hphp/tools/perf-script-raw.py

   1 #!/usr/bin/env python
   2 #
   3 # This script processes the output of 'perf script' and prints raw data in the
   4 # the following format:
   5 #
   6 # BINARY_NAME EVENT NUM_IPS
   7 # INSTRUCTION_PTR_1
   8 # INSTRUCTION_PTR_2
   9 # ...
  10 # INSTRUCTION_PTR_NUM_IPS
  11 #
  12 # The perf data may be recorded with -g or not.
  13 #
  14
  15 import string, sys
  16
  17 #
  18 # Positions of various fields within a line.  Assumes lines look like this:
  19 # BINARY  EVENT    IP     FUNCTION
  20 # hhvm    cycles:  d87415 HPHP::ObjectData::destruct()
  21 #
  22 BINARY_NAME_FIELD = 0
  23 EVENT_FIELD = 1
  24 IP_FIELD = 2
  25 FUNCTION_FIELD = 3
  26
  27 def remove_raw_annotation(header):
  28     """For non-generic events, perf adds a 'raw' annotation before the
  29        event id: remove it and return the modified header."""
  30     fields = header.split(None)
  31     if (fields[EVENT_FIELD] == 'raw'):
  32       del fields[4]
  33     return ' '.join(fields)
  34
  35 def extract_function(long_symbol):
  36     short_symbol = long_symbol.split('(', 1)[0]
  37     if short_symbol == "":
  38         return "unknown"
  39     return short_symbol
  40
  41 def process(record):
  42     if state == WITH_STACKTRACES:
  43         header = record[0]
  44     else:
  45         header = record
  46
  47     header = remove_raw_annotation(header)
  48
  49     fields            = header.split(None)
  50     binary_name       = fields[BINARY_NAME_FIELD]
  51     event             = fields[EVENT_FIELD][:-1]
  52     trace_entries     = []
  53
  54     if state == NORMAL:
  55         # perf record was called without "-g"
  56         try:
  57             entry = (int(fields[IP_FIELD], 16),
  58                      extract_function(fields[FUNCTION_FIELD]))
  59             trace_entries.append(entry)
  60         except ValueError:
  61             # Sometimes you get weird garbage from perf script (This
  62             # fix is motivated by a record that had spaces in the
  63             # binary name, sigh).  Ignore and move on.
  64             return
  65     elif state == WITH_STACKTRACES:
  66         # perf record was called with "-g"
  67         if len(record) == 1:
  68             return
  69         for line in record[1:]:
  70             line_fields = line.split(None)
  71             entry = (int(line_fields[0], 16),
  72                      extract_function(line_fields[1]))
  73             trace_entries.append(entry)
  74
  75     print("%s %s %u" % (binary_name, event, len(trace_entries)))
  76     for entry in trace_entries:
  77         print("0x%lx %s" % (entry[0], entry[1]))
  78
  79
  80 NORMAL            = 0
  81 WITH_STACKTRACES  = 1
  82
  83 def records():
  84     global state
  85
  86     # Skip comment lines.
  87     while True:
  88         line = sys.stdin.readline()
  89         if (not line):
  90             return
  91         if (line[0] != '#'):
  92             break
  93
  94     if len(line.split(None)) >= 4:
  95         state = NORMAL
  96     else:
  97         state = WITH_STACKTRACES
  98
  99     if state == NORMAL:
 100         yield line
 101         for line in sys.stdin:
 102             if (line[0] == "#"):
 103                 continue
 104             yield line
 105         return
 106
 107     # state == WITH_STACKTRACES
 108     lines = [line]
 109     for line in sys.stdin:
 110         if (line[0] == "#"):
 111             continue
 112         if line == '\n':
 113             yield lines
 114             lines = []
 115         else:
 116             lines.append(line)
 117     if (len(lines) > 0):
 118         yield lines
 119
 120 if __name__ == '__main__':
 121     for record in records():
 122         process(record)