[libvirt] [PATCH] qemu: hotplug: ensure address generation for vfio-ccw
When attaching a mediated host device of model vfio-ccw without specifying a guest-address, none is generated by libvirt. Let's fix this and make sure to generate a device address during live-hotplug. Reviewed-by: Boris Fiuczynski <fiuczy@linux.ibm.com> Signed-off-by: Bjoern Walk <bwalk@linux.ibm.com> --- src/qemu/qemu_hotplug.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index fd4bafef..7b775159 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -2790,6 +2790,8 @@ qemuDomainAttachMediatedDevice(virQEMUDriverPtr driver, { int ret = -1; g_autofree char *devstr = NULL; + char *devName = NULL; + bool releaseaddr = false; bool added = false; bool teardowncgroup = false; bool teardownlabel = false; @@ -2805,6 +2807,10 @@ qemuDomainAttachMediatedDevice(virQEMUDriverPtr driver, return -1; break; case VIR_MDEV_MODEL_TYPE_VFIO_CCW: + devName = hostdev->source.subsys.u.mdev.uuidstr; + if (qemuDomainEnsureVirtioAddress(&releaseaddr, vm, &dev, devName) < 0) + return -1; + break; case VIR_MDEV_MODEL_TYPE_LAST: break; } -- 2.21.0
Bjoern Walk <bwalk@linux.ibm.com> [2019-10-28, 09:06AM +0100]:
When attaching a mediated host device of model vfio-ccw without specifying a guest-address, none is generated by libvirt. Let's fix this and make sure to generate a device address during live-hotplug.
I know we are already in freeze, but it still would be good to have this fix for S390 in 5.9 if possible. -- IBM Systems Linux on Z & Virtualization Development -------------------------------------------------- IBM Deutschland Research & Development GmbH Schönaicher Str. 220, 71032 Böblingen Phone: +49 7031 16 1819 -------------------------------------------------- Vorsitzende des Aufsichtsrats: Matthias Hartmann Geschäftsführung: Dirk Wittkopp Sitz der Gesellschaft: Böblingen Registergericht: Amtsgericht Stuttgart, HRB 243294
ping... -- IBM Systems Linux on Z & Virtualization Development -------------------------------------------------- IBM Deutschland Research & Development GmbH Schönaicher Str. 220, 71032 Böblingen Phone: +49 7031 16 1819 -------------------------------------------------- Vorsitzende des Aufsichtsrats: Matthias Hartmann Geschäftsführung: Dirk Wittkopp Sitz der Gesellschaft: Böblingen Registergericht: Amtsgericht Stuttgart, HRB 243294
On 10/28/19 5:06 AM, Bjoern Walk wrote:
When attaching a mediated host device of model vfio-ccw without specifying a guest-address, none is generated by libvirt. Let's fix this and make sure to generate a device address during live-hotplug.
Reviewed-by: Boris Fiuczynski <fiuczy@linux.ibm.com> Signed-off-by: Bjoern Walk <bwalk@linux.ibm.com> ---
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
On Mon, Oct 28, 2019 at 09:06:13AM +0100, Bjoern Walk wrote:
When attaching a mediated host device of model vfio-ccw without specifying a guest-address, none is generated by libvirt. Let's fix this and make sure to generate a device address during live-hotplug.
Reviewed-by: Boris Fiuczynski <fiuczy@linux.ibm.com> Signed-off-by: Bjoern Walk <bwalk@linux.ibm.com> --- src/qemu/qemu_hotplug.c | 6 ++++++ 1 file changed, 6 insertions(+)
diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index fd4bafef..7b775159 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -2790,6 +2790,8 @@ qemuDomainAttachMediatedDevice(virQEMUDriverPtr driver, { int ret = -1; g_autofree char *devstr = NULL; + char *devName = NULL; + bool releaseaddr = false; bool added = false; bool teardowncgroup = false; bool teardownlabel = false; @@ -2805,6 +2807,10 @@ qemuDomainAttachMediatedDevice(virQEMUDriverPtr driver, return -1; break; case VIR_MDEV_MODEL_TYPE_VFIO_CCW: + devName = hostdev->source.subsys.u.mdev.uuidstr; + if (qemuDomainEnsureVirtioAddress(&releaseaddr, vm, &dev, devName) < 0)
vfio-ccw is not really a virtio device, is it? Jano
+ return -1; + break; case VIR_MDEV_MODEL_TYPE_LAST: break; } -- 2.21.0
-- libvir-list mailing list libvir-list@redhat.com https://www.redhat.com/mailman/listinfo/libvir-list
Ján Tomko <jtomko@redhat.com> [2019-11-12, 10:47AM +0100]:
diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index fd4bafef..7b775159 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -2790,6 +2790,8 @@ qemuDomainAttachMediatedDevice(virQEMUDriverPtr driver, { int ret = -1; g_autofree char *devstr = NULL; + char *devName = NULL; + bool releaseaddr = false; bool added = false; bool teardowncgroup = false; bool teardownlabel = false; @@ -2805,6 +2807,10 @@ qemuDomainAttachMediatedDevice(virQEMUDriverPtr driver, return -1; break; case VIR_MDEV_MODEL_TYPE_VFIO_CCW: + devName = hostdev->source.subsys.u.mdev.uuidstr; + if (qemuDomainEnsureVirtioAddress(&releaseaddr, vm, &dev, devName) < 0)
vfio-ccw is not really a virtio device, is it?
