#122 soft lockup while running SMP linux guest
I catched a error while runing a 32bit smp linux on pae host.
The linux guest hanged while running fsck on hda1.
The linux kernel is 2.6.16, vcpu# is 4.
Here is the command:
qemu . -m 512 -smp 4 -no-acpi -net nic,macaddr=00:16:3e:4c:c7:1c,model=rtl8139 -net tap,script=/etc/kvm/qemu-ifup -hda /share/xvs/var/tmp-img_gtp10_1191033976_1
Output from dmesg of host:
BUG: soft lockup detected on CPU#3!
[<c044a0b7>]</c044a0b7> softlockup_tick+0x98/0xa6
[<c042cc98>]</c042cc98> update_process_times+0x39/0x5c
[<c04176ec>]</c04176ec> smp_apic_timer_interrupt+0x5c/0x64
[<c04049bf>]</c04049bf> apic_timer_interrupt+0x1f/0x24
[<f8abe2dd>]</f8abe2dd> kvm_flush_remote_tlbs+0xce/0xdb [kvm]
[<f8ac3079>]</f8ac3079> kvm_mmu_pte_write+0x1e0/0x356 [kvm]
[<f8abe825>]</f8abe825> emulator_write_emulated_onepage+0xba/0x164 [kvm]
[<f8ac5e66>]</f8ac5e66> x86_emulate_insn+0x20ac/0x312e [kvm]
[<f8ac3b9a>]</f8ac3b9a> x86_decode_insn+0x624/0x844 [kvm]
[<f8abe63e>]</f8abe63e> emulate_instruction+0x12b/0x258 [kvm]
[<f8a17314>]</f8a17314> handle_exception+0x16a/0x246 [kvm_intel]
[<f8a170db>]</f8a170db> kvm_handle_exit+0x70/0x8a [kvm_intel]
[<f8abec7c>]</f8abec7c> kvm_vcpu_ioctl_run+0x230/0x335 [kvm]
[<f8abee91>]</f8abee91> kvm_vcpu_ioctl+0x0/0xc67 [kvm]
[<f8abef94>]</f8abef94> kvm_vcpu_ioctl+0x103/0xc67 [kvm]
[<c04049bf>]</c04049bf> apic_timer_interrupt+0x1f/0x24
[<c047abd9>]</c047abd9> do_select+0x3ae/0x3cb
[<c047b11f>]</c047b11f> __pollwait+0x0/0xb2
[<c0547843>]</c0547843> loopback_xmit+0x75/0x7a
[<c0427fd6>]</c0427fd6> local_bh_enable+0x74/0x7f
[<c05a5595>]</c05a5595> dev_queue_xmit+0x22e/0x24d
[<c05c2673>]</c05c2673> ip_output+0x1c0/0x1f9
[<c05c9b9e>]</c05c9b9e> tcp_cong_avoid+0x15/0x25
[<c05cbbc8>]</c05cbbc8> tcp_ack+0x1597/0x16c9
[<c05ceb20>]</c05ceb20> tcp_rcv_established+0x733/0x7c3
[<c05d3a67>]</c05d3a67> tcp_v4_do_rcv+0x25/0x2bc
[<c059ca38>]</c059ca38> release_sock+0x59/0x91
[<c04db744>]</c04db744> __next_cpu+0x12/0x21
[<c041c97f>]</c041c97f> find_busiest_group+0x177/0x462
[<c041d3cd>]</c041d3cd> enqueue_task+0x29/0x39
[<c041d3f9>]</c041d3f9> __activate_task+0x1c/0x29
[<c041dc43>]</c041dc43> try_to_wake_up+0x371/0x37b
[<c047ade5>]</c047ade5> core_sys_select+0x1ef/0x2ca
[<c041c41c>]</c041c41c> __wake_up_common+0x2f/0x53
[<c041d20d>]</c041d20d> __wake_up+0x2a/0x3d
[<c0437cef>]</c0437cef> wake_futex+0x3a/0x44
[<c042d849>]</c042d849> __dequeue_signal+0x151/0x15c
[<c042e942>]</c042e942> dequeue_signal+0x2d/0x9c
[<c042f4d2>]</c042f4d2> sys_rt_sigtimedwait+0xc5/0x2c2
[<f8abee91>]</f8abee91> kvm_vcpu_ioctl+0x0/0xc67 [kvm]
[<c0479cc8>]</c0479cc8> do_ioctl+0x1c/0x5d
[<c0479f53>]</c0479f53> vfs_ioctl+0x24a/0x25c
[<c0479fad>]</c0479fad> sys_ioctl+0x48/0x5f
[<c0403eff>]</c0403eff> syscall_call+0x7/0xb
Logged In: YES
user_id=1283543
Originator: YES
Tested it with "-no-kvm-irqchip", it has the same issue.
Logged In: YES
user_id=1283543
Originator: YES
I cannot produce this issue in latest commits.
It should be fixed already.