Symbol information for callq and similar instructions

gistfile1.txt

#!/usr/bin/env python
from optparse import OptionParser 
from commands import getstatusoutput
from sys import exit

if __name__ == "__main__":
  parser = OptionParser(usage="%prog <pid>")
  parser.add_option("--json", "-j", dest="json", help="dump output as JSON dictionary", action="store_true", default=False)
  opts, args = parser.parse_args()
  if len(args) != 1:
    parser.error("PID not specified.")
    exit(1)

  # Get the loaded libraries and the address at which they where loaded.
  addrs = {}
  vmbases = {}
  symbols = [] 
  callsites = [] 

  error, output = getstatusoutput('pmap %s | grep -v -e "[[][ ].*[ ][]]" | grep -e "r.x--" | sort -u -k 4' % args[0])
  if error:
    parser.error("Cannot run pmap on PID %s\n%s " % (args[0], output))
    exit(1)

  # Get the load address for all the loaded libraries.
  for line in output.splitlines():
    addr, size, perm, filename = (x for x in line.split(" ") if x)
    addrs[filename] = int(addr, 16)
  
  # Get the vmbase for the executable section of each loaded library.
  for filename in addrs.keys():
    error, output = getstatusoutput("objdump -p %s | grep LOAD | head -1" % filename) 
    if error:
      print "Error while parsing the VMBASE for %s" % filename
    # Format:
    #     LOAD off    0x0000000000000000 vaddr 0x00000036d2000000 paddr 0x00000036d2000000 align 2**21
    d1, d2, offset, d3, vmbase, d4, paddr, d5, d6 = (x for x in output.split(" ") if x)
    vmbases[filename] = int(vmbase, 16)

  # Parse the disassembly and get the symbols and the call / callq addresses.
  symbols.append("<unknown>")
  currentSymbolIndex = 0

  for filename in addrs.keys():
    error, output = getstatusoutput('objdump -D %s | grep "\(call\|>:\)"' % filename)
    if error:
      print "Error while disassebling."
    currentOffset = addrs[filename]
    currentBase = vmbases[filename]
    for line in output.splitlines():
      if line.startswith(" "):
        address = int([x for x in line.split(":", 1) if x][0], 16)
        # This is the code which calculates the actual virtual address.
        vaddress = address - currentBase + currentOffset
        callsites.append((vaddress, currentSymbolIndex))
        if not opts.json:
          print "%s, %s" % (vaddress, symbols[currentSymbolIndex])
      elif line.endswith(":"):
        symbol = line.rstrip(">:").split("<")[1]
        symbols.append(symbol)
        currentSymbolIndex += 1
 
  if opts.json:
    print "["
    print symbols, ","
    print dict(callsites)
    print "]"