Uploaded image for project: 'RHEL'
  1. RHEL
  2. RHEL-85193

Live migration fails while loading cpu#0 state: Failed to put registers after init: Invalid argument

Linking RHIVOS CVEs to...Migration: Automation ...SWIFT: POC ConversionSync from "Extern...XMLWordPrintable

    • No
    • Moderate
    • 1
    • rhel-virt-hwe-arm-1
    • ssg_virtualization
    • 5
    • False
    • False
    • Hide

      None

      Show
      None
    • None
    • Virt ARM 25-4
    • None
    • None
    • Unspecified
    • Unspecified
    • Unspecified
    • aarch64
    • None

      What were you trying to do that didn't work?

      Live migrate an Aarch64 guest from one host to the other.

      Host-1: hpe-apollo-cn99xx-16.khw.eng.rdu2.dc.redhat.com

      Host-2: hpe-apollo-cn99xx-22.khw.eng.rdu2.dc.redhat.com

      [root@hpe-apollo-cn99xx-16 ~]# uname -r
      5.14.0-503.11.1.el9_5.aarch64

      [root@hpe-apollo-cn99xx-22 ~]# uname -r
      5.14.0-503.11.1.el9_5.aarch64

      1. virsh migrate --verbose --live --auto-converge --parallel --parallel-connections 4 r91vm qemu+ssh://hpe-apollo-cn99xx-22.khw.eng.rdu2.dc.redhat.com/system

       

      What is the impact of this issue to you?

      • Guest live migration fails with an error about invalid vCPU register

      Please provide the package NVR for which the bug is seen:

      [root@hpe-apollo-cn99xx-16 ~]# rpm -q qemu-kvm libvirt kernel
      qemu-kvm-9.0.0-10.el9_5.aarch64
      libvirt-10.5.0-7.el9_5.aarch64
      kernel-5.14.0-503.11.1.el9_5.aarch64
      [root@hpe-apollo-cn99xx-16 ~]

      [root@hpe-apollo-cn99xx-22 ~]# rpm -q qemu-kvm libvirt kernel
      qemu-kvm-9.0.0-10.el9_5.aarch64
      libvirt-10.5.0-7.el9_5.aarch64
      kernel-5.14.0-503.11.1.el9_5.aarch64
      [root@hpe-apollo-cn99xx-22 ~]#

      How reproducible is this bug?:

      • 100%

      Steps to reproduce

      1. Set up NFS shares between above two machines
      2. Created an Aarch64 guest
      3. Live migrate the guest with the above virsh(1) command

      Expected results

      • Migration should complete successfully

      Actual results

      • Migration fails with an error message about invalid vCPU register
      2025-03-27T10:59:26.813434Z qemu-system-aarch64: error while loading state for instance 0x0 of device 'cpu'
      2025-03-27T10:59:26.816150Z qemu-system-aarch64: Failed to put registers after init: Invalid argument
      2025-03-27 10:59:27.220+0000: shutting down, reason=crashed
      
      
      #0  write_list_to_kvmstate (cpu=0xaaaad68747b0, level=3) at ../target/arm/kvm.c:950
      #1  cpu_post_load (opaque=0xaaaad68747b0, version_id=22) at ../target/arm/machine.c:830
      #2  vmstate_load_state (f=0xaaaad7774400, vmsd=0xaaaacea62860 <vmstate_arm_cpu>, opaque=0xaaaad68747b0, version_id=22) at ../migration/vmstate.c:186
      #3  vmstate_load (f=0xaaaad7774400, se=0xaaaad688b6f0) at ../migration/savevm.c:957
      #4  qemu_loadvm_section_start_full (f=0xaaaad7774400, mis=0xaaaad6668410, type=4 '\004') at ../migration/savevm.c:2635
      #5  qemu_loadvm_state_main (f=0xaaaad7774400, mis=0xaaaad6668410) at ../migration/savevm.c:2890
      #6  qemu_loadvm_state (f=0xaaaad7774400) at ../migration/savevm.c:2973
      #7  process_incoming_migration_co (opaque=0x0) at ../migration/migration.c:779
      #8  coroutine_trampoline (i0=-696287840, i1=43690) at ../util/coroutine-ucontext.c:175
      #9  0x0000ffff7fe36780 in ?? () from target:/lib64/libc.so.6
      
      Thread 1 "qemu-system-aar": i=252, 253,
            Thread 5 "CPU 0/KVM": i=252, 253
      
      write_list_to_kvmstate
             case KVM_REG_SIZE_U64:
                  ret = kvm_set_one_reg(cs, regidx, cpu->cpreg_values + i);

        1. dump_regs-host-1
          10 kB
        2. dump_regs-host-2
          10 kB
        3. cn99xx-17-dmidecode.out
          18 kB
        4. cn99xx-22-dmidecode.out
          17 kB

              rhn-engineering-ppandit Prasad Pandit
              rhn-engineering-ppandit Prasad Pandit
              virt-maint virt-maint
              Hana Holoubkova Hana Holoubkova
              Votes:
              0 Vote for this issue
              Watchers:
              14 Start watching this issue

                Created:
                Updated:
                Resolved: