add pass 4, 5, 6 and let it can print basic assembly code

README.md

@@ -0,0 +1,18 @@
+# TComp
+A practice of Essential of Complication in Julia
+
+## Dependencies
+ - julia
+ - [Match.jl](github.com/JuliaServices/Match.jl)
+
+## instruction
+`./src/TComp.jl [.tc file]`
+
+the output assembly code is `./a.c` in AT&T assembly langauge.
+
+to make it executable, please use `gcc`: `gcc ./a.c -o output.out`
+
+the example `.tc` file is in `./test`
+
+## Known issues
+ - parser for a + b + c .. and a * b * c

misc/vertexcoloring.jl

@@ -0,0 +1,59 @@
+graph = [['a', 'b'], ['b', 'c'], ['e', 'd'], ['e', 'a'], ['a', 'c'], ['b','e'], ['e','c']]
+
+
+function vertexColoring(graph)
+    notDefined = -1
+
+    function getColor(v, color)
+        if !(v in keys(color))
+            return -1
+        else
+            return color[v]
+        end
+    end
+
+    vertices = Set(vcat(graph...))
+    verticesList = collect(vertices)
+
+    verticesMapping = map(x -> [x, Set()], verticesList)
+
+    adjacentNodes = Dict(verticesMapping)
+
+    for link in graph
+        a = link[1]
+        b = link[2]
+        push!(adjacentNodes[a], b)
+        push!(adjacentNodes[b], a)
+    end
+
+    sort!(verticesList, by=x -> length(adjacentNodes[x]), rev=true)
+
+    color = Dict()
+
+    println(verticesList)
+
+
+    for i in verticesList
+        i_adjacents = adjacentNodes[i]
+        println(i_adjacents)
+        i_adjacents_color_set = Set(map(x -> getColor(x, color), collect(i_adjacents)))
+        i_adjacents_color_list = sort(collect(i_adjacents_color_set))
+
+        if i_adjacents_color_list == [notDefined]
+            color[i] = 0
+        else
+            tmpId = 0
+            for i in i_adjacents_color_list
+                if tmpId == i
+                    tmpId += 1
+                end
+            end
+            color[i] = tmpId
+        end
+    end
+
+    return color
+
+end
+
+println(vertexColoring(graph))

passesInstrument.md

@@ -7,7 +7,10 @@ to hold the results of complex subexpressions.
  3. **OK** explicate_control makes the execution order of the program explicit. It converts
 the abstract syntax tree representation into a graph in which each node is a
 labeled sequence of statements and the edges are goto statements.
- 4. select_instructions handles the difference between LVar operations and x86
+ 4. **OK** select_instructions handles the difference between LVar operations and x86
 instructions. This pass converts each LVar operation to a short sequence of
 instructions that accomplishes the same task.
 assign_homes
+ 5. **OK**assign homes (register allocation)
+ 6. **OK**patch instructions
+ 7. **OK**prelude & conclusion

src/TComp.jl

@@ -13,9 +13,9 @@ inp = ARGS
 f = open(ARGS[1], "r")
 prog = read(f, String)       
 
-print(prog)
+#(prog)
 parsed = Parser.totalParse(prog)
-print(parsed)
+#print(parsed)
 tmp_var_no = 0
 
 
@@ -93,7 +93,6 @@ function explicitControlRemoveComplex(prog)
                     varNo = res[2]
                     newBind = res[1].binds
                     if newBind != []
-                        println("NEW_BIND:", newBind)
                         newResList = vcat(newResList, newBind)
                         push!(new_exp_args, last(newBind)[2][2])
                     else
@@ -103,7 +102,6 @@ function explicitControlRemoveComplex(prog)
                 end
                 push!(new_exp_body, new_exp_args)
 
