[sanitizer] Extend sancov.py to show which PCs are missing from coverage.

[earthdok]

Summary

Example usage:

sancov.py print a.out.1234.sancov | sancov.py missing a.out

[earthdok]

• update regexp

[kcc]

looks good, but I wonder if a test can be added?
The pc-=4 hack is unfortunate, but I don't see a better fix...

[earthdok]

In D9228#165634, @kcc wrote:

looks good, but I wonder if a test can be added?

That would make python a requirement for running the test suite, not sure if good idea.

[kcc]

In D9228#165933, @earthdok wrote:

In D9228#165634, @kcc wrote:

looks good, but I wonder if a test can be added?

That would make python a requirement for running the test suite, not sure if good idea.

We already have tests that run sancov.py

[earthdok]

• update regexp
• fix regexp: accept "call", accept "@plt"
• 

[earthdok]

In D9228#166797, @kcc wrote:

We already have tests that run sancov.py

Done.

[earthdok]

• fix meaningless sorts

[kcc]

LGTM, thanks!

Once committed, please also update the docs at http://clang.llvm.org/docs/SanitizerCoverage.html#how-good-is-the-coverage

[earthdok]

Committed r236637

[earthdok]

In D9228#167162, @kcc wrote:

LGTM, thanks!

Once committed, please also update the docs at http://clang.llvm.org/docs/SanitizerCoverage.html#how-good-is-the-coverage

Committed r236641

[kcc]

Perfect!

lib/sanitizer_common/scripts/sancov.py

 #!/usr/bin/env python
 # Merge or print the coverage data collected by asan's coverage.
 # Input files are sequences of 4-byte integers.
 # We need to merge these integers into a set and then
 # either print them (as hex) or dump them into another file.
 import array
 import bisect
 import glob
 import os.path
 import struct
+import subprocess
 import sys
 
 prog_name = ""
 
 def Usage():
   print >> sys.stderr, "Usage: \n" + \
-      " " + prog_name + " [32|64] merge file1 [file2 ...]  > output\n" \
-      " " + prog_name + " [32|64] print file1 [file2 ...]\n" \
-      " " + prog_name + " [32|64] unpack file1 [file2 ...]\n" \
-      " " + prog_name + " [32|64] rawunpack file1 [file2 ...]\n"
+      " " + prog_name + " merge FILE [FILE...] > OUTPUT\n" \
+      " " + prog_name + " print FILE [FILE...]\n" \
+      " " + prog_name + " unpack FILE [FILE...]\n" \
+      " " + prog_name + " rawunpack FILE [FILE ...]\n" \
+      " " + prog_name + " missing BINARY < LIST_OF_PCS\n"
   exit(1)
 
 def CheckBits(bits):
   if bits != 32 and bits != 64:
     raise Exception("Wrong bitness: %d" % bits)
 
 def TypeCodeForBits(bits):
   CheckBits(bits)

 
 def RawUnpack(files):
   for f in files:
     if not f.endswith('.sancov.raw'):
       raise Exception('Unexpected raw file name %s' % f)
     f_map = f[:-3] + 'map'
     UnpackOneRawFile(f, f_map)
 
+def GetInstrumentedPCs(binary):
+  cmd = "objdump -d %s | " \
+        "grep '^\s\+[0-9a-f]\+:.*\scallq\s\+[0-9a-f]\+ <__sanitizer_cov>' | " \
+        "grep '^\s\+[0-9a-f]\+' -o" % binary
+  proc = subprocess.Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.PIPE,
+                          shell=True)
+  proc.stdin.close()
+  # The PCs we get from objdump are off by 4 bytes, as they point to the
+  # beginning of the callq instruction. Empirically this is true on x86 and
+  # x86_64.
+  return set(int(line.strip(), 16) + 4 for line in proc.stdout)
+
+def PrintMissing(binary):
+  if not os.path.isfile(binary):
+    raise Exception('File not found: %s' % binary)
+  instrumented = GetInstrumentedPCs(binary)
+  print >> sys.stderr, "%s: found %d instrumented PCs in %s" % (prog_name,
+                                                                len(instrumented),
+                                                                binary)
+  covered = set(int(line, 16) for line in sys.stdin)
+  print >> sys.stderr, "%s: read %d PCs from stdin" % (prog_name, len(covered))
+  missing = instrumented - covered
+  print >> sys.stderr, "%s: %d PCs missing from coverage" % (prog_name, len(missing))
+  if (len(missing) > len(instrumented) - len(covered)):
+    print >> sys.stderr, \
+        "%s: WARNING: stdin contains PCs not found in binary" % prog_name
+  for pc in sorted(missing):
+    print "0x%x" % pc
+
 if __name__ == '__main__':
   prog_name = sys.argv[0]
   if len(sys.argv) <= 2:
     Usage();
 
+  if sys.argv[1] == "missing":
+    if len(sys.argv) != 3:
+      Usage()
+    PrintMissing(sys.argv[2])
+    exit(0)
+
   file_list = []
   for f in sys.argv[2:]:
     file_list += glob.glob(f)
   if not file_list:
     Usage()
 
   if sys.argv[1] == "print":
     PrintFiles(file_list)
   elif sys.argv[1] == "merge":
     MergeAndPrint(file_list)
   elif sys.argv[1] == "unpack":
     Unpack(file_list)
   elif sys.argv[1] == "rawunpack":
     RawUnpack(file_list)
   else:
     Usage()