historical: restrict the number of cache misses to the actual compilation

experiments/historical_build.py

@@ -90,10 +90,6 @@ class HistoricalCompilation(Experiment, ClangHashHelper):
                            "commit-hash": self.metadata["project-hash"],
                            'builds': []}
 
-        hash_log = self.tmp_directory.new_file("clang-hash.log")
-        if "clang-hash" in self.mode.value:
-            os.environ["CLANG_HASH_LOGFILE"] = hash_log.path
-
         with self.project as src_path:
             (commits, _) = shell("cd %s; git log --no-merges --oneline --topo-order --format='%%H %%P %%s'", src_path)
             # [0] is hash. [1] is parent, [2] rest
@@ -151,13 +147,6 @@ class HistoricalCompilation(Experiment, ClangHashHelper):
         with open(self.stats.path, "w+") as fd:
             fd.write(repr(self.build_info))
 
-        if "ccache" in self.mode.value:
-            shell("ccache -s > %s", self.ccache_stats.path)
-
-        # Copy Log to the result directory
-        with open(hash_log.path) as fd:
-            self.clang_hash_stats.value = fd.read()
-
     def variant_name(self):
         return "%s-%s"%(self.project_name(), self.metadata['mode'])
 

experiments/lib.py

@@ -3,6 +3,7 @@ import fnmatch
 import time
 from versuchung.execute import shell, CommandFailed, shell_failok
 import logging
+import tempfile
 
 def read_hash_directory(hash_dir, remove_keys = []):
     """Read in all records from a hash dir
@@ -88,7 +89,25 @@ class ClangHashHelper:
         else:
             return shell("cd %s; make -j %s", path, str(self.jobs.value))
 
+    def ccache_hits(self):
+        (lines, _) = shell("ccache -s")
+        ccache_hits = 0
+        ccache_misses = 0
+        for line in lines:
+            if "cache hit" in line and "rate" not in line:
+                ccache_hits += int(line[line.index(")")+1:].strip())
+            if "cache miss" in line:
+                ccache_misses += int(line[line.index("miss")+4:].strip())
+        return ccache_hits, ccache_misses
+
     def rebuild(self, path, info, fail_ok=False):
+        if "ccache" in self.mode.value:
+            old_ccache_hits, old_ccache_misses = self.ccache_hits()
+
+        if "clang-hash" in self.mode.value:
+            hash_log = tempfile.NamedTemporaryFile()
+            os.environ["CLANG_HASH_LOGFILE"] = hash_log.name
+
         # Recompile!
         start_time = time.time()
         try:
@@ -101,9 +120,20 @@ class ClangHashHelper:
                 ret = ("", 1)
         end_time = time.time()
 
-        # Account only nano seconds, everywhere
         build_time = int((end_time - start_time) * 1e9)
         info['build-time'] = build_time
+
+        # Record Cache misses and hits
+        if "ccache" in self.mode.value:
+            ccache_hits, ccache_misses = self.ccache_hits()
+            info['ccache-hits'] = ccache_hits - old_ccache_hits
+            info['ccache-misses'] = ccache_misses - old_ccache_misses
+
+        if "clang-hash" in self.mode.value:
+            log = hash_log.read()
+            info['clang-hash-hits'] = log.count("H")
+            info['clang-hash-misses'] = log.count("M")
+
         logging.info("Rebuild done[%s]: %s s; failed=%s",
                      info.get("filename") or info.get("commit"),
                      build_time / 1e9,