[Qemu-devel] [PATCH] save kvm-specific msrs over save/load

[Qemu-devel] [PATCH] save kvm-specific msrs over save/load

[Glauber Costa]

Although we currently do not register a pvclock, there is no harm in saving the
values of the involved msrs. We'll just load an empty value. qemu-kvm, OTOH,
will make the correct use of it, so I think it is better to do it here, than
to augment the diff.

Signed-off-by: Glauber Costa <glommer@redhat.com>
---
 target-i386/cpu.h     |    4 +++-
 target-i386/kvm.c     |   11 +++++++++++
 target-i386/machine.c |    3 +++
 3 files changed, 17 insertions(+), 1 deletions(-)

diff --git a/target-i386/cpu.h b/target-i386/cpu.h
index 5929d28..ff7ae99 100644
--- a/target-i386/cpu.h
+++ b/target-i386/cpu.h
@@ -693,6 +693,8 @@ typedef struct CPUX86State {
     /* For KVM */
     uint64_t interrupt_bitmap[256 / 64];
     uint32_t mp_state;
+    uint64_t system_time_msr;
+    uint64_t wall_clock_msr;
 
     /* in order to simplify APIC support, we leave this pointer to the
        user */
@@ -870,7 +872,7 @@ uint64_t cpu_get_tsc(CPUX86State *env);
 #define cpu_signal_handler cpu_x86_signal_handler
 #define cpu_list x86_cpu_list
 
-#define CPU_SAVE_VERSION 11
+#define CPU_SAVE_VERSION 12
 
 /* MMU modes definitions */
 #define MMU_MODE0_SUFFIX _kernel
diff --git a/target-i386/kvm.c b/target-i386/kvm.c
index 7010999..a1c2fae 100644
--- a/target-i386/kvm.c
+++ b/target-i386/kvm.c
@@ -17,6 +17,7 @@
 #include <sys/mman.h>
 
 #include <linux/kvm.h>
+#include <linux/kvm_para.h>
 
 #include "qemu-common.h"
 #include "sysemu.h"
@@ -484,6 +485,8 @@ static int kvm_put_msrs(CPUState *env)
     kvm_msr_entry_set(&msrs[n++], MSR_FMASK, env->fmask);
     kvm_msr_entry_set(&msrs[n++], MSR_LSTAR, env->lstar);
 #endif
+    kvm_msr_entry_set(&msrs[n++], MSR_KVM_SYSTEM_TIME, env->system_time_msr);
+    kvm_msr_entry_set(&msrs[n++], MSR_KVM_WALL_CLOCK, env->wall_clock_msr);
     msr_data.info.nmsrs = n;
 
     return kvm_vcpu_ioctl(env, KVM_SET_MSRS, &msr_data);
@@ -617,6 +620,8 @@ static int kvm_get_msrs(CPUState *env)
     msrs[n++].index = MSR_FMASK;
     msrs[n++].index = MSR_LSTAR;
 #endif
+    msrs[n++].index = MSR_KVM_SYSTEM_TIME;
+    msrs[n++].index = MSR_KVM_WALL_CLOCK;
     msr_data.info.nmsrs = n;
     ret = kvm_vcpu_ioctl(env, KVM_GET_MSRS, &msr_data);
     if (ret < 0)
@@ -653,6 +658,12 @@ static int kvm_get_msrs(CPUState *env)
         case MSR_IA32_TSC:
             env->tsc = msrs[i].data;
             break;
+        case MSR_KVM_SYSTEM_TIME:
+            env->system_time_msr = msrs[i].data;
+            break;
+        case MSR_KVM_WALL_CLOCK:
+            env->wall_clock_msr = msrs[i].data;
+            break;
         }
     }
 
diff --git a/target-i386/machine.c b/target-i386/machine.c
index b13eff4..5ba2b6c 100644
--- a/target-i386/machine.c
+++ b/target-i386/machine.c
@@ -475,6 +475,9 @@ const VMStateDescription vmstate_cpu = {
         VMSTATE_UINT64_ARRAY_V(mce_banks, CPUState, MCE_BANKS_DEF *4, 10),
         /* rdtscp */
         VMSTATE_UINT64_V(tsc_aux, CPUState, 11),
+        /* kvm specific msrs */
+        VMSTATE_UINT64_V(system_time_msr, CPUState, 12),
+        VMSTATE_UINT64_V(wall_clock_msr, CPUState, 12),
         VMSTATE_END_OF_LIST()
     }
 };
-- 
1.6.2.2

Re: [Qemu-devel] [PATCH] save kvm-specific msrs over save/load

[Anthony Liguori]

Glauber Costa wrote:
> Although we currently do not register a pvclock, there is no harm in saving the
> values of the involved msrs. We'll just load an empty value. qemu-kvm, OTOH,
> will make the correct use of it, so I think it is better to do it here, than
> to augment the diff.
>   

How does the cpu versioning match up with qemu-kvm?

Has this gone into the qemu-kvm tree yet?

We've already bumped the version once for the 0.12 release so it's not 
necessary to bump it again.

Regards,

Anthony Liguori

Re: [Qemu-devel] [PATCH] save kvm-specific msrs over save/load

[Glauber Costa]

On Mon, Oct 12, 2009 at 09:21:39AM -0500, Anthony Liguori wrote:
> Glauber Costa wrote:
>> Although we currently do not register a pvclock, there is no harm in saving the
>> values of the involved msrs. We'll just load an empty value. qemu-kvm, OTOH,
>> will make the correct use of it, so I think it is better to do it here, than
>> to augment the diff.
>>   
>
> How does the cpu versioning match up with qemu-kvm?
>
> Has this gone into the qemu-kvm tree yet?
No. I opted by sendind it here first.

I can do the very opposite, however, if you prefer. Just let me know.

>
> We've already bumped the version once for the 0.12 release so it's not  
> necessary to bump it again.
okay.