From 89439593e4c39f80e6bb6cab71a43adfe526adaa Mon Sep 17 00:00:00 2001
From: cfreksen <casper@freksen.dk>
Date: Mon, 30 Oct 2017 00:00:56 +0100
Subject: [PATCH] Handle getelementptr.

---
 ll.py      |  7 +++++
 stepper.py | 87 ++++++++++++++++++++++++++++++++++++++++++++++++++----
 2 files changed, 89 insertions(+), 5 deletions(-)

diff --git a/ll.py b/ll.py
index 31d706a..45cbfa0 100644
--- a/ll.py
+++ b/ll.py
@@ -137,6 +137,13 @@ def insn2s(insn):
         return ('bitcast {} {} to {}'
                 .format(ty2s(insn.from_ty), oper2s(insn.oper),
                         ty2s(insn.to_ty)))
+    elif isinstance(insn, Gep):
+        return ('getelementptr {}, {} {}, {}'
+                .format(ty2s(insn.base_ty), ty2s(insn.oper_ty),
+                        oper2s(insn.oper),
+                        ', '.join('{} {}'.format(ty2s(t), oper2s(o))
+                                  for t, o in insn.steps)))
+
     elif isinstance(insn, Zext):
         return ('zext {} {} to {}'
                 .format(ty2s(insn.from_ty), oper2s(insn.oper),
diff --git a/stepper.py b/stepper.py
index 1edf7e4..4493084 100644
--- a/stepper.py
+++ b/stepper.py
@@ -162,7 +162,33 @@ def step(insns, terminator, blocks, stack_frames, ssa_env, global_env, heap,
         print('bitcast {} {} to {}'
               .format(ll.ty2s(from_ty), oper_v, ll.ty2s(to_ty)))
     elif isinstance(next_insn, ll.Gep):
-        TODO('Gep')
+        res = 0
+        base_ty = next_insn.base_ty
+        oper_ty = next_insn.oper_ty
+        oper = next_insn.oper
+        steps = next_insn.steps
+
+        actual_base_ty = ty2base_ty(base_ty, tdecs)
+        actual_oper_ty = ty2base_ty(oper_ty, tdecs)
+
+        if not isinstance(actual_oper_ty, ll.PointerType):
+            err('Type of main operand to getelementptr must be a pointer type. It was {}'
+                .format(ll.ty2s(actual_oper_ty)))
+        elif actual_base_ty != actual_oper_ty.inner_ty:
+            err(('Type of the main operand does not match the type getelementptr'
+                 ' navigates through.\n'
+                 '  Getelementptr type: {}\n'
+                 '  Operand type:       {}')
+                .format(ll.ty2s(actual_base_ty), ll.ty2s(actual_oper_ty.inner_ty)))
+        else:
+            oper_v = eval_oper(oper, ssa_env, global_env)
+            gep_res, formula = handle_gep(oper_v, actual_base_ty, steps, ssa_env,
+                                          global_env)
+            res = gep_res
+
+            # TODO
+            print('Gep formula: {}'
+                  .format(formula))
     elif isinstance(next_insn, ll.Zext):
         oper = next_insn.oper
         from_ty = next_insn.from_ty
@@ -346,6 +372,57 @@ def eval_icmp(cnd, left, right):
         return 0
 
 
+def handle_gep(starting_location, starting_type, starting_steps, ssa_env, global_env):
+    def visit(current_location, current_type, current_steps, current_formula):
+        if len(current_steps) == 0:
+            return current_location, current_formula
+        else:
+            s_ty, s_oper = current_steps[0]
+            s_oper_v = eval_oper(s_oper, ssa_env, global_env)
+            next_steps = current_steps[1:]
+            if isinstance(current_type, ll.StructType):
+                if not isinstance(s_oper, ll.Const):
+                    err('Index into struct must be a constant. It was: {}'
+                        .format(ll.oper2s(s_oper)))
+                    return current_location, current_formula + ' + ???'
+                jumps = [base_ty2size(current_type.fields[i])
+                         for i in range(s_oper_v)]
+                next_location = current_location + sum(jumps)
+                next_type = current_type.fields[s_oper_v]
+                if len(jumps) > 0:
+                    next_formula = ('{} + ({})'
+                                    .format(current_formula,
+                                            ' + '.join(map(str, jumps))))
+                else:
+                    next_formula = ('{} + 0'
+                                    .format(current_formula))
+
+                return visit(next_location, next_type, next_steps, next_formula)
+            elif isinstance(current_type, ll.ArrayType):
+                TODO('gep arrays')
+                return current_location, current_formula + ' + ???'
+            elif isinstance(current_type, ll.PointerType):
+                err(('Cannot use getelementptr to traverse pointers.'
+                     ' Use Load, and getelementptr on the result from that'
+                     ' to go through a pointer.'))
+                return current_location, current_formula + ' + ???'
+            else:
+                err('Unknown type to getelementptr on: {}'
+                    .format(ll.ty2s(current_type)))
+                return current_location, current_formula + ' + ???'
+
+    if len(starting_steps) == 0:
+        err('There must be at least one stepping argument to a getelementptr instruction')
+        return 0, ''
+    s_ty, s_oper = starting_steps[0]
+    s_oper_v = eval_oper(s_oper, ssa_env, global_env)
+    size = base_ty2size(starting_type)
+    next_location = starting_location + s_oper_v * size
+    formula = ('{} + {} * {}'
+               .format(starting_location, s_oper_v, size))
+    return visit(next_location, starting_type, starting_steps[1:], formula)
+
+
 def ty2base_ty(ty, tdecs, seen=[]):
     if isinstance(ty, ll.SimpleType):
         return ty
@@ -392,14 +469,14 @@ def gogo():
     p = parser.LLVMParser()
     p.build()
     data = r'''
-%T_tigermain = type { i64, i64 }
+%T_tigermain = type { i64, i64, {i64, i64}, i64 }
 
 define i64 @tigermain (i64 %U_mainSL_8, i64 %U_mainDummy_9) {
-    %a = alloca i64
     %t = alloca %T_tigermain
+    %a = getelementptr %T_tigermain, %T_tigermain* %t, i32 0, i32 2, i32 1
     store i64 9, i64* %a
-    %b = load i64, i64* %a
-    ret i64 %b
+    %r = load i64, i64* %a
+    ret i64 %r
 }
 '''