metasm 1.0.0 → 1.0.5

data/{lib/metasm → metasm/cpu}/mips/parse.rb

@@ -4,7 +4,7 @@
 #    Licence is LGPL, see LICENCE in the top-level directory
 
 
-require 'metasm/mips/opcodes'
+require 'metasm/cpu/mips/opcodes'
 require 'metasm/parse'
 
 module Metasm

data/{lib/metasm → metasm/cpu}/mips/render.rb

@@ -4,7 +4,7 @@
 #    Licence is LGPL, see LICENCE in the top-level directory
 
 
-require 'metasm/mips/opcodes'
+require 'metasm/cpu/mips/opcodes'
 require 'metasm/render'
 
 module Metasm

data/{lib/metasm/dalvik.rb → metasm/cpu/msp430.rb}

@@ -5,4 +5,4 @@
 
 
 require 'metasm/main'
-require 'metasm/dalvik/decode'
+require 'metasm/cpu/msp430/decode'

data/metasm/cpu/msp430/decode.rb

@@ -0,0 +1,243 @@
+#    This file is part of Metasm, the Ruby assembly manipulation suite
+#    Copyright (C) 2006-2010 Yoann GUILLOT
+#
+#    Licence is LGPL, see LICENCE in the top-level directory
+
+require 'metasm/cpu/msp430/opcodes'
+require 'metasm/decode'
+
+module Metasm
+class MSP430
+	def build_opcode_bin_mask(op)
+		op.bin_mask = 0
+		op.fields.each_key { |f|
+			op.bin_mask |= @fields_mask[f] << @fields_shift[f]
+		}
+		op.bin_mask ^= 0xffff
+	end
+
+	def build_bin_lookaside
+		lookaside = Array.new(256) { [] }
+		opcode_list.each { |op|
+			build_opcode_bin_mask op
+			b   = (op.bin >> 8) & 255
+			msk = (op.bin_mask >> 8) & 255
+
+			for i in b..(b | (255^msk))
+				lookaside[i] << op if i & msk == b & msk
+			end
+		}
+		lookaside
+	end
+
+	def decode_findopcode(edata)
+		di = DecodedInstruction.new(self)
+		val = edata.decode_imm(:u16, @endianness)
+		edata.ptr -= 2
+		di.opcode = @bin_lookaside[(val >> 8) & 0xff].find { |opcode| (val & opcode.bin_mask) == opcode.bin }
+		di if di.opcode
+	end
+
+	def decode_instr_op(edata, di)
+		before_ptr = edata.ptr
+		op = di.opcode
+		di.instruction.opname = op.name
+		val = edata.decode_imm(:u16, @endianness)
+
+		field_val = lambda{ |f|
+			(val >> @fields_shift[f]) & @fields_mask[f]
+		}
+
+		# must decode rs first
+		vals = {}
+		([:rs, :rd, :r_pc] & op.args).each { |a|
+			mod = { :rs => :as, :rd => :ad, :r_pc => :ad }[a]
+			mod = :as if mod == :ad and not op.fields[mod]	# addop_macro1 -> rs + ad
+
+			if a == :r_pc
+				r = Reg.new(0)
+			else
+				r = Reg.new(field_val[a])
+			end
+
+			w = op.props[:byte] ? 1 : 2
+
+			case field_val[mod]
+			when 0
+				if r.i == 3 and a == :rs
+					vals[a] = Expression[0]
+				else
+					vals[a] = r
+				end
+			when 1
+				if r.i == 3 and a == :rs
+					vals[a] = Expression[1]
+				else
+					imm = edata.decode_imm(:u16, @endianness)
+					r = nil if r.i == 2	# [imm]
+					vals[a] = Memref.new(r, imm, w)
+				end
+			when 2
+				if r.i == 3
+					vals[a] = Expression[2]
+				elsif r.i == 2
+					vals[a] = Expression[4]
+				else
+					vals[a] = Memref.new(r, 0, w)
+				end
+			when 3
+				if r.i == 3
+					vals[a] = Expression[-1]
+				elsif r.i == 2
+					vals[a] = Expression[8]
+				elsif r.i == 0 # pc++
+					# XXX order wrt other edata.decode_imm ?
+					vals[a] = Expression[edata.decode_imm(:u16, @endianness)]
+				else
+					vals[a] = Memref.new(r, 0, w, true)
+				end
+			end
+		}
+
+		op.args.each { |a|
+			di.instruction.args << case a
+			when :joff; Expression[2 * Expression.make_signed(field_val[a], 10)]
+			when :rs, :rd, :r_pc; vals[a]
+			else raise SyntaxError, "Internal error: invalid argument #{a} in #{op.name}"
+			end
+		}
+
+		di.bin_length += edata.ptr - before_ptr
+
+		return if edata.ptr > edata.length
+
+		di
+	end
+
+	def decode_instr_interpret(di, addr)
+		if di.opcode.props[:setip] and di.opcode.name =~ /^j/
+			delta = di.instruction.args.last.reduce
