Small ruby script to prettify dump output from pin

prettydump.rb

#!/usr/bin/env ruby

# Dump format should be like this
#   (STAGE) OP (OP_ID) ... cycle (CYCLE)
# Matching is case insensitive
#
# Examples:
#   FE_STAGE OP 0 is fetched at cycle 1
#   ID_STAGE OP 0 is scheduled at cycle 3
#   IS_STAGE OP 0 is issued at cycle 3
#   EX_STAGE OP 0 finished execution at cycle 5
#   WB_STAGE OP 0 is retired at cycle 6
#
# Look at the addstage function for stages defined
#
# The following are the current stages
#   FE_stage: after instruction is fetched
#   ID_stage: after instruction is added to scheduler
#   IS_stage: instruction is selected for execution i.e will be in EX stage next cycle
#   EX_stage: instruction finished execution 
#   WB_stage: instruction retired
#
# Dont be confused by ID_stage, Op is added to scheduler in IS_stage, just easier
# to differentiate in the parsing
#
# Usage:
# ../../../pin -t obj-intel64/sim.so -readtrace 1 -printinst 0 -max_sim_count 0 -max_inst_count 2 -print_pipe_freq 0 # -tracename trace2.pzip -- /bin/ls | ruby prettydump.rb 
#
# Higher the tab width better the formatting of output


# Monkey patch to NilClass to print "-" instead of "" for nil case
class NilClass
  def to_s
    return "-"
  end
end

# Op class
class Op
  attr_accessor :op_id, :fe, :id, :is, :ex, :wb

  def initialize(id)
    @op_id = id
  end

  def addstage(stage, cycle)
    case stage
      when 'FE_STAGE'
        @fe = cycle
      when 'ID_STAGE'
        @id = cycle
      when 'IS_STAGE'
        @is = cycle
      when 'EX_STAGE'
        @ex = cycle
      when 'WB_STAGE'
        @wb = cycle
    end
  end

  def to_s
    @fe.to_s+"\t"[email protected]_s+"\t"[email protected]_s+"\t"[email protected]_s+"\t"[email protected]_s
  end
end

ops = Array.new
op_cnt = -1
op_ex = Regexp.new(/(.+)\s+OP\s+(\d+).+cycle\s+(\d+)/i)

# Iterate over the dump file
ARGF.each{
  |line|
  matchdata = op_ex.match(line)

  # Add stage to the op found in ops, if not create and then add
  ops.find(lambda{
    op = Op.new(op_cnt+=1)
    ops.push(op)
    return op
  }) { 
    |op| 
    op.op_id == matchdata[2].to_i
  }.addstage(matchdata[1], matchdata[3]) if matchdata
}

# the actual pretty print
puts <<-STAGE_INFORMATION
# The following are the current stages
#   FE: after instruction is fetched
#   ID: after instruction is added to scheduler
#   IS: instruction is selected for execution
#   EX: instruction finished execution 
#   WB: instruction retired
STAGE_INFORMATION

puts "\n\n "+"\tFE\tID\tIS\tEX\tWB\n"
ops.each{
  |op|
  puts op.op_id.to_s + "\t" + op.to_s
}