| Message ID | 20151005090722.20420.71571.stgit@bahia.huguette.org |
|---|---|
| State | New |
| Headers | show |
On Mon, Oct 05, 2015 at 11:07:23AM +0200, Greg Kurz wrote: > If the user tries to hot unplug a virtio-9p device, it seems to succeed but > in fact: > - virtio-9p coroutines thread pool and async queue are leaked > - QEMU crashes in virtio_vmstate_change() if the user tries to live migrate > > This patch brings hot unplug support to virtio-9p-device. It fixes both > above issues. > > Signed-off-by: Greg Kurz <gkurz@linux.vnet.ibm.com> > --- > hw/9pfs/virtio-9p-device.c | 12 ++++++++++++ > 1 file changed, 12 insertions(+) What happens to in-flight I/O requests? We cannot assume that the guest driver quiesces the device.
On Tue, 6 Oct 2015 09:09:14 +0100 Stefan Hajnoczi <stefanha@gmail.com> wrote: > On Mon, Oct 05, 2015 at 11:07:23AM +0200, Greg Kurz wrote: > > If the user tries to hot unplug a virtio-9p device, it seems to succeed but > > in fact: > > - virtio-9p coroutines thread pool and async queue are leaked > > - QEMU crashes in virtio_vmstate_change() if the user tries to live migrate > > > > This patch brings hot unplug support to virtio-9p-device. It fixes both > > above issues. > > > > Signed-off-by: Greg Kurz <gkurz@linux.vnet.ibm.com> > > --- > > hw/9pfs/virtio-9p-device.c | 12 ++++++++++++ > > 1 file changed, 12 insertions(+) > > What happens to in-flight I/O requests? We cannot assume that the guest > driver quiesces the device. > We can assume that the guest has unmounted the 9p share otherwise migration is blocked... is it possible we still have in-flight I/O requests in this scenario ?
Stefan Hajnoczi <stefanha@gmail.com> writes: > On Mon, Oct 05, 2015 at 11:07:23AM +0200, Greg Kurz wrote: >> If the user tries to hot unplug a virtio-9p device, it seems to succeed but >> in fact: >> - virtio-9p coroutines thread pool and async queue are leaked >> - QEMU crashes in virtio_vmstate_change() if the user tries to live migrate >> >> This patch brings hot unplug support to virtio-9p-device. It fixes both >> above issues. >> >> Signed-off-by: Greg Kurz <gkurz@linux.vnet.ibm.com> >> --- >> hw/9pfs/virtio-9p-device.c | 12 ++++++++++++ >> 1 file changed, 12 insertions(+) > > What happens to in-flight I/O requests? We cannot assume that the guest > driver quiesces the device. We enable migration blocker when we have an active mount. So if we get here, that should indicate no active 9p mounts. -aneesh
On Wed, 07 Oct 2015 14:20:28 +0530 "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com> wrote: > Stefan Hajnoczi <stefanha@gmail.com> writes: > > > On Mon, Oct 05, 2015 at 11:07:23AM +0200, Greg Kurz wrote: > >> If the user tries to hot unplug a virtio-9p device, it seems to succeed but > >> in fact: > >> - virtio-9p coroutines thread pool and async queue are leaked > >> - QEMU crashes in virtio_vmstate_change() if the user tries to live migrate > >> > >> This patch brings hot unplug support to virtio-9p-device. It fixes both > >> above issues. > >> > >> Signed-off-by: Greg Kurz <gkurz@linux.vnet.ibm.com> > >> --- > >> hw/9pfs/virtio-9p-device.c | 12 ++++++++++++ > >> 1 file changed, 12 insertions(+) > > > > What happens to in-flight I/O requests? We cannot assume that the guest > > driver quiesces the device. > > We enable migration blocker when we have an active mount. So if we get > here, that should indicate no active 9p mounts. > > -aneesh Oops.. Stefan is talking about hot-unplug versus in-flight requests... not about migration. And there is no such thing as a hot-unplug blocker... -- Greg
