[libvirt] [PATCH v2] qemu: migration: Don't crash on access to 'current' job

When a VM is destroyed while being migrated (waiting in qemuMigrationSrcWaitForCompletion) the private object cleanup code frees the 'current' job info. Since the migration code attempts to setup various aspects of the current job even on failure this results into a crash. Job data is cleared in qemuDomainObjPrivateDataClear since commit 888aa4b6b9db Fix this by skipping all of the code which requires the qemu process to be alive if the VM is not active any more. Signed-off-by: Peter Krempa <pkrempa@redhat.com> --- src/qemu/qemu_migration.c | 33 +++++++++++++++++---------------- 1 file changed, 17 insertions(+), 16 deletions(-) diff --git a/src/qemu/qemu_migration.c b/src/qemu/qemu_migration.c index b913e99928..1a857329ec 100644 --- a/src/qemu/qemu_migration.c +++ b/src/qemu/qemu_migration.c @@ -3561,27 +3561,28 @@ qemuMigrationSrcRun(virQEMUDriverPtr driver, error: orig_err = virSaveLastError(); - if (cancel && - priv->job.current->status != QEMU_DOMAIN_JOB_STATUS_QEMU_COMPLETED && - virDomainObjIsActive(vm) && - qemuDomainObjEnterMonitorAsync(driver, vm, - QEMU_ASYNC_JOB_MIGRATION_OUT) == 0) { - qemuMonitorMigrateCancel(priv->mon); - ignore_value(qemuDomainObjExitMonitor(driver, vm)); - } + if (virDomainObjIsActive(vm)) { + if (cancel && + priv->job.current->status != QEMU_DOMAIN_JOB_STATUS_QEMU_COMPLETED && + qemuDomainObjEnterMonitorAsync(driver, vm, + QEMU_ASYNC_JOB_MIGRATION_OUT) == 0) { + qemuMonitorMigrateCancel(priv->mon); + ignore_value(qemuDomainObjExitMonitor(driver, vm)); + } - /* cancel any outstanding NBD jobs */ - if (mig && mig->nbd) - qemuMigrationSrcCancelDriveMirror(driver, vm, false, - QEMU_ASYNC_JOB_MIGRATION_OUT, - dconn); + /* cancel any outstanding NBD jobs */ + if (mig && mig->nbd) + qemuMigrationSrcCancelDriveMirror(driver, vm, false, + QEMU_ASYNC_JOB_MIGRATION_OUT, + dconn); + + if (priv->job.current->status != QEMU_DOMAIN_JOB_STATUS_CANCELED) + priv->job.current->status = QEMU_DOMAIN_JOB_STATUS_FAILED; + } if (iothread) qemuMigrationSrcStopTunnel(iothread, true); - if (priv->job.current->status != QEMU_DOMAIN_JOB_STATUS_CANCELED) - priv->job.current->status = QEMU_DOMAIN_JOB_STATUS_FAILED; - goto cleanup; exit_monitor: -- 2.16.2

On Mon, Apr 30, 2018 at 13:36:39 +0200, Peter Krempa wrote:
When a VM is destroyed while being migrated (waiting in qemuMigrationSrcWaitForCompletion) the private object cleanup code frees the 'current' job info. Since the migration code attempts to setup various aspects of the current job even on failure this results into a crash.
Job data is cleared in qemuDomainObjPrivateDataClear since commit 888aa4b6b9db
Fix this by skipping all of the code which requires the qemu process to be alive if the VM is not active any more.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
participants (2)
-
Jiri Denemark
-
Peter Krempa