+			arg = Expression[[addr, :+, di.bin_length], :+, delta].reduce
+			di.instruction.args[-1] = Expression[arg]
+		end
+
+		di
+	end
+
+	def init_backtrace_binding
+		@backtrace_binding ||= {}
+
+		opcode_list.map { |ol| ol.name }.uniq.each { |op|
+			@backtrace_binding[op] ||= case op
+			when 'mov'; lambda { |di, a0, a1| { a0 => Expression[a1] }}
+			when 'cmp', 'test'; lambda { |di, *a| {} }	# TODO
+			when 'add', 'adc' ; lambda { |di, a0, a1| { a0 => Expression[a0, :+, a1] } }
+			when 'sub', 'sbc';  lambda { |di, a0, a1| { a0 => Expression[a0, :-, a1] } }
+			when 'and'; lambda { |di, a0, a1| { a0 => Expression[a0, :&, a1] } }
+			when 'or';  lambda { |di, a0, a1| { a0 => Expression[a0, :|, a1] } }
+			when 'xor'; lambda { |di, a0, a1| { a0 => Expression[a0, :^, a1] } }
+			when 'push'; lambda { |di, a0| { Indirection[:sp, 2] => Expression[a0],
+							 :sp => Expression[:sp, :-, 2] } }
+			when 'call'; lambda { |di, a0| { Indirection[:sp, 2] => Expression[di.next_addr],
+							 :sp => Expression[:sp, :-, 2] } }
+			when 'pop';  lambda { |di, a0| { a0 => Expression[Indirection[:sp, 2]],
+							 :sp => Expression[:sp, :+, 2] } }
+			when 'ret';  lambda { |di| { :sp => Expression[:sp, :+, 2] } }
+			when 'reti'; lambda { |di| { :sp => Expression[:sp, :+, 4] } }
+			when /^j/; lambda { |di, a0| {} }
+			end
+		}
+
+		@backtrace_binding
+	end
+
+	def get_backtrace_binding(di)
+		a = di.instruction.args.map { |arg|
+			case arg
+			when Reg; arg.symbolic
+			when Memref; arg.symbolic(di)
+			else arg
+			end
+		}
+
+		if binding = backtrace_binding[di.opcode.basename]
+			bd = binding[di, *a] || {}
+			di.instruction.args.grep(Memref).each { |m|
+				next unless r = m.base and m.postincr
+				r = m.base.symbolic
+				bd[r] ||= Expression[r, :+, m.size]
+			}
+			bd
+		else
+			puts "unhandled instruction to backtrace: #{di}" if $VERBOSE
+			{ :incomplete_binding => Expression[1] }
+		end
+	end
+
+	def get_xrefs_x(dasm, di)
+		return [] if not di.opcode.props[:setip]
+
+		case di.instruction.opname
+		when 'ret'
+			return [Indirection[:sp, 2, di.address]]
+		when 'reti'
+			return [Indirection[[:sp, :+, 2], 2, di.address]]
+		end
+
+		# XXX add pc, 42 ?
+		val = di.instruction.args[0]
+		case val
+		when Reg; val = val.symbolic
+		when Memref; val = val.symbolic(di)
+		end
+
+		[Expression[val]]
+	end
+
+	def backtrace_is_function_return(expr, di=nil)
+		expr = Expression[expr].reduce_rec
+		expr.kind_of?(Indirection) and expr.len == 2 and expr.target == Expression[:sp]
+	end
+
+	# updates the function backtrace_binding
+	# if the function is big and no specific register is given, do nothing (the binding will be lazily updated later, on demand)
+	def backtrace_update_function_binding(dasm, faddr, f, retaddrlist, *wantregs)
+		b = f.backtrace_binding
+
+		bt_val = lambda { |r|
+			next if not retaddrlist
+			b[r] = Expression::Unknown
+			bt = []
+			retaddrlist.each { |retaddr|
+				bt |= dasm.backtrace(Expression[r], retaddr, :include_start => true,
+					     :snapshot_addr => faddr, :origin => retaddr)
+			}
+			if bt.length != 1
+				b[r] = Expression::Unknown
+			else
+				b[r] = bt.first
+			end
+		}
+
+		if not wantregs.empty?
+			wantregs.each(&bt_val)
+		else
+			bt_val[:sp]
+		end
+
+		b
+	end
+
+	def replace_instr_arg_immediate(i, old, new)
+		i.args.map! { |a|
+			case a
+			when Expression; a == old ? new : Expression[a.bind(old => new).reduce]
+			when Memref
+				a.base = (a.base == old ? new : Expression[a.base.bind(old => new).reduce]) if a.base.kind_of?(Expression)
+				a
+			else a
+			end
+		}
+	end
+end
+end

