From 4b6c0b986907c44e816520531b84293339496fac Mon Sep 17 00:00:00 2001
From: Brendan Jackman <brendan.jackman@arm.com>
Date: Wed, 11 Oct 2017 15:24:22 +0100
Subject: [PATCH] schedstats: Add schedstats instrument

---
 wa/instrumentation/schedstats.py | 75 ++++++++++++++++++++++++++++++++
 1 file changed, 75 insertions(+)
 create mode 100644 wa/instrumentation/schedstats.py

diff --git a/wa/instrumentation/schedstats.py b/wa/instrumentation/schedstats.py
new file mode 100644
index 000000000..f96982a4d
--- /dev/null
+++ b/wa/instrumentation/schedstats.py
@@ -0,0 +1,75 @@
+#    Copyright 2017 ARM Limited
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+from devlib import SchedstatsInstrument as _Schedstats
+
+from wa import Instrument
+from wa.framework.exception import InstrumentError
+
+class SchedstatsInstrument(Instrument):
+    name = 'schedstats'
+
+    description = """
+    Collects scheduler statistics info and reports them as Metrics
+
+    This uses the devlib schedstats instrument to collect schedstats information
+    at the beginning and at the end of the workload. It then reports the delta
+    in each statistic as a Metric.
+
+    The metrics are named according to corresponding identifiers in the
+    kernel scheduler code. The names for ``sched_domain.lb_*`` stats, which are
+    recorded per ``cpu_idle_type`` are suffixed with a ':' followed by the idle
+    type, for example ``'lb_balanced:CPU_NEWLY_IDLE'``.
+
+    A classifier called 'node' is used to differentiate between the metrics from
+    different sources (each CPU and sched_domain). Therefore you might want to
+    make sure your result_processors report that classifier. For the CSV
+    processor, for example, the simplest way is to set ``use_all_classifiers``
+    to ``True``.
+    """
+
+    def initialize(self, context):
+        if not self.target.file_exists(_Schedstats.schedstat_path) and \
+           not self.target.file_exists(_Schedstats.sysctl_path):
+            raise InstrumentError('schedstats not supported by target. '
+                                  'Ensure CONFIG_SCHEDSTATS is enabled.')
+
+    def setup(self, context):
+        self.instrument = _Schedstats(self.target)
+        self.instrument.reset()
+
+    def start(self, context):
+        self.before = self.instrument.take_measurement()
+
+    def stop(self, context):
+        self.after = self.instrument.take_measurement()
+
+    def update_result(self, context):
+        before = {m.channel.label: m for m in self.before}
+        after = {m.channel.label: m for m in self.after}
+
+        if set(before.keys()) != set(after.keys()):
+            raise InstrumentError('schedstats measurements returned different entries!')
+
+        for label, measurement_after in after.iteritems():
+            measurement_before = before[label]
+            diff = measurement_after.value - measurement_before.value
+            channel = measurement_after.channel
+            context.add_metric(channel.kind, diff,
+                               classifiers={'node': channel.site})
+
+    def teardown(self, context):
+        self.instrument.teardown()
+