-                println(newResList)
                 newBindVar = [("int", "id"), ("tmp" * string(varNo) , "id")]
                 varNo += 1 
                 newBind = [("%let", "id"), newBindVar, new_exp_body]
@@ -118,10 +116,17 @@ function explicitControlRemoveComplex(prog)
     end
     
     function splitLet(binds, exp, varNo)
+
         if exp[1] == ("%let", "id")
             res = rmComplexAux1(exp[3], varNo)
+
             binds = vcat(binds, res[1].binds)
             new_exp = res[1].body
+
+            #fix bug[("int", "id"), ("tmp1", "id")] => ("tmp1", "id")
+            if new_exp[1] == ("int", "id")
+                new_exp = new_exp[2]
+            end
             new_bind = [("%let", "id"), exp[2], new_exp]
             push!(binds, new_bind)
     
@@ -135,7 +140,17 @@ function explicitControlRemoveComplex(prog)
 
 
     raw_res =  rmComplex(prog)[1]
-    res = push!(raw_res.binds, raw_res.body)
+    raw_res_body = raw_res.body
+    #fix bug[("int", "id"), ("tmp1", "id")] => ("tmp1", "id")
+    if  raw_res_body[1] == ("int", "id")
+        raw_res_body =  [("%return", "id"), raw_res_body[2]]
+    end
+
+    if raw_res_body[2] == "int" # ("$8", "int")
+        raw_res_body =  [("%return", "id"), raw_res_body]
+    end
+
+    res = push!(raw_res.binds, raw_res_body)
     return res
 end
 
@@ -143,17 +158,30 @@ end
 function assignInstruction(inp)
     resList = []
     for i in inp
-        println(i)
         @match i begin
+            [("%return", "id"), (val, t_val)] => begin
+                if t_val == "int"
+                    val = "\$" * val
+                end
+                push!(resList, ["movq", val, "%rax"])
+            end
             [("%let", "id"), [_ty, (id, "id")],
-            [("%prime", "id"), (op, _), [(rhs, _), (lhs, _)]]] =>
+               [("%prime", "id"), (op, _), [(lhs, lhs_t), (rhs, rhs_t)]]] =>
             begin
                 instr = ""
                 ops = ["+", "-", "*", "/"]
-                instrs = ["addq", "subq", "mulq", "divq"]
+                instrs = ["addq", "subq", "imulq", "divq"]
                 opIndex = findfirst(x -> x == op, ops)
                 instr = instrs[opIndex]
 
+                if lhs_t == "int"
+                    lhs = "\$" * lhs
+                end
+
+                if rhs_t == "int"
+                    rhs = "\$" * rhs
+                end
+
                 if rhs == id
                     line1 = [instr, lhs, id]
                     push!(resList, line1)
@@ -169,13 +197,19 @@ function assignInstruction(inp)
                 #TODO　[("%call", "id"), (op, _), args] => ...
             end
 
-            [("%let", "id"), [_ty, (id, "id")], (val, _)] =>
+            [("%let", "id"), [_ty, (id, "id")], (val, t_val)] =>
             begin
+                if t_val == "int"
+                    val = "\$" * val
+                end
                 line = ["movq", val, id]
                 push!(resList, line)
             end
 
-            (c, "int") => push!(resList, [c])
+            (c, "int") => begin
+                c_modified = "\$" * c
+                push!(resList, [c_modified])
+            end
 
             (v, "id") => push!(resList, [v])
             _ => println("Error")
@@ -189,16 +223,134 @@ end
 
 emptyEnv = []
 res = uniquifyVar(parsed, emptyEnv)
-println("PASS1", res)
+#println("PASS1", res)
 res2 = explicitControlRemoveComplex(res)
 
