summary refs log tree commit diff stats
path: root/hw/i386/kvm/clock.c
diff options
context:
space:
mode:
authorLiang Li <liang.z.li@intel.com>2015-11-05 11:51:03 +0800
committerPaolo Bonzini <pbonzini@redhat.com>2015-11-05 11:28:10 +0100
commit0fd7e098db30e302d27920487f0afec33be8982a (patch)
tree87cfe5cc1dc319430ae507e6ecf76199b349e9b3 /hw/i386/kvm/clock.c
parentb553a0428014636bce4951098e97c60dc18a83ed (diff)
downloadfocaccia-qemu-0fd7e098db30e302d27920487f0afec33be8982a.tar.gz
focaccia-qemu-0fd7e098db30e302d27920487f0afec33be8982a.zip
kvmclock: add a new function to update env->tsc.
The commit 317b0a6d8 fixed an issue which caused by the outdated
env->tsc value, but the fix lead to 'cpu_synchronize_all_states()'
called twice during live migration. The 'cpu_synchronize_all_states()'
takes about 130us for a VM which has 4 vcpus, it's a bit expensive.

Synchronize the whole CPU context just for updating env->tsc is too
wasting, this patch use a new function to update the env->tsc.
Comparing to 'cpu_synchronize_all_states()', it only takes about 20us.

Signed-off-by: Liang Li <liang.z.li@intel.com>
Message-Id: <1446695464-27116-2-git-send-email-liang.z.li@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Diffstat (limited to 'hw/i386/kvm/clock.c')
-rw-r--r--hw/i386/kvm/clock.c18
1 files changed, 2 insertions, 16 deletions
diff --git a/hw/i386/kvm/clock.c b/hw/i386/kvm/clock.c
index efdf165848..0593a3f1f5 100644
--- a/hw/i386/kvm/clock.c
+++ b/hw/i386/kvm/clock.c
@@ -17,7 +17,7 @@
 #include "qemu/host-utils.h"
 #include "sysemu/sysemu.h"
 #include "sysemu/kvm.h"
-#include "sysemu/cpus.h"
+#include "kvm_i386.h"
 #include "hw/sysbus.h"
 #include "hw/kvm/clock.h"
 
@@ -125,21 +125,7 @@ static void kvmclock_vm_state_change(void *opaque, int running,
             return;
         }
 
-        cpu_synchronize_all_states();
-        /* In theory, the cpu_synchronize_all_states() call above wouldn't
-         * affect the rest of the code, as the VCPU state inside CPUState
-         * is supposed to always match the VCPU state on the kernel side.
-         *
-         * In practice, calling cpu_synchronize_state() too soon will load the
-         * kernel-side APIC state into X86CPU.apic_state too early, APIC state
-         * won't be reloaded later because CPUState.vcpu_dirty==true, and
-         * outdated APIC state may be migrated to another host.
-         *
-         * The real fix would be to make sure outdated APIC state is read
-         * from the kernel again when necessary. While this is not fixed, we
-         * need the cpu_clean_all_dirty() call below.
-         */
-        cpu_clean_all_dirty();
+        kvm_synchronize_all_tsc();
 
         ret = kvm_vm_ioctl(kvm_state, KVM_GET_CLOCK, &data);
         if (ret < 0) {