[Qemu-devel] migration: qemu-coroutine-lock.c:141: qemu_co_mutex_unlock: Assertion `mutex->locked == 1' failed

[Alexey Kardashevskiy]

Hi!

I am hitting an racing issue with migration.

I migrate a guest from one machine to another using libvirt:

virsh migrate --live --persistent --undefinesource --copy-storage-all
--verbose --desturi qemu+ssh://legkvm/system --domain chig1

I.e. it copied the full disk which is qcow2, 20G virtual, 10GB of real disk
space.

When migration finished, process_incoming_migration_co() calls
bdrv_invalidate_cache_all() which calls qcow2_invalidate_cache() which does
qcow2_close() and the latter destroys l2_table_cache and
refcount_block_cache. It also calls qcow2_cache_flush(). All good.

However somehow after migration is completed as described above,
qcow2_co_flush_to_os() is called again and I either get crash in
qcow2_cache_flush (as @c==NULL) or I get assert like below as s->lock is
not set, the backtrace is below.

In qcow2_co_flush_to_os(), @bs points to valid data and @s is empty.

The xml I used is at the end of this email. But it does not seem essential
for the problem - it only happens on two selected machines and I cannot
reproduce it on the machines I got locally.

It sounds to me like qcow2_close() is called when there is still a
qcow2_co_flush_to_os() "coroutine" in flight (there is always one, I believe).

Why is this happening and how to fix it properly? Thanks.




Program received signal SIGSEGV, Segmentation fault.
0x0000000010504204 in qcow2_cache_flush (bs=0x10019aab420, c=0x0) at
/home/alexey/p/qemu/block/qcow2-cache.c:174
174         for (i = 0; i < c->size; i++) {
(gdb) bt
#0  0x0000000010504204 in qcow2_cache_flush (bs=0x10019aab420, c=0x0) at
/home/alexey/p/qemu/block/qcow2-cache.c:174
#1  0x00000000104f557c in qcow2_co_flush_to_os (bs=0x10019aab420) at
/home/alexey/p/qemu/block/qcow2.c:2162
#2  0x00000000104c126c in bdrv_co_flush (bs=0x10019aab420) at
/home/alexey/p/qemu/block.c:4971
#3  0x00000000104b2000 in nbd_trip (opaque=0x10019cf75c0) at
/home/alexey/p/qemu/nbd.c:1259
#4  0x00000000104d17d4 in coroutine_trampoline (i0=0x100, i1=0x19cd5c00) at
/home/alexey/p/qemu/coroutine-ucontext.c:118
#5  0x00003fff94ff099c in .__makecontext () from /usr/lib64/libc.so.6
#6  0x0eeabf8ea4edc5a2 in ?? ()
Backtrace stopped: previous frame inner to this frame (corrupt stack?)
(gdb) up
#1  0x00000000104f557c in qcow2_co_flush_to_os (bs=0x10019aab420) at
/home/alexey/p/qemu/block/qcow2.c:2162
2162        ret = qcow2_cache_flush(bs, s->l2_table_cache);
(gdb) p s
$1 = (BDRVQcowState *) 0x10019aaf300
(gdb) p s->l2_table_cache
$2 = (Qcow2Cache *) 0x0
(gdb) p *s
$3 = {
  cluster_bits = 0x0,
  cluster_size = 0x0,
  cluster_sectors = 0x0,
  l2_bits = 0x0,
  l2_size = 0x0,
  l1_size = 0x0,
  l1_vm_state_index = 0x0,
  csize_shift = 0x0,
  csize_mask = 0x0,
  cluster_offset_mask = 0x0,
  l1_table_offset = 0x0,
  l1_table = 0x0,
  l2_table_cache = 0x0,
  refcount_block_cache = 0x0,
  cluster_cache = 0x0,
  cluster_data = 0x0,
  cluster_cache_offset = 0x0,
  cluster_allocs = {
    lh_first = 0x0
  },
  refcount_table = 0x0,
  refcount_table_offset = 0x0,
  refcount_table_size = 0x0,
  free_cluster_index = 0x0,
  free_byte_offset = 0x0,
  lock = {
    locked = 0x1,
    queue = {
      entries = {
        tqh_first = 0x0,
        tqh_last = 0x0
      }
    }
  },
  crypt_method = 0x0,
  crypt_method_header = 0x0,
  aes_encrypt_key = {
    rd_key =       {0x0 <repeats 60 times>},
    rounds = 0x0
  },
  aes_decrypt_key = {
    rd_key =       {0x0 <repeats 60 times>},
    rounds = 0x0
  },
  snapshots_offset = 0x0,
  snapshots_size = 0x0,
  nb_snapshots = 0x0,
  snapshots = 0x0,
  flags = 0x0,
  qcow_version = 0x0,
  use_lazy_refcounts = 0x0,
  refcount_order = 0x0,
  discard_passthrough =     {0x0,
    0x0,
    0x0,
    0x0,
    0x0},
  overlap_check = 0x0,
  incompatible_features = 0x0,
  compatible_features = 0x0,
  autoclear_features = 0x0,
  unknown_header_fields_size = 0x0,
  unknown_header_fields = 0x0,
  unknown_header_ext = {
    lh_first = 0x0
  },
  discards = {
    tqh_first = 0x0,
    tqh_last = 0x0
  },
  cache_discards = 0x0
}
(gdb) p *bs
$4 = {
  total_sectors = 0x2800000,
  read_only = 0x0,
  open_flags = 0x2062,
  encrypted = 0x0,
  valid_key = 0x0,
  sg = 0x0,
  copy_on_read = 0x0,
  drv = 0x1078d440 <bdrv_qcow2>,
  opaque = 0x10019aaf300,
  dev = 0x10019a90b38,
  dev_ops = 0x105b49b0 <virtio_block_ops>,
  dev_opaque = 0x10019a90b38,
  aio_context = 0x10019a7f270,
  aio_notifiers = {
    lh_first = 0x10019cf74f0
  },
  filename =     "/var/lib/libvirt/images/chig1.qcow2",
  backing_file =     "",
  backing_format =     "",
  full_open_options = 0x10019c2a030,
  exact_filename =     "/var/lib/libvirt/images/chig1.qcow2",
  backing_hd = 0x0,
  file = 0x10019aae3b0,
  close_notifiers = {
    notifiers = {
      lh_first = 0x10019cf6460
    }
  },
  before_write_notifiers = {
    notifiers = {
      lh_first = 0x0
    }
  },
  serialising_in_flight = 0x0,
  throttle_state = {
    cfg = {
      buckets =         {{
          avg = 0,
          max = 0,
          level = 0
        },
        {
          avg = 0,
          max = 0,
          level = 0
        },
        {
          avg = 0,
          max = 0,
          level = 0
        },
        {
          avg = 0,
          max = 0,
          level = 0
        },
        {
          avg = 0,
          max = 0,
          level = 0
        },
        {
          avg = 0,
          max = 0,
          level = 0
        }},
      op_size = 0x0
    },
    previous_leak = 0x0,
    timers =       {0x0,
      0x0},
    clock_type = QEMU_CLOCK_REALTIME,
    read_timer_cb = 0x0,
    write_timer_cb = 0x0,
    timer_opaque = 0x0
  },
  throttled_reqs =     {{
      entries = {
        tqh_first = 0x0,
        tqh_last = 0x10019aac188
      }
    },
    {
      entries = {
        tqh_first = 0x0,
        tqh_last = 0x10019aac198
      }
    }},
  io_limits_enabled = 0x0,
  nr_bytes =     {0x0,
    0x0,
    0x0},
  nr_ops =     {0x0,
    0x0,
    0x0},
  total_time_ns =     {0x0,
    0x0,
    0x0},
  wr_highest_sector = 0x27fffff,
  bl = {
    max_discard = 0x0,
    discard_alignment = 0x0,
    max_write_zeroes = 0x0,
    write_zeroes_alignment = 0x80,
    opt_transfer_length = 0x0,
    opt_mem_alignment = 0x1000
  },
  growable = 0x0,
  zero_beyond_eof = 0x1,
  request_alignment = 0x200,
  guest_block_size = 0x200,
  enable_write_cache = 0x1,
  on_read_error = BLOCKDEV_ON_ERROR_REPORT,
  on_write_error = BLOCKDEV_ON_ERROR_ENOSPC,
  iostatus_enabled = 0x1,
  iostatus = BLOCK_DEVICE_IO_STATUS_OK,
  node_name =     "",
  node_list = {
    tqe_next = 0x0,
    tqe_prev = 0x0
  },
  device_name =     "drive-virtio-disk0",
  device_list = {
    tqe_next = 0x0,
    tqe_prev = 0x1078b328 <bdrv_states>
  },
  dirty_bitmaps = {
    lh_first = 0x0
  },
  refcnt = 0x2,
  tracked_requests = {
    lh_first = 0x0
  },
  op_blockers =     {{
      lh_first = 0x0
    } <repeats 14 times>},
  job = 0x0,
  options = 0x10019aa9db0,
  detect_zeroes = BLOCKDEV_DETECT_ZEROES_OPTIONS_OFF,
  backing_blocker = 0x0
}





