llvm--emulator/stepper.py

import ll
import parser


def TODO(msg):
    print('TODO, not implemented yet at {}'
          .format(msg))


def err(msg):
    print('ERROR: {}'
          .format(msg))


def warn(msg):
    print('WARNING: {}'
          .format(msg))


def step(insns, terminator, blocks, stack_frames, ssa_env, global_env, memory,
         tdecs, fdecs, call_res):
    if len(insns) == 0:
        return terminate(terminator, blocks, stack_frames, ssa_env, global_env, memory)
    ssa_target, next_insn = insns[0]
    insns_rest = insns[1:]

    # TODO
    print('Evaluating {}'
          .format(ll.insn2s(next_insn)))

    res = None
    if isinstance(next_insn, ll.Binop):
        bop = next_insn.bop
        left = next_insn.left
        right = next_insn.right
        left_v = eval_oper(left, ssa_env, global_env)
        right_v = eval_oper(right, ssa_env, global_env)
        res = eval_binop(bop, left_v, right_v)

        # TODO
        print('{} {}, {}'
              .format(bop, left_v, right_v))
    elif isinstance(next_insn, ll.Icmp):
        cnd = next_insn.cnd
        left = next_insn.left
        right = next_insn.right
        left_v = eval_oper(left, ssa_env, global_env)
        right_v = eval_oper(right, ssa_env, global_env)
        res = eval_icmp(cnd, left_v, right_v)

        # TODO
        print('icmp {} {}, {}'
              .format(cnd,  left_v, right_v))
    elif isinstance(next_insn, ll.Bitcast):
        oper = next_insn.oper
        from_ty = next_insn.from_ty
        to_ty = next_insn.to_ty
        oper_v = eval_oper(oper, ssa_env, global_env)
        res = oper_v

        # TODO
        print('bitcast {} {} to {}'
              .format(ll.ty2s(from_ty), oper_v, ll.ty2s(to_ty)))
    elif isinstance(next_insn, ll.Zext):
        oper = next_insn.oper
        from_ty = next_insn.from_ty
        to_ty = next_insn.to_ty
        oper_v = eval_oper(oper, ssa_env, global_env)
        res = oper_v

        # TODO
        print('zext {} {} to {}'
              .format(ll.ty2s(from_ty), oper_v, ll.ty2s(to_ty)))
    elif isinstance(next_insn, ll.Ptrtoint):
        oper = next_insn.oper
        pointer_ty = next_insn.pointer_ty
        to_ty = next_insn.to_ty
        oper_v = eval_oper(oper, ssa_env, global_env)
        res = oper_v

        # TODO
        print('ptrtoint {}* {} to {}'
              .format(ll.ty2s(pointer_ty), oper_v, ll.ty2s(to_ty)))
    else:
        err('Unknown LLVM instruction: {}'
            .format(next_insn))

    if ssa_target is not None:
        if ssa_target in ssa_env:
            err('Cannot assign to variable twice: {}'
                .format(ssa_target))
        else:
            # TODO
            print('%{} <- {}'
                  .format(ssa_target, res))
            ssa_env[ssa_target] = res

    return insns_rest, terminator, blocks, stack_frames, ssa_env, memory, None


def terminate(terminator, blocks, stack_frames, ssa_env, global_env, memory):
    def clear_block_from_ssa_env(insns, ssa_env):
        for (id, insn) in insns:
            if id is not None and id in ssa_env:
                del ssa_env[id]

    print('Evaluating {}'
          .format(ll.terminator2s(terminator)))

    if isinstance(terminator, ll.Ret):
        oper = terminator.oper
        if oper is None:
            oper_v = None
        else:
            oper_v = eval_oper(oper, ssa_env, global_env)

        # TODO
        print('Returning {}'
              .format(oper_v))
        if len(stack_frames) == 0:
            new_insns = []
            new_terminator = None
            new_blocks = {}
            new_ssa_env = ssa_env
            new_stack_frames = []
        else:
            new_insns, new_terminator, new_blocks, new_ssa_env = stack_frames[0]
            new_stack_frames = stack_frames[1:]
        return (new_insns, new_terminator, new_blocks, new_stack_frames,
                new_ssa_env, memory, oper_v)
    elif isinstance(terminator, ll.Br):
        label = terminator.label
        next_block = blocks[label]
        new_insns = next_block.insns
        new_terminator = next_block.terminator

