Class: SyntaxTree::YARV::DataFlowGraph

Inherits:
Object
  • Object
show all
Defined in:
lib/syntax_tree/yarv/data_flow_graph.rb

Overview

Constructs a data-flow-graph of a YARV instruction sequence, via a control-flow-graph. Data flow is discovered locally and then globally. The graph only considers data flow through the stack - local variables and objects are considered fully escaped in this analysis.

You can use this class by calling the ::compile method and passing it a control flow graph. It will return a data flow graph object.

iseq = RubyVM::InstructionSequence.compile("1 + 2")
iseq = SyntaxTree::YARV::InstructionSequence.from(iseq.to_a)
cfg = SyntaxTree::YARV::ControlFlowGraph.compile(iseq)
dfg = SyntaxTree::YARV::DataFlowGraph.compile(cfg)

Defined Under Namespace

Classes: BlockArgument, Compiler, DataFlow, LocalArgument

Instance Attribute Summary collapse

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(cfg, insn_flows, block_flows) ⇒ DataFlowGraph

Returns a new instance of DataFlowGraph.



68
69
70
71
72
 
# File 'lib/syntax_tree/yarv/data_flow_graph.rb', line 68

def initialize(cfg, insn_flows, block_flows)
  @cfg = cfg
  @insn_flows = insn_flows
  @block_flows = block_flows
end

Instance Attribute Details

#block_flows ⇒ Object (readonly)

Returns the value of attribute block_flows.



66
67
68
 
# File 'lib/syntax_tree/yarv/data_flow_graph.rb', line 66

def block_flows
  @block_flows
end

#cfg ⇒ Object (readonly)

Returns the value of attribute cfg.



66
67
68
 
# File 'lib/syntax_tree/yarv/data_flow_graph.rb', line 66

def cfg
  @cfg
end

#insn_flows ⇒ Object (readonly)

Returns the value of attribute insn_flows.



66
67
68
 
# File 'lib/syntax_tree/yarv/data_flow_graph.rb', line 66

def insn_flows
  @insn_flows
end

Class Method Details

.compile(cfg) ⇒ Object 



204
205
206
 
# File 'lib/syntax_tree/yarv/data_flow_graph.rb', line 204

def self.compile(cfg)
  Compiler.new(cfg).compile
end

Instance Method Details

#blocks ⇒ Object 



74
75
76
 
# File 'lib/syntax_tree/yarv/data_flow_graph.rb', line 74

def blocks
  cfg.blocks
end

#disasm ⇒ Object 



78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
 
# File 'lib/syntax_tree/yarv/data_flow_graph.rb', line 78

def disasm
  fmt = Disassembler.new(cfg.iseq)
  fmt.puts("== dfg: #{cfg.iseq.inspect}")

  blocks.each do |block|
    fmt.puts(block.id)
    fmt.with_prefix("    ") do |prefix|
      unless block.incoming_blocks.empty?
        from = block.incoming_blocks.map(&:id)
        fmt.puts("#{prefix}== from: #{from.join(", ")}")
      end

      block_flow = block_flows.fetch(block.id)
      unless block_flow.in.empty?
        fmt.puts("#{prefix}== in: #{block_flow.in.join(", ")}")
      end

      fmt.format_insns!(block.insns, block.block_start) do |_, length|
        insn_flow = insn_flows[length]
        next if insn_flow.in.empty? && insn_flow.out.empty?

        fmt.print(" # ")
        unless insn_flow.in.empty?
          fmt.print("in: #{insn_flow.in.join(", ")}")
          fmt.print("; ") unless insn_flow.out.empty?
        end

        unless insn_flow.out.empty?
          fmt.print("out: #{insn_flow.out.join(", ")}")
        end
      end

      to = block.outgoing_blocks.map(&:id)
      to << "leaves" if block.insns.last.leaves?
      fmt.puts("#{prefix}== to: #{to.join(", ")}")

      unless block_flow.out.empty?
        fmt.puts("#{prefix}== out: #{block_flow.out.join(", ")}")
      end
    end
  end

  fmt.string
end

#to_mermaid ⇒ Object 



127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
 
# File 'lib/syntax_tree/yarv/data_flow_graph.rb', line 127

def to_mermaid
  Mermaid.flowchart do |flowchart|
    disasm = Disassembler::Squished.new

    blocks.each do |block|
      block_flow = block_flows.fetch(block.id)
      graph_name =
        if block_flow.in.any?
          "#{block.id} #{block_flows[block.id].in.join(", ")}"
        else
          block.id
        end

      flowchart.subgraph(graph_name) do
        previous = nil

        block.each_with_length do |insn, length|
          node =
            flowchart.node(
              "node_#{length}",
              "%04d %s" % [length, insn.disasm(disasm)],
              shape: :rounded
            )

          flowchart.link(previous, node, color: :red) if previous
          insn_flows[length].in.each do |input|
            if input.is_a?(LocalArgument)
              from = flowchart.fetch("node_#{input.length}")
              flowchart.link(from, node, color: :green)
            end
          end

          previous = node
        end
      end
    end

    blocks.each do |block|
      block.outgoing_blocks.each do |outgoing|
        offset =
          block.block_start + block.insns.sum(&:length) -
            block.insns.last.length

        from = flowchart.fetch("node_#{offset}")
        to = flowchart.fetch("node_#{outgoing.block_start}")
        flowchart.link(from, to, color: :red)
      end
    end
  end
end

#to_son ⇒ Object 



123
124
125
 
# File 'lib/syntax_tree/yarv/data_flow_graph.rb', line 123

def to_son
  SeaOfNodes.compile(self)
end

#verify ⇒ Object

Verify that we constructed the data flow graph correctly.



179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
 
# File 'lib/syntax_tree/yarv/data_flow_graph.rb', line 179

def verify
  # Check that the first block has no arguments.
  raise unless block_flows.fetch(blocks.first.id).in.empty?

  # Check all control flow edges between blocks pass the right number of
  # arguments.
  blocks.each do |block|
    block_flow = block_flows.fetch(block.id)

    if block.outgoing_blocks.empty?
      # With no outgoing blocks, there should be no output arguments.
      raise unless block_flow.out.empty?
    else
      # Check with outgoing blocks...
      block.outgoing_blocks.each do |outgoing_block|
        outgoing_flow = block_flows.fetch(outgoing_block.id)

        # The block should have as many output arguments as the
        # outgoing block has input arguments.
        raise unless block_flow.out.size == outgoing_flow.in.size
      end
    end
  end
end