address@hidden ~]# cat chig1-aik.xml
<domain type='kvm'>
  <name>chig1</name>
  <uuid>bbf91237-3c78-489e-b426-ab593806c78b</uuid>
  <memory unit='KiB'>4194304</memory>
  <currentMemory unit='KiB'>4194304</currentMemory>
  <vcpu placement='static'>1</vcpu>
  <resource>
    <partition>/machine</partition>
  </resource>
  <os>
    <type arch='ppc64' machine='pseries'>hvm</type>
    <boot dev='hd'/>
    <boot dev='network'/>
    <bootmenu enable='yes'/>
  </os>
  <features>
    <acpi/>
    <apic/>
  </features>
  <cpu>
  </cpu>
  <clock offset='utc'/>
  <on_poweroff>destroy</on_poweroff>
  <on_reboot>restart</on_reboot>
  <on_crash>restart</on_crash>
  <devices>
    <emulator>/usr/bin/qemu-system-ppc64.aik</emulator>
    <disk type='file' device='disk'>
      <driver name='qemu' type='qcow2' cache='none'/>
      <source file='/var/lib/libvirt/images/chig1.qcow2'/>
      <target dev='vda' bus='virtio'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x04'
function='0x0'/>
    </disk>
    <controller type='pci' index='0' model='pci-root'/>
    <controller type='usb' index='0'>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x03'
function='0x0'/>
    </controller>
    <interface type='bridge'>
      <mac address='52:54:00:27:70:6f'/>
      <source bridge='brenP1p9s0f0'/>
      <driver name='qemu'/>
      <model type='virtio'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x02'
function='0x0'/>
    </interface>
    <serial type='pty'>
      <target port='0'/>
      <address type='spapr-vio' reg='0x30000000'/>
    </serial>
    <console type='pty'>
      <target type='serial' port='0'/>
      <address type='spapr-vio' reg='0x30000000'/>
    </console>
    <video>
      <model type='vga' vram='9216' heads='1'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x01'
function='0x0'/>
    </video>
    <memballoon model='virtio'>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x05'
function='0x0'/>
    </memballoon>
  </devices>
  <seclabel type='dynamic' model='selinux' relabel='yes'/>
</domain>






-- 
Alexey