| Message ID | 1693333086-392798-2-git-send-email-steven.sistare@oracle.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | fix migration of suspended runstate | expand |
On Tue, Aug 29, 2023 at 11:17:56AM -0700, Steve Sistare wrote: > When a vm in the suspended state is migrated, we must call vm_prepare_start > on the destination, so a later system_wakeup properly resumes the guest, > when main_loop_should_exit callsresume_all_vcpus. However, the runstate > should remain suspended until system_wakeup is called, so allow the caller > to pass the new state to vm_prepare_start, rather than assume the new state > is RUN_STATE_RUNNING. Modify vm state change handlers that check > RUN_STATE_RUNNING to instead use the running parameter. > > No functional change. > > Suggested-by: Peter Xu <peterx@redhat.com> > Signed-off-by: Steve Sistare <steven.sistare@oracle.com> I think all the call sites should be covered indeed, via: qemu_add_vm_change_state_handler_prio qdev_add_vm_change_state_handler virtio_blk_device_realize[1653] qdev_add_vm_change_state_handler(dev, virtio_blk_dma_restart_cb, s); scsi_qdev_realize[289] dev->vmsentry = qdev_add_vm_change_state_handler(DEVICE(dev), vfio_migration_init[796] migration->vm_state = qdev_add_vm_change_state_handler(vbasedev->dev, virtio_init[3189] vdev->vmstate = qdev_add_vm_change_state_handler(DEVICE(vdev), qemu_add_vm_change_state_handler xen_init[106] qemu_add_vm_change_state_handler(xen_change_state_handler, NULL); audio_init[1827] e = qemu_add_vm_change_state_handler (audio_vm_change_state_handler, s); tpm_emulator_inst_init[978] qemu_add_vm_change_state_handler(tpm_emulator_vm_state_change, blk_root_activate[223] blk->vmsh = qemu_add_vm_change_state_handler(blk_vm_state_changed, gdbserver_start[384] qemu_add_vm_change_state_handler(gdb_vm_state_change, NULL); pflash_post_load[1038] pfl->vmstate = qemu_add_vm_change_state_handler(postload_update_cb, qxl_realize_common[2202] qemu_add_vm_change_state_handler(qxl_vm_change_state_handler, qxl); kvmclock_realize[233] qemu_add_vm_change_state_handler(kvmclock_vm_state_change, s); kvm_pit_realizefn[298] qemu_add_vm_change_state_handler(kvm_pit_vm_state_change, s); vapic_post_load[796] qemu_add_vm_change_state_handler(kvmvapic_vm_state_change, s); ide_bus_register_restart_cb[2767] bus->vmstate = qemu_add_vm_change_state_handler(ide_restart_cb, bus); kvm_arm_its_realize[122] qemu_add_vm_change_state_handler(vm_change_state_handler, s); kvm_arm_gicv3_realize[888] qemu_add_vm_change_state_handler(vm_change_state_handler, s); kvmppc_xive_connect[794] xive->change = qemu_add_vm_change_state_handler( via1_post_load[971] v1s->vmstate = qemu_add_vm_change_state_handler( e1000e_core_pci_realize[3379] qemu_add_vm_change_state_handler(e1000e_vm_state_change, core); igb_core_pci_realize[4012] core->vmstate = qemu_add_vm_change_state_handler(igb_vm_state_change, core); spapr_nvram_post_load[235] nvram->vmstate = qemu_add_vm_change_state_handler(postload_update_cb, ppc_booke_timers_init[366] qemu_add_vm_change_state_handler(cpu_state_change_handler, cpu); spapr_machine_init[3070] qemu_add_vm_change_state_handler(cpu_ppc_clock_vm_state_change, kvm_s390_tod_realize[133] qemu_add_vm_change_state_handler(kvm_s390_tod_vm_state_change, td); usb_ehci_realize[2540] s->vmstate = qemu_add_vm_change_state_handler(usb_ehci_vm_state_change, s); usb_host_auto_check[1912] usb_vmstate = qemu_add_vm_change_state_handler(usb_host_vm_state, NULL); usbredir_realize[1466] qemu_add_vm_change_state_handler(usbredir_vm_state_change, dev); virtio_rng_device_realize[226] vrng->vmstate = qemu_add_vm_change_state_handler(virtio_rng_vm_state_change, xen_do_ioreq_register[825] qemu_add_vm_change_state_handler(xen_hvm_change_state_handler, state); net_init_clients[1644] qemu_add_vm_change_state_handler(net_vm_change_state_handler, NULL); memory_global_dirty_log_stop[2978] vmstate_change = qemu_add_vm_change_state_handler( hvf_arch_init[2036] qemu_add_vm_change_state_handler(hvf_vm_state_change, &vtimer); kvm_arch_init_vcpu[567] qemu_add_vm_change_state_handler(kvm_arm_vm_state_change, cs); kvm_arch_init_vcpu[2191] cpu->vmsentry = qemu_add_vm_change_state_handler(cpu_update_state, env); sev_kvm_init[1014] qemu_add_vm_change_state_handler(sev_vm_state_change, sev); whpx_init_vcpu[2248] qemu_add_vm_change_state_handler(whpx_cpu_update_state, cpu->env_ptr); kvm_arch_init_vcpu[70] qemu_add_vm_change_state_handler(kvm_mips_update_state, cs); kvm_arch_init_vcpu[891] qemu_add_vm_change_state_handler(kvm_riscv_vm_state_change, cs); gtk_display_init[2410] qemu_add_vm_change_state_handler(gd_change_runstate, s); qemu_spice_display_init_done[651] qemu_add_vm_change_state_handler(vm_change_state_handler, NULL); qemu_spice_add_interface[868] qemu_add_vm_change_state_handler(vm_change_state_handler, NULL); Looks all correct: Reviewed-by: Peter Xu <peterx@redhat.com>
On 8/30/2023 11:52 AM, Peter Xu wrote: > On Tue, Aug 29, 2023 at 11:17:56AM -0700, Steve Sistare wrote: >> When a vm in the suspended state is migrated, we must call vm_prepare_start >> on the destination, so a later system_wakeup properly resumes the guest, >> when main_loop_should_exit callsresume_all_vcpus. However, the runstate >> should remain suspended until system_wakeup is called, so allow the caller >> to pass the new state to vm_prepare_start, rather than assume the new state >> is RUN_STATE_RUNNING. Modify vm state change handlers that check >> RUN_STATE_RUNNING to instead use the running parameter. >> >> No functional change. >> >> Suggested-by: Peter Xu <peterx@redhat.com> >> Signed-off-by: Steve Sistare <steven.sistare@oracle.com> > > I think all the call sites should be covered indeed, via: Indeed, I laboriously checked every handler, like you, and then I realized that searching for RUN_STATE_RUNNING outside of runstate.c yielded many fewer call sites to check! - Steve > qemu_add_vm_change_state_handler_prio > qdev_add_vm_change_state_handler > virtio_blk_device_realize[1653] qdev_add_vm_change_state_handler(dev, virtio_blk_dma_restart_cb, s); > scsi_qdev_realize[289] dev->vmsentry = qdev_add_vm_change_state_handler(DEVICE(dev), > vfio_migration_init[796] migration->vm_state = qdev_add_vm_change_state_handler(vbasedev->dev, > virtio_init[3189] vdev->vmstate = qdev_add_vm_change_state_handler(DEVICE(vdev), > qemu_add_vm_change_state_handler > xen_init[106] qemu_add_vm_change_state_handler(xen_change_state_handler, NULL); > audio_init[1827] e = qemu_add_vm_change_state_handler (audio_vm_change_state_handler, s); > tpm_emulator_inst_init[978] qemu_add_vm_change_state_handler(tpm_emulator_vm_state_change, > blk_root_activate[223] blk->vmsh = qemu_add_vm_change_state_handler(blk_vm_state_changed, > gdbserver_start[384] qemu_add_vm_change_state_handler(gdb_vm_state_change, NULL); > pflash_post_load[1038] pfl->vmstate = qemu_add_vm_change_state_handler(postload_update_cb, > qxl_realize_common[2202] qemu_add_vm_change_state_handler(qxl_vm_change_state_handler, qxl); > kvmclock_realize[233] qemu_add_vm_change_state_handler(kvmclock_vm_state_change, s); > kvm_pit_realizefn[298] qemu_add_vm_change_state_handler(kvm_pit_vm_state_change, s); > vapic_post_load[796] qemu_add_vm_change_state_handler(kvmvapic_vm_state_change, s); > ide_bus_register_restart_cb[2767] bus->vmstate = qemu_add_vm_change_state_handler(ide_restart_cb, bus); > kvm_arm_its_realize[122] qemu_add_vm_change_state_handler(vm_change_state_handler, s); > kvm_arm_gicv3_realize[888] qemu_add_vm_change_state_handler(vm_change_state_handler, s); > kvmppc_xive_connect[794] xive->change = qemu_add_vm_change_state_handler( > via1_post_load[971] v1s->vmstate = qemu_add_vm_change_state_handler( > e1000e_core_pci_realize[3379] qemu_add_vm_change_state_handler(e1000e_vm_state_change, core); > igb_core_pci_realize[4012] core->vmstate = qemu_add_vm_change_state_handler(igb_vm_state_change, core); > spapr_nvram_post_load[235] nvram->vmstate = qemu_add_vm_change_state_handler(postload_update_cb, > ppc_booke_timers_init[366] qemu_add_vm_change_state_handler(cpu_state_change_handler, cpu); > spapr_machine_init[3070] qemu_add_vm_change_state_handler(cpu_ppc_clock_vm_state_change, > kvm_s390_tod_realize[133] qemu_add_vm_change_state_handler(kvm_s390_tod_vm_state_change, td); > usb_ehci_realize[2540] s->vmstate = qemu_add_vm_change_state_handler(usb_ehci_vm_state_change, s); > usb_host_auto_check[1912] usb_vmstate = qemu_add_vm_change_state_handler(usb_host_vm_state, NULL); > usbredir_realize[1466] qemu_add_vm_change_state_handler(usbredir_vm_state_change, dev); > virtio_rng_device_realize[226] vrng->vmstate = qemu_add_vm_change_state_handler(virtio_rng_vm_state_change, > xen_do_ioreq_register[825] qemu_add_vm_change_state_handler(xen_hvm_change_state_handler, state); > net_init_clients[1644] qemu_add_vm_change_state_handler(net_vm_change_state_handler, NULL); > memory_global_dirty_log_stop[2978] vmstate_change = qemu_add_vm_change_state_handler( > hvf_arch_init[2036] qemu_add_vm_change_state_handler(hvf_vm_state_change, &vtimer); > kvm_arch_init_vcpu[567] qemu_add_vm_change_state_handler(kvm_arm_vm_state_change, cs); > kvm_arch_init_vcpu[2191] cpu->vmsentry = qemu_add_vm_change_state_handler(cpu_update_state, env); > sev_kvm_init[1014] qemu_add_vm_change_state_handler(sev_vm_state_change, sev); > whpx_init_vcpu[2248] qemu_add_vm_change_state_handler(whpx_cpu_update_state, cpu->env_ptr); > kvm_arch_init_vcpu[70] qemu_add_vm_change_state_handler(kvm_mips_update_state, cs); > kvm_arch_init_vcpu[891] qemu_add_vm_change_state_handler(kvm_riscv_vm_state_change, cs); > gtk_display_init[2410] qemu_add_vm_change_state_handler(gd_change_runstate, s); > qemu_spice_display_init_done[651] qemu_add_vm_change_state_handler(vm_change_state_handler, NULL); > qemu_spice_add_interface[868] qemu_add_vm_change_state_handler(vm_change_state_handler, NULL); > > Looks all correct: > > Reviewed-by: Peter Xu <peterx@redhat.com> >
diff --git a/backends/tpm/tpm_emulator.c b/backends/tpm/tpm_emulator.c index 402a2d6..a8e559a 100644 --- a/backends/tpm/tpm_emulator.c +++ b/backends/tpm/tpm_emulator.c @@ -907,7 +907,7 @@ static void tpm_emulator_vm_state_change(void *opaque, bool running, trace_tpm_emulator_vm_state_change(running, state); - if (!running || state != RUN_STATE_RUNNING || !tpm_emu->relock_storage) { + if (!running || !tpm_emu->relock_storage) { return; } diff --git a/gdbstub/softmmu.c b/gdbstub/softmmu.c index f509b72..a43e832 100644 --- a/gdbstub/softmmu.c +++ b/gdbstub/softmmu.c @@ -565,7 +565,7 @@ int gdb_continue_partial(char *newstates) } } - if (vm_prepare_start(step_requested)) { + if (vm_prepare_start(step_requested, RUN_STATE_RUNNING)) { return 0; } diff --git a/hw/usb/hcd-ehci.c b/hw/usb/hcd-ehci.c index c930c60..e436f5c 100644 --- a/hw/usb/hcd-ehci.c +++ b/hw/usb/hcd-ehci.c @@ -2451,7 +2451,7 @@ static void usb_ehci_vm_state_change(void *opaque, bool running, RunState state) * USB-devices which have async handled packages have a packet in the * ep queue to match the completion with. */ - if (state == RUN_STATE_RUNNING) { + if (running) { ehci_advance_async_state(ehci); } diff --git a/hw/usb/redirect.c b/hw/usb/redirect.c index 39fbaaa..1ec5909 100644 --- a/hw/usb/redirect.c +++ b/hw/usb/redirect.c @@ -1403,7 +1403,7 @@ static void usbredir_vm_state_change(void *priv, bool running, RunState state) { USBRedirDevice *dev = priv; - if (state == RUN_STATE_RUNNING && dev->parser != NULL) { + if (running && dev->parser != NULL) { usbredirparser_do_write(dev->parser); /* Flush any pending writes */ } } diff --git a/hw/xen/xen-hvm-common.c b/hw/xen/xen-hvm-common.c index 565dc39..47e6cb1 100644 --- a/hw/xen/xen-hvm-common.c +++ b/hw/xen/xen-hvm-common.c @@ -623,7 +623,7 @@ void xen_hvm_change_state_handler(void *opaque, bool running, xen_set_ioreq_server_state(xen_domid, state->ioservid, - (rstate == RUN_STATE_RUNNING)); + running); } void xen_exit_notifier(Notifier *n, void *data) diff --git a/include/sysemu/runstate.h b/include/sysemu/runstate.h index 7beb29c..7d889ab 100644 --- a/include/sysemu/runstate.h +++ b/include/sysemu/runstate.h @@ -39,8 +39,9 @@ void vm_start(void); * vm_prepare_start: Prepare for starting/resuming the VM * * @step_pending: whether any of the CPUs is about to be single-stepped by gdb + * @state: the vm state to setup */ -int vm_prepare_start(bool step_pending); +int vm_prepare_start(bool step_pending, RunState state); int vm_stop(RunState state); int vm_stop_force_state(RunState state); int vm_shutdown(void); diff --git a/softmmu/cpus.c b/softmmu/cpus.c index fed20ff..0a082d3 100644 --- a/softmmu/cpus.c +++ b/softmmu/cpus.c @@ -681,7 +681,7 @@ int vm_stop(RunState state) * Returns -1 if the vCPUs are not to be restarted (e.g. if they are already * running or in case of an error condition), 0 otherwise. */ -int vm_prepare_start(bool step_pending) +int vm_prepare_start(bool step_pending, RunState state) { RunState requested; @@ -713,14 +713,14 @@ int vm_prepare_start(bool step_pending) qapi_event_send_resume(); cpu_enable_ticks(); - runstate_set(RUN_STATE_RUNNING); - vm_state_notify(1, RUN_STATE_RUNNING); + runstate_set(state); + vm_state_notify(1, state); return 0; } void vm_start(void) { - if (!vm_prepare_start(false)) { + if (!vm_prepare_start(false, RUN_STATE_RUNNING)) { resume_all_vcpus(); } }
When a vm in the suspended state is migrated, we must call vm_prepare_start on the destination, so a later system_wakeup properly resumes the guest, when main_loop_should_exit callsresume_all_vcpus. However, the runstate should remain suspended until system_wakeup is called, so allow the caller to pass the new state to vm_prepare_start, rather than assume the new state is RUN_STATE_RUNNING. Modify vm state change handlers that check RUN_STATE_RUNNING to instead use the running parameter. No functional change. Suggested-by: Peter Xu <peterx@redhat.com> Signed-off-by: Steve Sistare <steven.sistare@oracle.com> --- backends/tpm/tpm_emulator.c | 2 +- gdbstub/softmmu.c | 2 +- hw/usb/hcd-ehci.c | 2 +- hw/usb/redirect.c | 2 +- hw/xen/xen-hvm-common.c | 2 +- include/sysemu/runstate.h | 3 ++- softmmu/cpus.c | 8 ++++---- 7 files changed, 11 insertions(+), 10 deletions(-)