| Message ID | 20221012123229.1196007-1-nrb@linux.ibm.com |
|---|---|
| State | New |
| Headers | show |
| Series | [v1] s390x/tod-kvm: don't save/restore the TOD in PV guests | expand |
On 12/10/2022 14.32, Nico Boehr wrote: > Under PV, the guest's TOD clock is under control of the ultravisor and the > hypervisor cannot change it. > > With upcoming kernel changes[1], the Linux kernel will reject QEMU's > request to adjust the guest's clock in this case, so don't attempt to set > the clock. > > This avoids the following warning message on save/restore of a PV guest: > > warning: Unable to set KVM guest TOD clock: Operation not supported > > [1] https://lore.kernel.org/all/20221011160712.928239-2-nrb@linux.ibm.com/ > > Fixes: c3347ed0d2ee ("s390x: protvirt: Support unpack facility") > Signed-off-by: Nico Boehr <nrb@linux.ibm.com> > --- > hw/s390x/tod-kvm.c | 8 ++++++++ > 1 file changed, 8 insertions(+) > > diff --git a/hw/s390x/tod-kvm.c b/hw/s390x/tod-kvm.c > index 9d0cbfbce2bf..303bd67ee64f 100644 > --- a/hw/s390x/tod-kvm.c > +++ b/hw/s390x/tod-kvm.c > @@ -13,6 +13,7 @@ > #include "qemu/module.h" > #include "sysemu/runstate.h" > #include "hw/s390x/tod.h" > +#include "hw/s390x/pv.h" > #include "kvm/kvm_s390x.h" > > static void kvm_s390_get_tod_raw(S390TOD *tod, Error **errp) > @@ -84,6 +85,13 @@ static void kvm_s390_tod_vm_state_change(void *opaque, bool running, > S390TODState *td = opaque; > Error *local_err = NULL; > > + /* > + * Under PV, the clock is under ultravisor control, hence we cannot restore > + * it on resume. > + */ > + if (s390_is_pv()) > + return; Hi Nico, I know it's annoying when switching between kernel coding style and QEMU coding style, but please use curly braces when doing QEMU patches. I wonder why checkpatch.pl does not print any warnings here...? Anyway, since it's a trivial patch, I fixed it up on my own and queued your patch to my s390x-next branch: https://gitlab.com/thuth/qemu/-/commits/s390x-next/ Thomas
Quoting Thomas Huth (2022-10-17 09:30:04) [...] > I know it's annoying when switching between kernel coding style and QEMU > coding style, but please use curly braces when doing QEMU patches. I wonder > why checkpatch.pl does not print any warnings here...? Ooops, sorry for the oversight. You are right, thanks for making me aware of the different coding styles.
diff --git a/hw/s390x/tod-kvm.c b/hw/s390x/tod-kvm.c index 9d0cbfbce2bf..303bd67ee64f 100644 --- a/hw/s390x/tod-kvm.c +++ b/hw/s390x/tod-kvm.c @@ -13,6 +13,7 @@ #include "qemu/module.h" #include "sysemu/runstate.h" #include "hw/s390x/tod.h" +#include "hw/s390x/pv.h" #include "kvm/kvm_s390x.h" static void kvm_s390_get_tod_raw(S390TOD *tod, Error **errp) @@ -84,6 +85,13 @@ static void kvm_s390_tod_vm_state_change(void *opaque, bool running, S390TODState *td = opaque; Error *local_err = NULL; + /* + * Under PV, the clock is under ultravisor control, hence we cannot restore + * it on resume. + */ + if (s390_is_pv()) + return; + if (running && td->stopped) { /* Set the old TOD when running the VM - start the TOD clock. */ kvm_s390_set_tod_raw(&td->base, &local_err);
Under PV, the guest's TOD clock is under control of the ultravisor and the hypervisor cannot change it. With upcoming kernel changes[1], the Linux kernel will reject QEMU's request to adjust the guest's clock in this case, so don't attempt to set the clock. This avoids the following warning message on save/restore of a PV guest: warning: Unable to set KVM guest TOD clock: Operation not supported [1] https://lore.kernel.org/all/20221011160712.928239-2-nrb@linux.ibm.com/ Fixes: c3347ed0d2ee ("s390x: protvirt: Support unpack facility") Signed-off-by: Nico Boehr <nrb@linux.ibm.com> --- hw/s390x/tod-kvm.c | 8 ++++++++ 1 file changed, 8 insertions(+)