target-i386: kvm: cache KVM_GET_SUPPORTED_CPUID data

KVM_GET_SUPPORTED_CPUID ioctl is called frequently when initializing
CPU. Depends on CPU features and CPU count, the number of calls can be
extremely high which slows down QEMU booting significantly. In our
testing, we saw 5922 calls with switches:

    -cpu SandyBridge -smp 6,sockets=6,cores=1,threads=1

This ioctl takes more than 100ms, which is almost half of the total
QEMU startup time.

While for most cases the data returned from two different invocations
are not changed, that means, we can cache the data to avoid trapping
into kernel for the second time. To make sure the cache safe one
assumption is desirable: the ioctl is stateless. This is not true for
CPUID leaves in general (such as CPUID leaf 0xD, whose value depends
on guest XCR0 and IA32_XSS) but it is true of KVM_GET_SUPPORTED_CPUID,
which runs before there is a value for XCR0 and IA32_XSS.

Signed-off-by: Chao Peng <chao.p.peng@linux.intel.com>
Message-Id: <1465784487-23482-1-git-send-email-chao.p.peng@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

1 files changed, 7 insertions, 2 deletions

diff --git a/target-i386/kvm.c b/target-i386/kvm.c
index 7b092ee385..ff92b1d118 100644
--- a/target-i386/kvm.c
+++ b/target-i386/kvm.c
@@ -106,6 +106,8 @@ static int has_xsave;
 static int has_xcrs;
 static int has_pit_state2;
 
+static struct kvm_cpuid2 *cpuid_cache;
+
 int kvm_has_pit_state2(void)
 {
     return has_pit_state2;
@@ -199,9 +201,14 @@ static struct kvm_cpuid2 *get_supported_cpuid(KVMState *s)
 {
     struct kvm_cpuid2 *cpuid;
     int max = 1;
+
+    if (cpuid_cache != NULL) {
+        return cpuid_cache;
+    }
     while ((cpuid = try_get_cpuid(s, max)) == NULL) {
         max *= 2;
     }
+    cpuid_cache = cpuid;
     return cpuid;
 }
 
@@ -319,8 +326,6 @@ uint32_t kvm_arch_get_supported_cpuid(KVMState *s, uint32_t function,
         ret |= cpuid_1_edx & CPUID_EXT2_AMD_ALIASES;
     }
 
-    g_free(cpuid);
-
     /* fallback for older kernels */
     if ((function == KVM_CPUID_FEATURES) && !found) {
         ret = get_para_features(s);