On Mon, Feb 04, 2019 at 13:36:24 +0100, Marc Hartmayer wrote: > For normal starts (no incoming migration) the refresh of the QEMU > state must be done before the VCPUs getting started since otherwise > there might be a race condition between a possible shutdown of the > guest OS and the QEMU monitor queries. > > This fixes "qemu: migration: Refresh device information after > transferring state" (93db7eea1b864). > > Signed-off-by: Marc Hartmayer <mhartmay(a)linux.ibm.com&gt; > --- > src/qemu/qemu_process.c | 20 +++++++++++--------- > 1 file changed, 11 insertions(+), 9 deletions(-) > > diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c > index dace5aaca102..2a3763f40d49 100644 > --- a/src/qemu/qemu_process.c > +++ b/src/qemu/qemu_process.c > @@ -6929,10 +6929,17 @@ qemuProcessStart(virConnectPtr conn, > } > relabel = true; > > - if (incoming && > - incoming->deferredURI && > - qemuMigrationDstRun(driver, vm, incoming->deferredURI, asyncJob) < 0) > - goto stop; > + if (incoming) { > + if (incoming->deferredURI && > + qemuMigrationDstRun(driver, vm, incoming->deferredURI, asyncJob) < 0) > + goto stop; > + } else { This logic does not seem right ...

> + /* Refresh state of devices from QEMU. During migration this > + * needs to happen after the state information is fully > + * transferred. */ as this comment clearly states that this should happen after migration.

Here it would happen only when migration is not done.

> + if (qemuProcessRefreshState(driver, vm, asyncJob) < 0) > + goto stop; > + } > > if (qemuProcessFinishStartup(driver, vm, asyncJob, > !(flags & VIR_QEMU_PROCESS_START_PAUSED), > @@ -6945,11 +6952,6 @@ qemuProcessStart(virConnectPtr conn, > /* Keep watching qemu log for errors during incoming migration, otherwise > * unset reporting errors from qemu log. */ > qemuMonitorSetDomainLog(priv->mon, NULL, NULL, NULL); > - > - /* Refresh state of devices from qemu. During migration this needs to > - * happen after the state information is fully transferred. */ > - if (qemuProcessRefreshState(driver, vm, asyncJob) < 0) > - goto stop; > } > > ret = 0; > -- > 2.17.0 >

On Mon, Feb 04, 2019 at 02:37 PM +0100, Marc Hartmayer <mhartmay(a)linux.ibm.com&gt; wrote: > On Mon, Feb 04, 2019 at 01:41 PM +0100, Peter Krempa <pkrempa(a)redhat.com&gt; wrote: >> On Mon, Feb 04, 2019 at 13:36:24 +0100, Marc Hartmayer wrote: >>> For normal starts (no incoming migration) the refresh of the QEMU >>> state must be done before the VCPUs getting started since otherwise >>> there might be a race condition between a possible shutdown of the >>> guest OS and the QEMU monitor queries. >>> >>> This fixes "qemu: migration: Refresh device information after >>> transferring state" (93db7eea1b864). >>> >>> Signed-off-by: Marc Hartmayer <mhartmay(a)linux.ibm.com&gt; >>> --- >>> src/qemu/qemu_process.c | 20 +++++++++++--------- >>> 1 file changed, 11 insertions(+), 9 deletions(-) >>> >>> diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c >>> index dace5aaca102..2a3763f40d49 100644 >>> --- a/src/qemu/qemu_process.c >>> +++ b/src/qemu/qemu_process.c >>> @@ -6929,10 +6929,17 @@ qemuProcessStart(virConnectPtr conn, >>> } >>> relabel = true; >>> >>> - if (incoming && >>> - incoming->deferredURI && >>> - qemuMigrationDstRun(driver, vm, incoming->deferredURI, asyncJob) < 0) >>> - goto stop; >>> + if (incoming) { >>> + if (incoming->deferredURI && >>> + qemuMigrationDstRun(driver, vm, incoming->deferredURI, asyncJob) < 0) >>> + goto stop; >>> + } else { >> >> This logic does not seem right ... > > I’m not familiar with the usage of this function for migration… so > there's a good chance I'm missing something. > >> >>> + /* Refresh state of devices from QEMU. During migration this >>> + * needs to happen after the state information is fully >>> + * transferred. */ >> >> as this comment clearly states that this should happen after >> migration. > > Is qemuProcessFinishStartup ^^^^^^^^^^^^^^^^^^^^^^^^ qemuProcessRefreshState > not called explicitly in qemuMigrationFinish > for migration? >

>> >> Here it would happen only when migration is not done. > > Without this patch qemuProcessRefreshState is called after > qemuProcessFinishStartup if (!incoming). Now it’s called directly before > qemuProcessFinishStartup if (!incoming). Why should this be wrong now? > What happens in qemuProcessFinishStartup? > > Before your change in 93db7eea1b86408e the function call > 'qemuProcessRefreshState' was done in qemuProcessFinishStartup (without > any condition).

On Mon, Feb 04, 2019 at 01:36:24PM +0100, Marc Hartmayer wrote: > For normal starts (no incoming migration) the refresh of the QEMU > state must be done before the VCPUs getting started since otherwise > there might be a race condition between a possible shutdown of the > guest OS and the QEMU monitor queries. > > This fixes "qemu: migration: Refresh device information after > transferring state" (93db7eea1b864). Hmm, this is the second bug that this patch has been responsible for now It has also resulted in a deadlock in libguestfs during startup when there is a chardev in blocking connect mode in QEMU. If the guest writes data to the chardev before it is accepted by the server libvirt can deadlock in startup as QEMU is blocked and won't respond to QMP. The core problem is that the startup code

must not rely on being able to use QMP /after/ the vCPUs have been started untill the virDomainCreate API has completed https://bugzilla.redhat.com/show_bug.cgi?id=1661940 IMHO we need to just revert that original broken commit entirely and find a different way to fix the problem it was attacking.

> > Signed-off-by: Marc Hartmayer <mhartmay(a)linux.ibm.com&gt; > --- > src/qemu/qemu_process.c | 20 +++++++++++--------- > 1 file changed, 11 insertions(+), 9 deletions(-) > > diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c > index dace5aaca102..2a3763f40d49 100644 > --- a/src/qemu/qemu_process.c > +++ b/src/qemu/qemu_process.c > @@ -6929,10 +6929,17 @@ qemuProcessStart(virConnectPtr conn, > } > relabel = true; > > - if (incoming && > - incoming->deferredURI && > - qemuMigrationDstRun(driver, vm, incoming->deferredURI, asyncJob) < 0) > - goto stop; > + if (incoming) { > + if (incoming->deferredURI && > + qemuMigrationDstRun(driver, vm, incoming->deferredURI, asyncJob) < 0) > + goto stop; > + } else { > + /* Refresh state of devices from QEMU. During migration this > + * needs to happen after the state information is fully > + * transferred. */