module Core class Kernel #there are no Kernel instances, only class methods. # We use this module syntax to avoid the (ugly) self (also eases searching). module ClassMethods def main_start block #TODO extract args into array of strings Vm::RegisterMachine.instance.main_start block block end def main_exit block # Machine.exit mov r7 , 0 + swi 0 Vm::RegisterMachine.instance.main_exit block block end def function_entry block , f_name Vm::RegisterMachine.instance.function_entry block , f_name end def function_exit block , f_name Vm::RegisterMachine.instance.function_exit block , f_name end #TODO this is in the wrong place. It is a function that returns a function object # while all other methods add their code into some block. --> kernel def putstring context , string = Vm::Integer , length = Vm::Integer function = Vm::Function.new(:putstring , [string , length ] , string) block = function.body # should be another level of indirection, ie write(io,str) ret = Vm::RegisterMachine.instance.write_stdout(block) function.set_return ret function end def putint context putint_function = Vm::Function.new(:putint , Vm::Integer , [] , Vm::Integer ) buffer = Vm::StringConstant.new(" ") # create a buffer context.object_space.add_object buffer # and save it (function local variable: a no no) int = putint_function.args.first moved_int = putint_function.new_local utoa = context.object_space.get_or_create_class(:Object).get_or_create_function(:utoa) b = putint_function.body b.mov( moved_int , int ) #move arg up #b.a buffer => int # string to write to b.add( int , buffer ,nil ) # string to write to b.a int + (buffer.length-3) => int b.call( utoa ) # And now we "just" have to print it, using the write_stdout b.add( int , buffer , nil ) # string to write to b.mov( moved_int , buffer.length ) Vm::RegisterMachine.instance.write_stdout(putint_function.body) putint_function end # The conversion to base10 is quite a bit more complicated than i thought. The bulk of it is in div10 # We set up variables, do the devision and write the result to the string # then check if were done and recurse if neccessary # As we write before we recurse (save a push) we write the number backwards # arguments: string address , integer def utoa context utoa_function = Vm::Function.new(:utoa , Vm::Integer , [ Vm::Integer ] , Vm::Integer ) str_addr = utoa_function.args[0] number = utoa_function.args[1] remainder = utoa_function.new_local Vm::RegisterMachine.instance.div10( utoa_function.body , number , remainder ) # make char out of digit (by using ascii encoding) 48 == "0" b = utoa_function.body.scope binding b.remainder = remainder + 48 b.strb( remainder, str_addr ) b.sub( str_addr, str_addr , 1 ) b.cmp( number , 0 ) b.callne( utoa_function ) return utoa_function end # testing method, hand coded fibo, expects arg in 1 # result comes in 7 # a hand coded version of the fibonachi numbers # not my hand off course, found in the net from a basic introduction def fibo context fibo_function = Vm::Function.new(:fibo , Vm::Integer , [] , Vm::Integer ) result = Vm::Integer.new(7) int = fibo_function.args[0] count = fibo_function.new_local f1 = fibo_function.new_local f2 = fibo_function.new_local b = fibo_function.body.scope binding b.a int == 1 b.mov( result, int , condition_code: :le) b.mov( :pc , :lr , condition_code: :le) b.push [ count , f1 , f2 , :lr] b.f1 = 1 b.f2 = 0 b.count = int - 2 l = fibo_function.body.new_block("loop").scope binding l.f1 = f1 + f2 l.f2 = f1 - f2 l.count = (count - 1).set_update_status l.bpl( l ) l.mov( result , f1 ) fibo_function.set_return result l.pop [ count , f1 , f2 , :pc] fibo_function end end extend ClassMethods end end