 55.012881] [<ffffffff81005
[ 55.012919] [<ffffffff81006
[ 55.012955] [<ffffffff815ca
[ 55.012990] [<ffffffff8130e
[ 55.013027] [<ffffffff812ed
[ 55.013062] [<ffffffff812ed
[ 55.013095] [<ffffffff812eb
[ 55.013129] [<ffffffff81080
[ 55.013162] [<ffffffff8105b
[ 55.013195] [<ffffffff8105c
[ 55.013226] [<ffffffff8105c
[ 55.013260] [<ffffffff815bf
[ 55.013294] [<ffffffff8140c
[ 55.013328] [<ffffffff8140c
[ 55.013360] [<ffffffff81221
[ 55.013394] [<ffffffff811b1
[ 55.013426] [<ffffffff811b2
[ 55.013462] [<ffffffff815d0
[ 55.013517] [<00007f9fbf380
I can reproduce it on a Ubuntu 14.10 host with stock KVM and on Debian host with a custom KVM. The oops sometimes causes the virtual machine to freeze.