        # TODO: Might need to find a better solution as we will ignore
        # multiple assignments, if they are spread over multiple
        # blocks.
        clear_block_from_ssa_env(new_insns, ssa_env)

        # TODO
        print('Jumping unconditionally to {}'
              .format(label))

        return (new_insns, new_terminator, blocks, stack_frames,
                ssa_env, memory, None)
    elif isinstance(terminator, ll.Cbr):
        ty = terminator.ty
        if ty != ll.SimpleType.I1:
            warn('Branching based on value of type {}. You ought to branch on {}'
                 .format(ll.ty2s(ty), ll.ty2s(ll.SimpleType.I1)))
        operand = terminator.oper
        operand_v = eval_oper(operand, ssa_env, global_env)

        if operand_v:
            label = terminator.then_label
        else:
            label = terminator.else_label

        next_block = blocks[label]
        new_insns = next_block.insns
        new_terminator = next_block.terminator
        clear_block_from_ssa_env(new_insns, ssa_env)

        # TODO
        print('Operand was {}. Branching to {}'
              .format(operand_v, label))

        return (new_insns, new_terminator, blocks, stack_frames,
                ssa_env, memory, None)
    else:
        err('Unknown LLVM terminator: {}'
            .format(terminator))


def eval_oper(operand, ssa_env, global_env):
    if isinstance(operand, ll.Null):
        return 0
    elif isinstance(operand, ll.Const):
        return operand.val
    elif isinstance(operand, ll.Gid):
        TODO('eval_oper Gid')
    elif isinstance(operand, ll.Id):
        id = operand.val
        try:
            return ssa_env[id]
        except KeyError:
            err('Unable to find %{} in environment:\n{}'
                .format(id, ssa_env))


def eval_binop(bop, left, right):
    if bop == 'add':
        return left + right
    elif bop == 'sub':
        return left - right
    elif bop == 'mul':
        return left * right
    elif bop == 'sdiv':
        return left // right
    elif bop == 'shl':
        return left << right
    elif bop == 'ashr':
        return left >> right
    elif bop == 'lshr':
        return (left >> right) % 0x10000000000000000
    elif bop == 'and':
        return left & right
    elif bop == 'or':
        return left | right
    elif bop == 'xor':
        return left ^ right
    else:
        err('Unknown LLVM Binary operator: {}'
            .format(bop))


def eval_icmp(cnd, left, right):
    if cnd == 'eq':
        return left == right
    elif cnd == 'ne':
        return left != right
    elif cnd == 'slt':
        return left < right
    elif cnd == 'sle':
        return left <= right
    elif cnd == 'sgt':
        return left > right
    elif cnd == 'sge':
        return left >= right
    else:
        err('eval_icmp: Unknown cnd: {}'
            .format(cnd))
        return 0


def gogo():
    p = parser.LLVMParser()
    p.build()
    data = r'''
define i64 @tigermain (i64 %U_mainSL_8, i64 %U_mainDummy_9) {
    %a = add i64 3, 5 ; please be 8
    %c = icmp eq i64 %a, 9
    br i1 %c, label %L1, label %L2
L1:
    %b = add i64 %a, %a
    ret i1 %c
L2:
    %d = add i64 %a, 1
    %e = add i64 10, %d
    ret i64 %e
}
'''

    print(data)
    ast = p.parse(data)

    tdecs = ast.tdecls
    fdecs = ast.fdecls
    global_env = ast.gdecls

    tigermain = ast.fdecls['tigermain']
    first_block = tigermain.body.first_block
    blocks = tigermain.body.named_blocks
    insns = first_block.insns
    terminator = first_block.terminator
    stack_frames = []
    ssa_env = {}
    # TODO: memory structure has not been decided yet
    memory = [None]
    call_res = None

    while True:
        (insns, terminator, blocks,
         stack_frames, ssa_env, memory, call_res) = step(insns, terminator, blocks,
                                                         stack_frames, ssa_env,
                                                         global_env, memory, tdecs,
                                                         fdecs, call_res)

        if terminator is None:
            print('Stepping done! Final ssa_env:\n{}'
                  .format(ssa_env))
            print('Program resulted in {}'.
                  format(call_res))
            break


if __name__ == '__main__':
    gogo()