-println("PASS2", Parser.prettyStringLisp(res2))
+#println("PASS2", Parser.prettyStringLisp(res2))
 res3 = assignInstruction(res2)
-println("PASS3", res3)
+#println("PASS3", res3)
+
+
+# PASS4 assign home
+function assignHomes(inp)
+    varRegex = r"(^[^\$%].*)"
+    res = []
+    vars = []
+    for i in inp
+        orig = i[2]
+        dest = i[3]
+        if match(varRegex, orig) != nothing # i.e. orig is a var and not a reg.
+            if !(orig in vars)
+                push!(vars, orig)
+            end
+        end 
+        if match(varRegex, dest) != nothing # i.e. dest is a var and not a reg.
+            if !(dest in vars)
+                push!(vars, dest)
+            end
+        end 
+    end
+    #println("ALL_VAR", vars)
+
+
+    varsLength = length(vars)
+    for i in inp
+        instr = i[1]
+        orig = i[2]
+        dest = i[3]
+
+        origIdx = findfirst(x -> x == orig,vars)
+        if origIdx != nothing
+            realAddressIdx = varsLength - origIdx + 1
+            realAddress = "-$(realAddressIdx * 8)(%rbp)"
+            orig = realAddress
+        end
+
+        destIdx = findfirst(x -> x == dest,vars)
+        if destIdx != nothing
+            realAddressIdx = varsLength - destIdx + 1
+            realAddress = "-$(realAddressIdx * 8)(%rbp)"
+            dest = realAddress
+        end
+
+        push!(res, [instr, orig, dest])
+    end
+    return (res, varsLength)
+end
+
+
+# PASS5 patch instruction (ensure "instr x(rbp) y(rbp)" not happened)
+function patchInstruction(inp)
+    memoryRegex = r".+[(]%rbp[)]$"
+    res = []
+
+    for i in inp
+        inst = i[1]
+        orig = i[2]
+        dest = i[3]
+        if (match(memoryRegex, orig) != nothing) & (match(memoryRegex, dest) != nothing)
+            cmd1 = ["movq", orig, "%rax"]
+            push!(res, cmd1)
+
+            cmd2 = [inst, "%rax", dest]
+            push!(res, cmd2)
+        elseif (inst == "imulq") & (match(r"^%.+", dest) == nothing)
+            cmd1 = ["movq", dest, "%rax"]
+            cmd2 = ["imulq", orig, "%rax"]
+            cmd3 = ["movq", "%rax", dest]
+            push!(res, cmd1)
+            push!(res, cmd2)
+            push!(res, cmd3)
+
+        else
+            push!(res, i)
+        end
+    end
+    return res
+end
+
+res4 = assignHomes(res3)
+res4_prog = res4[1]
+varNumber = res4[2]
+res5 = patchInstruction(res4_prog)
+#println("PASS5",res5)
+
+
+## PASS6 add prelude and conclude
+function preludeConclude(prog, varNumber)
+    rspSubqMax = varNumber * 8
+
+    body = "start:\n"
+
+    for i in prog
+        ln_cmd = ""
+        if length(i) == 3
+            ln_cmd = "\t$(i[1])\t$(i[2]), $(i[3])\n"
+            body = body * ln_cmd
+        end
+    end
+    body *= "\tjmp\tconclusion\n\n\n"
+
+    prelude = """
+    .globl	main
+    main:
+    	pushq %rbp
+    	movq	 %rsp, %rbp\n""" * "\tsubq	 \$$rspSubqMax, %rsp\n\tjmp start\n\n"
+
+    conclude = """\nconclusion:\n""" * "\taddq	 \$$rspSubqMax, %rsp\n\tpopq	%rbp\n\tretq"
+
+    assemblyProg = prelude * body * conclude
+    return assemblyProg
+end
+
+res6 = preludeConclude(res5, varNumber)
+# println("PASS6",res6) # emit assembly code
+f2 = open("./a.s", "w")
+write(f2, res6) #write the assembly code
 
 
 close(f)
-
+close(f2)
 
 
 end # module

src/parser.jl

@@ -436,12 +436,10 @@ function totalParse(prog)
     matchedList = map((x)->x.match, collect(mI))
     groupNameList = map(processKeys, collect(mI))
     zippedTokenList = collect(zip(matchedList, groupNameList))
-    print(zippedTokenList)
 
     withoutSpaces = filter((x)-> x[2] != "sp", zippedTokenList)
     initWrapped = ParserResult([], withoutSpaces)
     res = initWrapped >> body
-    println(prettyStringLisp(res))
     return res.matched
 end
 

test/prog1.tc

@@ -1,5 +1,6 @@
-int a = 12;
+int a = 10;
+int a = 13;
 int b = (12 + (0 - a));
-int a = (15 + b);
+int c = (14 + b);
 int d = 20;
-a + d((0 - a), b)
+a * 2 + ((b - c) - d)