The naming of that function is rather odd and misleading. It sets the address type of the device if not set and generates the correct address, exactly what we need here. Somewhere on my TODO-list I have a point to go over the address generation for CCW and clean it up. For now, this must suffice. -- IBM Systems Linux on Z & Virtualization Development -------------------------------------------------- IBM Deutschland Research & Development GmbH Schönaicher Str. 220, 71032 Böblingen Phone: +49 7031 16 1819 -------------------------------------------------- Vorsitzende des Aufsichtsrats: Matthias Hartmann Geschäftsführung: Dirk Wittkopp Sitz der Gesellschaft: Böblingen Registergericht: Amtsgericht Stuttgart, HRB 243294
On Tue, Nov 12, 2019 at 12:06:46PM +0100, Bjoern Walk wrote:
Ján Tomko <jtomko@redhat.com> [2019-11-12, 10:47AM +0100]:
diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index fd4bafef..7b775159 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -2790,6 +2790,8 @@ qemuDomainAttachMediatedDevice(virQEMUDriverPtr driver, { int ret = -1; g_autofree char *devstr = NULL; + char *devName = NULL; + bool releaseaddr = false;
Having these scoped for the whole function is excessive. Especially the misleading releaseaddr, which is never used.
bool added = false; bool teardowncgroup = false; bool teardownlabel = false; @@ -2805,6 +2807,10 @@ qemuDomainAttachMediatedDevice(virQEMUDriverPtr driver, return -1; break; case VIR_MDEV_MODEL_TYPE_VFIO_CCW: + devName = hostdev->source.subsys.u.mdev.uuidstr; + if (qemuDomainEnsureVirtioAddress(&releaseaddr, vm, &dev, devName) < 0)
vfio-ccw is not really a virtio device, is it?
The naming of that function is rather odd and misleading. It sets the address type of the device if not set and generates the correct address, exactly what we need here.
Oh right, not even VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_S390 seems to be virtio-related.
Somewhere on my TODO-list I have a point to go over the address generation for CCW and clean it up. For now, this must suffice.
It does seem to be more complicated than it needs to be. With the declarations moved into the switch case: Reviewed-by: Ján Tomko <jtomko@redhat.com> and pushed. Jano
Ján Tomko <jtomko@redhat.com> [2019-11-12, 04:45PM +0100]:
On Tue, Nov 12, 2019 at 12:06:46PM +0100, Bjoern Walk wrote:
Somewhere on my TODO-list I have a point to go over the address generation for CCW and clean it up. For now, this must suffice.
It does seem to be more complicated than it needs to be.
Yes, indeed.
With the declarations moved into the switch case: Reviewed-by: Ján Tomko <jtomko@redhat.com>
and pushed.
Thank you! -- IBM Systems Linux on Z & Virtualization Development -------------------------------------------------- IBM Deutschland Research & Development GmbH Schönaicher Str. 220, 71032 Böblingen Phone: +49 7031 16 1819 -------------------------------------------------- Vorsitzende des Aufsichtsrats: Matthias Hartmann Geschäftsführung: Dirk Wittkopp Sitz der Gesellschaft: Böblingen Registergericht: Amtsgericht Stuttgart, HRB 243294
participants (3)
-
Bjoern Walk -
Daniel Henrique Barboza -
Ján Tomko