kvm: Set the perf exclude_host attribute if available

The performance counting framework in Linux 3.2 and onwards supports
an attribute to exclude events generated by the host when running
KVM. Setting this attribute allows us to get more reliable
measurements of the guest machine. For example, on a highly loaded
system, the instruction counts from the guest can be severely
distorted by the host kernel (e.g., by page fault handlers).

This changeset introduces a check for the attribute and enables it in
the KVM CPU if present.

2 files changed, 47 insertions, 0 deletions

diff --git a/src/cpu/kvm/base.cc b/src/cpu/kvm/base.cc
index a8e115da7..d25e145a5 100644
--- a/src/cpu/kvm/base.cc
+++ b/src/cpu/kvm/base.cc
@@ -1133,6 +1133,12 @@ BaseKvmCPU::setupCounters()
     cfgCycles.disabled(true)
         .pinned(true);
 
+    // Try to exclude the host. We set both exclude_hv and
+    // exclude_host since different architectures use slightly
+    // different APIs in the kernel.
+    cfgCycles.exclude_hv(true)
+        .exclude_host(true);
+
     if (perfControlledByTimer) {
         // We need to configure the cycles counter to send overflows
         // since we are going to use it to trigger timer signals that
@@ -1206,6 +1212,12 @@ BaseKvmCPU::setupInstCounter(uint64_t period)
     PerfKvmCounterConfig cfgInstructions(PERF_TYPE_HARDWARE,
                                          PERF_COUNT_HW_INSTRUCTIONS);
 
+    // Try to exclude the host. We set both exclude_hv and
+    // exclude_host since different architectures use slightly
+    // different APIs in the kernel.
+    cfgInstructions.exclude_hv(true)
+        .exclude_host(true);
+
     if (period) {
         // Setup a sampling counter if that has been requested.
         cfgInstructions.wakeupEvents(1)
diff --git a/src/cpu/kvm/perfevent.hh b/src/cpu/kvm/perfevent.hh
index eed900994..b1b0da283 100644
--- a/src/cpu/kvm/perfevent.hh
+++ b/src/cpu/kvm/perfevent.hh
@@ -45,6 +45,8 @@
 
 #include <inttypes.h>
 
+#include "config/have_perf_attr_exclude_host.hh"
+
 /**
  * PerfEvent counter configuration.
  */
@@ -125,6 +127,39 @@ class PerfKvmCounterConfig
         return *this;
     }
 
+    /**
+     * Exclude the events from the host (i.e., only include events
+     * from the guest system).
+     *
+     * Intel CPUs seem to support this attribute from Linux 3.2 and
+     * onwards. Non-x86 architectures currently ignore this attribute
+     * (Linux 3.12-rc5).
+     *
+     * @warn This attribute is ignored if it isn't present in the
+     * kernel headers or if the kernel doesn't support it.
+     *
+     * @param val true to exclude host events
+     */
+    PerfKvmCounterConfig &exclude_host(bool val) {
+#if HAVE_PERF_ATTR_EXCLUDE_HOST == 1
+        attr.exclude_host = val;
+#endif
+        return *this;
+    }
+
+    /**
+     * Exclude the hyper visor (i.e., only include events from the
+     * guest system).
+     *
+     * @warn This is attribute only seems to be ignored on Intel.
+     *
+     * @param val true to exclude host events
+     */
+    PerfKvmCounterConfig &exclude_hv(bool val) {
+        attr.exclude_hv = val;
+        return *this;
+    }
+
     /** Underlying perf_event_attr structure describing the counter */
     struct perf_event_attr attr;
 };