1 files changed, 46 insertions, 0 deletions

diff --git a/sanitizer.py b/sanitizer.py
new file mode 100644
index 0000000..d1764dd
--- /dev/null
+++ b/sanitizer.py
@@ -0,0 +1,46 @@
+###############################################################################
+# Description
+###############################################################################
+
+# Sanitize input. (There are a number of goofy issues with the sched_trace
+# output.)
+
+###############################################################################
+# Public functions
+###############################################################################
+
+def sanitizer(stream):
+    jobs_released = []
+    for record in stream:
+
+        # All records with job < 2 are garbage
+        if record.job < 2:
+            continue
+
+        # Some records with job == 2 are garbage
+        if record.job==2:
+
+            # There is a duplicate release of every job 2
+            # This will throw away the second one
+            if (record.type_name == 'release' and
+                    (record.pid,record.job) in jobs_released):
+                continue
+
+            # Job 2 has a resume that is garbage
+            if record.type_name == 'resume':
+                continue
+
+        # Record job releases. This helps with filtering out
+        # the second release of job 2 and renumbering final switch_aways
+        # (see below).
+        if record.type_name == 'release':
+            jobs_released.append((record.pid,record.job))
+
+        # By default, the switch_away for a job (after it has completed)
+        # is maked as being for job+1, which has not been released.
+        # Correct this.
+        if record.type_name == 'switch_away':
+            if (record.pid,record.job) not in jobs_released:
+                record.job -= 1
+        
+        yield record