On Fri, Oct 09, 2015 at 09:23:49AM +0200, Greg Kurz wrote: > On Wed, 07 Oct 2015 14:20:28 +0530 > "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com> wrote: > > > Stefan Hajnoczi <stefanha@gmail.com> writes: > > > > > On Mon, Oct 05, 2015 at 11:07:23AM +0200, Greg Kurz wrote: > > >> If the user tries to hot unplug a virtio-9p device, it seems to succeed but > > >> in fact: > > >> - virtio-9p coroutines thread pool and async queue are leaked > > >> - QEMU crashes in virtio_vmstate_change() if the user tries to live migrate > > >> > > >> This patch brings hot unplug support to virtio-9p-device. It fixes both > > >> above issues. > > >> > > >> Signed-off-by: Greg Kurz <gkurz@linux.vnet.ibm.com> > > >> --- > > >> hw/9pfs/virtio-9p-device.c | 12 ++++++++++++ > > >> 1 file changed, 12 insertions(+) > > > > > > What happens to in-flight I/O requests? We cannot assume that the guest > > > driver quiesces the device. > > > > We enable migration blocker when we have an active mount. So if we get > > here, that should indicate no active 9p mounts. > > > > -aneesh > > Oops.. Stefan is talking about hot-unplug versus in-flight requests... not > about migration. And there is no such thing as a hot-unplug blocker... If unplug request fails, that should be enough. > -- > Greg
On Fri, 9 Oct 2015 10:36:56 +0300 "Michael S. Tsirkin" <mst@redhat.com> wrote: > On Fri, Oct 09, 2015 at 09:23:49AM +0200, Greg Kurz wrote: > > On Wed, 07 Oct 2015 14:20:28 +0530 > > "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com> wrote: > > > > > Stefan Hajnoczi <stefanha@gmail.com> writes: > > > > > > > On Mon, Oct 05, 2015 at 11:07:23AM +0200, Greg Kurz wrote: > > > >> If the user tries to hot unplug a virtio-9p device, it seems to succeed but > > > >> in fact: > > > >> - virtio-9p coroutines thread pool and async queue are leaked > > > >> - QEMU crashes in virtio_vmstate_change() if the user tries to live migrate > > > >> > > > >> This patch brings hot unplug support to virtio-9p-device. It fixes both > > > >> above issues. > > > >> > > > >> Signed-off-by: Greg Kurz <gkurz@linux.vnet.ibm.com> > > > >> --- > > > >> hw/9pfs/virtio-9p-device.c | 12 ++++++++++++ > > > >> 1 file changed, 12 insertions(+) > > > > > > > > What happens to in-flight I/O requests? We cannot assume that the guest > > > > driver quiesces the device. > > > > > > We enable migration blocker when we have an active mount. So if we get > > > here, that should indicate no active 9p mounts. > > > > > > -aneesh > > > > Oops.. Stefan is talking about hot-unplug versus in-flight requests... not > > about migration. And there is no such thing as a hot-unplug blocker... > > If unplug request fails, that should be enough. > Like setting @hotpluggable to false for the virtio-9p-pci class ? I see no other way for the unplug request to fail... but I will happily accept all suggestions :) -- Greg
diff --git a/hw/9pfs/virtio-9p-device.c b/hw/9pfs/virtio-9p-device.c index 93a407c45926..ed133c40493a 100644 --- a/hw/9pfs/virtio-9p-device.c +++ b/hw/9pfs/virtio-9p-device.c @@ -138,6 +138,17 @@ out: v9fs_path_free(&path); } +static void virtio_9p_device_unrealize(DeviceState *dev, Error **errp) +{ + VirtIODevice *vdev = VIRTIO_DEVICE(dev); + V9fsState *s = VIRTIO_9P(dev); + + v9fs_release_worker_threads(); + g_free(s->ctx.fs_root); + g_free(s->tag); + virtio_cleanup(vdev); +} + /* virtio-9p device */ static Property virtio_9p_properties[] = { @@ -154,6 +165,7 @@ static void virtio_9p_class_init(ObjectClass *klass, void *data) dc->props = virtio_9p_properties; set_bit(DEVICE_CATEGORY_STORAGE, dc->categories); vdc->realize = virtio_9p_device_realize; + vdc->unrealize = virtio_9p_device_unrealize; vdc->get_features = virtio_9p_get_features; vdc->get_config = virtio_9p_get_config; }
If the user tries to hot unplug a virtio-9p device, it seems to succeed but in fact: - virtio-9p coroutines thread pool and async queue are leaked - QEMU crashes in virtio_vmstate_change() if the user tries to live migrate This patch brings hot unplug support to virtio-9p-device. It fixes both above issues. Signed-off-by: Greg Kurz <gkurz@linux.vnet.ibm.com> --- hw/9pfs/virtio-9p-device.c | 12 ++++++++++++ 1 file changed, 12 insertions(+)