Message ID | 1553528210-30562-1-git-send-email-frank.heimes@canonical.com |
---|---|
Headers | show |
Series | Fixes for 1818854 | expand |
On Mon, Mar 25, 2019 at 04:36:49PM +0100, frank.heimes@canonical.com wrote: > Buglink: https://bugs.launchpad.net/bugs/1818854 > > SRU Justification: > > [Impact] > > * The vfio-ap driver will create a matrix device in sysfs without a subsystem link. > > * This causes failures in libudev that might also lead to libvirt errors. > > * A fix is upstream in master branch for kernel 5.1 > > [Fix] > > * 36360658eb5a6cf04bb9f2704d1e4ce54037ec99 3636065 "s390: vfio_ap: link the vfio_ap devices to the vfio_ap bus subsystem" > > [Test Case] > > * An s390x machine with at least one CryptoExpress card, > where at least one AP and one 'Domain' is assigned to a particular LPAR. > > * Now running virsh nodedev-list before (and later after) construction the matrix device should expose the issue. > > * For details about how to setup a vfio-ap matrix device see: http://kvmonz.blogspot.com/2018/12/qemu-v31-released.html > (see 2nd bullet: virtio-ap) > > [Regression Potential] > > * The regression potential can be considered as low since it only affects the s390x platform > > * and there it only affects the usage of AP (CryptoExpress) cards and it's driver/module > > * and again only affects the recently introduced virtual IO function of AP (vfio-ap). > > [Other Info] > > * It was already briefly discussed here: > https://www.redhat.com/archives/libvir-list/2019-February/msg00837.html > as well as reviewed and pushed. > > * Commit 3636065 from v5.1-rc1 > > * This affects ccw cards and their vf only, NOT vf of PCI/PCIe cards! > > * * For details on virtio-ap/vfio-ap see bullet #2 here: http://kvmonz.blogspot.com/2018/12/qemu-v31-released.html This patch is in upstream stable updates, so therefore it also meets the SRU criteria for Ubuntu. For B/C: Acked-by: Seth Forshee <seth.forshee@canonical.com> Disco already got this from upstream stable, so not needed there.
On 2019-03-25 16:36:49 , frank.heimes@canonical.com wrote: > Buglink: https://bugs.launchpad.net/bugs/1818854 > > SRU Justification: > > [Impact] > > * The vfio-ap driver will create a matrix device in sysfs without a subsystem link. > > * This causes failures in libudev that might also lead to libvirt errors. > > * A fix is upstream in master branch for kernel 5.1 > > [Fix] > > * 36360658eb5a6cf04bb9f2704d1e4ce54037ec99 3636065 "s390: vfio_ap: link the vfio_ap devices to the vfio_ap bus subsystem" > > [Test Case] > > * An s390x machine with at least one CryptoExpress card, > where at least one AP and one 'Domain' is assigned to a particular LPAR. > > * Now running virsh nodedev-list before (and later after) construction the matrix device should expose the issue. > > * For details about how to setup a vfio-ap matrix device see: http://kvmonz.blogspot.com/2018/12/qemu-v31-released.html > (see 2nd bullet: virtio-ap) > > [Regression Potential] > > * The regression potential can be considered as low since it only affects the s390x platform > > * and there it only affects the usage of AP (CryptoExpress) cards and it's driver/module > > * and again only affects the recently introduced virtual IO function of AP (vfio-ap). > > [Other Info] > > * It was already briefly discussed here: > https://www.redhat.com/archives/libvir-list/2019-February/msg00837.html > as well as reviewed and pushed. > > * Commit 3636065 from v5.1-rc1 > > * This affects ccw cards and their vf only, NOT vf of PCI/PCIe cards! > > * * For details on virtio-ap/vfio-ap see bullet #2 here: http://kvmonz.blogspot.com/2018/12/qemu-v31-released.html > > Pierre Morel (1): > s390: vfio_ap: link the vfio_ap devices to the vfio_ap bus subsystem > --- > drivers/s390/crypto/vfio_ap_drv.c | 44 ++++++++++++++++++++++++++++------- > drivers/s390/crypto/vfio_ap_ops.c | 4 ++-- > drivers/s390/crypto/vfio_ap_private.h | 1 + > 3 files changed, 38 insertions(+), 11 deletions(-) > > > -- > kernel-team mailing list > kernel-team@lists.ubuntu.com > https://lists.ubuntu.com/mailman/listinfo/kernel-team