Re: [PATCH v4 07/10] qemu-kvm: Cleanup/fix TSC and PV clock writeback

[Jan Kiszka <jan.kiszka@siemens.com>]

Drop kvm_load_tsc in favor of level-dependent writeback in
kvm_arch_load_regs. KVM's PV clock MSRs fall in the same category and
should therefore only be written back on full sync.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
---

Changes in v4:
 - only write TSC MSRs of SMP guests if they are non-zero, ie. after
   migration (behavior of original code)

 qemu-kvm-x86.c        |   27 +++++++++++++--------------
 qemu-kvm.h            |    4 ----
 target-i386/machine.c |    5 -----
 3 files changed, 13 insertions(+), 23 deletions(-)

diff --git a/qemu-kvm-x86.c b/qemu-kvm-x86.c
index 840c1c9..285c05a 100644
--- a/qemu-kvm-x86.c
+++ b/qemu-kvm-x86.c
@@ -965,8 +965,19 @@ void kvm_arch_load_regs(CPUState *env, int level)
         set_msr_entry(&msrs[n++], MSR_LSTAR  ,           env->lstar);
     }
 #endif
-    set_msr_entry(&msrs[n++], MSR_KVM_SYSTEM_TIME,  env->system_time_msr);
-    set_msr_entry(&msrs[n++], MSR_KVM_WALL_CLOCK,  env->wall_clock_msr);
+    if (level == KVM_PUT_FULL_STATE) {
+        /*
+         * KVM is yet unable to synchronize TSC values of multiple VCPUs on
+         * writeback. Until this is fixed, we only write the offset to SMP
+         * guests after migration, desynchronizing the VCPUs, but avoiding
+         * huge jump-backs that would occur without any writeback at all.
+         */
+        if (smp_cpus == 1 || env->tsc != 0) {
+            set_msr_entry(&msrs[n++], MSR_IA32_TSC, env->tsc);
+        }
+        set_msr_entry(&msrs[n++], MSR_KVM_SYSTEM_TIME, env->system_time_msr);
+        set_msr_entry(&msrs[n++], MSR_KVM_WALL_CLOCK, env->wall_clock_msr);
+    }
 
     rc = kvm_set_msrs(env, msrs, n);
     if (rc == -1)
@@ -986,18 +997,6 @@ void kvm_arch_load_regs(CPUState *env, int level)
     kvm_guest_debug_workarounds(env);
 }
 
-void kvm_load_tsc(CPUState *env)
-{
-    int rc;
-    struct kvm_msr_entry msr;
-
-    set_msr_entry(&msr, MSR_IA32_TSC, env->tsc);
-
-    rc = kvm_set_msrs(env, &msr, 1);
-    if (rc == -1)
-        perror("kvm_set_tsc FAILED.\n");
-}
-
 void kvm_arch_save_regs(CPUState *env)
 {
     struct kvm_regs regs;
diff --git a/qemu-kvm.h b/qemu-kvm.h
index c6ff652..827cac5 100644
--- a/qemu-kvm.h
+++ b/qemu-kvm.h
@@ -958,7 +958,6 @@ int handle_tpr_access(void *opaque, CPUState *env, uint64_t rip,
 #ifdef TARGET_I386
 #define qemu_kvm_has_pit_state2() kvm_has_pit_state2(kvm_context)
 #endif
-void kvm_load_tsc(CPUState *env);
 #else
 #define kvm_nested 0
 #define qemu_kvm_has_gsi_routing() (0)
@@ -966,9 +965,6 @@ void kvm_load_tsc(CPUState *env);
 #define qemu_kvm_has_pit_state2() (0)
 #endif
 #define qemu_kvm_cpu_stop(env) do {} while(0)
-static inline void kvm_load_tsc(CPUState *env)
-{
-}
 #endif
 
 void kvm_mutex_unlock(void);
diff --git a/target-i386/machine.c b/target-i386/machine.c
index bcc315b..b547e2a 100644
--- a/target-i386/machine.c
+++ b/target-i386/machine.c
@@ -354,11 +354,6 @@ static int cpu_post_load(void *opaque, int version_id)
         hw_breakpoint_insert(env, i);
 
     tlb_flush(env, 1);
-
-    if (kvm_enabled()) {
-        kvm_load_tsc(env);
-    }
-
     return 0;
 }