data/metasm/cpu/msp430/main.rb

@@ -0,0 +1,62 @@
+#    This file is part of Metasm, the Ruby assembly manipulation suite
+#    Copyright (C) 2006-2010 Yoann GUILLOT
+#
+#    Licence is LGPL, see LICENCE in the top-level directory
+
+require 'metasm/main'
+
+module Metasm
+
+class MSP430 < CPU
+	def initialize(e = :little)
+		super()
+		@endianness = e
+		@size = 16
+	end
+
+	class Reg
+		include Renderable
+		Sym = (4..15).inject(0 => :pc, 1 => :sp, 2 => :flags, 3 => :rzero) { |h, i| h.update i => "r#{i}".to_sym }
+
+		attr_accessor :i
+		def initialize(i) ; @i = i end
+		def symbolic(di=nil) ; Sym[@i] end
+		def render ; [Sym[@i].to_s] end
+		def ==(o) ; o.class == self.class and o.i == @i end
+	end
+
+	class Memref
+		attr_accessor :base, :offset, :size, :postincr
+
+		def initialize(base, offset = 0, size = nil, postincr = false)
+			@base = base
+			@offset = Expression[offset]
+			@size = size
+			@postincr = postincr
+		end
+
+		def symbolic(di=nil)
+			r = @base.symbolic if @base
+			e = Expression[r, :+, @offset].reduce
+			Indirection[e, (@size || 1), (di.address if di)]
+		end
+
+		include Renderable
+
+		def render
+			b = @base
+			b = @base.to_s + '++' if @base and @postincr
+			p = Expression[b, :+, @offset].reduce
+			Indirection[p, @size].render
+		end
+	end
+
+	def init_opcode_list
+		init
+	end
+
+	def dbg_register_list
+		@dbg_register_list ||= Reg::Sym.sort.transpose.last
+	end
+end
+end

data/metasm/cpu/msp430/opcodes.rb

@@ -0,0 +1,101 @@
+#    This file is part of Metasm, the Ruby assembly manipulation suite
+#    Copyright (C) 2006-2010 Yoann GUILLOT
+#
+#    Licence is LGPL, see LICENCE in the top-level directory
+
+require 'metasm/cpu/msp430/main'
+
+module Metasm
+class MSP430
+	def addop(name, bin, *args)
+		o = Opcode.new name, bin
+
+		args.each { |a|
+			o.args << a if @valid_args[a]
+			o.props[a] = true if @valid_props[a]
+			o.fields[a] = [@fields_mask[a], @fields_shift[a]] if @fields_mask[a]
+		}
+
+		@opcode_list << o
+	end
+
+	def init
+		@opcode_list = []
+
+		@fields_mask = {
+			:as => 3,	# adressing mode
+			:ad => 1,	# adressing mode
+			:rd => 0xf,
+			:rs => 0xf,
+			:joff => 0x3ff,	# signed offset for jumps
+		}
+		@fields_shift = {
+			:as => 4,
+			:ad => 7,
+			:rd => 0,
+			:rs => 8,
+			:joff => 0,
+		}
+		@valid_args = { :r_pc => true, :rd => true, :rs => true, :joff => true }
+		@valid_props = { :setip => true, :stopexec => true, :saveip => true, :byte => true }
+
+		# https://en.wikipedia.org/wiki/TI_MSP430
+
+		addop_macro1 'rrc',  0, :byte
+		addop_macro1 'swpb', 1
+		addop_macro1 'rra',  2, :byte
+		addop_macro1 'sxt',  3
+		addop_macro1 'push', 4, :byte
+		addop_macro1 'call', 5, :setip, :stopexec, :saveip
+
+		addop 'reti', 0b000100_110_0000000
+
+		addop_macro2 'jnz', 0
+		addop_macro2 'jz',  1
+		addop_macro2 'jnc', 2
+		addop_macro2 'jc',  3
+		addop_macro2 'jb',  4	# 'jn' jump if negative => jl unsigned ?
+		addop_macro2 'jge', 5
+		addop_macro2 'jl',  6
+		addop_macro2 'jmp', 7, :stopexec
+
+		addop 'ret', 0x4130, :setip, :stopexec	# mov pc, [sp++]
+		addop 'pop', 0x4130, :rd, :ad		# mov rd, [sp++]
+
+		addop_macro3 'mov', 4
+		addop_macro3 'add', 5
+		addop_macro3 'adc', 6	# 'addc'
+		addop_macro3 'sbc', 7
+		addop_macro3 'sub', 8
+		addop_macro3 'cmp', 9
+		addop_macro3 'dadd',10	# decimal add with carry
+		addop_macro3 'test',11	# 'bit'
+		addop_macro3 'andn',12	# 'bic'
+		addop_macro3 'or',  13	# 'bis'
+		addop_macro3 'xor', 14
+		addop_macro3 'and', 15
+	end
+
+	def addop_macro1(name, bin, *props)
+		if props.delete :byte
+			addop_byte name, (0b000100 << 10) | (bin << 7), :as, :rd, *props
+		else
+			addop name, (0b000100 << 10) | (bin << 7), :as, :rd, *props
+		end
+	end
+
+	def addop_macro2(name, bin, *props)
+		addop name, (0b001 << 13) | (bin << 10), :joff, :setip, *props
+	end
+
+	def addop_macro3(name, bin, *props)
+		addop_byte name, (bin << 12), :r_pc, :ad, :as, :rs, :setip, :stopexec	# dst == pc
+		addop_byte name, (bin << 12), :rd, :ad, :as, :rs
+	end
+
+	def addop_byte(name, bin, *props)
+		addop name, bin, *props
+		addop name + '.b', bin | (1 << 6), :byte, *props
+	end
+end
+end

data/metasm/cpu/openrisc.rb

@@ -0,0 +1,11 @@
+#    This file is part of Metasm, the Ruby assembly manipulation suite
+#    Copyright (C) 2006-2009 Yoann GUILLOT
+#
+#    Licence is LGPL, see LICENCE in the top-level directory
+
+
+require 'metasm/main'
+require 'metasm/cpu/openrisc/decode'
+require 'metasm/cpu/openrisc/render'
+require 'metasm/cpu/openrisc/debug'
+require 'metasm/cpu/openrisc/decompile'