Welcome to Chapter 7 of the "Implementing a language with LLVM" tutorial. In chapters 1 through 6, we've built a very respectable, albeit simple, functional programming language. In our journey, we learned some parsing techniques, how to build and represent an AST, how to build LLVM IR, and how to optimize the resultant code as well as JIT compile it.

While Kaleidoscope is interesting as a functional language, the fact that it is functional makes it "too easy" to generate LLVM IR for it. In particular, a functional language makes it very easy to build LLVM IR directly in SSA form. Since LLVM requires that the input code be in SSA form, this is a very nice property and it is often unclear to newcomers how to generate code for an imperative language with mutable variables.

The short (and happy) summary of this chapter is that there is no need for your front-end to build SSA form: LLVM provides highly tuned and well tested support for this, though the way it works is a bit unexpected for some.

To understand why mutable variables cause complexities in SSA construction, consider this extremely simple C example:

int G, H;
int test(_Bool Condition) {
  int X;
  if (Condition)
    X = G;
  else
    X = H;
  return X;
}

In this case, we have the variable "X", whose value depends on the path executed in the program. Because there are two different possible values for X before the return instruction, a PHI node is inserted to merge the two values. The LLVM IR that we want for this example looks like this:

@G = weak global i32 0   ; type of @G is i32*
@H = weak global i32 0   ; type of @H is i32*

define i32 @test(i1 %Condition) {
entry:
	br i1 %Condition, label %cond_true, label %cond_false

cond_true:
	%X.0 = load i32* @G
	br label %cond_next

cond_false:
	%X.1 = load i32* @H
	br label %cond_next

cond_next:
	%X.2 = phi i32 [ %X.1, %cond_false ], [ %X.0, %cond_true ]
	ret i32 %X.2
}

In this example, the loads from the G and H global variables are explicit in the LLVM IR, and they live in the then/else branches of the if statement (cond_true/cond_false). In order to merge the incoming values, the X.2 phi node in the cond_next block selects the right value to use based on where control flow is coming from: if control flow comes from the cond_false block, X.2 gets the value of X.1. Alternatively, if control flow comes from cond_true, it gets the value of X.0. The intent of this chapter is not to explain the details of SSA form. For more information, see one of the many online references.

The question for this article is "who places the phi nodes when lowering assignments to mutable variables?". The issue here is that LLVM requires that its IR be in SSA form: there is no "non-ssa" mode for it. However, SSA construction requires non-trivial algorithms and data structures, so it is inconvenient and wasteful for every front-end to have to reproduce this logic.

The 'trick' here is that while LLVM does require all register values to be in SSA form, it does not require (or permit) memory objects to be in SSA form. In the example above, note that the loads from G and H are direct accesses to G and H: they are not renamed or versioned. This differs from some other compiler systems, which do try to version memory objects. In LLVM, instead of encoding dataflow analysis of memory into the LLVM IR, it is handled with Analysis Passes which are computed on demand.

With this in mind, the high-level idea is that we want to make a stack variable (which lives in memory, because it is on the stack) for each mutable object in a function. To take advantage of this trick, we need to talk about how LLVM represents stack variables.

In LLVM, all memory accesses are explicit with load/store instructions, and it is carefully designed not to have (or need) an "address-of" operator. Notice how the type of the @G/@H global variables is actually "i32*" even though the variable is defined as "i32". What this means is that @G defines space for an i32 in the global data area, but its name actually refers to the address for that space. Stack variables work the same way, except that instead of being declared with global variable definitions, they are declared with the LLVM alloca instruction:

define i32 @example() {
entry:
	%X = alloca i32           ; type of %X is i32*.
	...
	%tmp = load i32* %X       ; load the stack value %X from the stack.
	%tmp2 = add i32 %tmp, 1   ; increment it
	store i32 %tmp2, i32* %X  ; store it back
	...

This code shows an example of how you can declare and manipulate a stack variable in the LLVM IR. Stack memory allocated with the alloca instruction is fully general: you can pass the address of the stack slot to functions, you can store it in other variables, etc. In our example above, we could rewrite the example to use the alloca technique to avoid using a PHI node:

@G = weak global i32 0   ; type of @G is i32*
@H = weak global i32 0   ; type of @H is i32*

define i32 @test(i1 %Condition) {
entry:
	%X = alloca i32           ; type of %X is i32*.
	br i1 %Condition, label %cond_true, label %cond_false

cond_true:
	%X.0 = load i32* @G
        store i32 %X.0, i32* %X   ; Update X
	br label %cond_next

cond_false:
	%X.1 = load i32* @H
        store i32 %X.1, i32* %X   ; Update X
	br label %cond_next

cond_next:
	%X.2 = load i32* %X       ; Read X
	ret i32 %X.2
}

With this, we have discovered a way to handle arbitrary mutable variables without the need to create Phi nodes at all:

1. Each mutable variable becomes a stack allocation.
2. Each read of the variable becomes a load from the stack.
3. Each update of the variable becomes a store to the stack.
4. Taking the address of a variable just uses the stack address directly.

While this solution has solved our immediate problem, it introduced another one: we have now apparently introduced a lot of stack traffic for very simple and common operations, a major performance problem. Fortunately for us, the LLVM optimizer has a highly-tuned optimization pass named "mem2reg" that handles this case, promoting allocas like this into SSA registers, inserting Phi nodes as appropriate. If you run this example through the pass, for example, you'll get:

$ llvm-as < example.ll | opt -mem2reg | llvm-dis
@G = weak global i32 0
@H = weak global i32 0

define i32 @test(i1 %Condition) {
entry:
	br i1 %Condition, label %cond_true, label %cond_false

cond_true:
	%X.0 = load i32* @G
	br label %cond_next

cond_false:
	%X.1 = load i32* @H
	br label %cond_next

cond_next:
	%X.01 = phi i32 [ %X.1, %cond_false ], [ %X.0, %cond_true ]
	ret i32 %X.01
}

The mem2reg pass implements the standard "iterated dominance frontier" algorithm for constructing SSA form and has a number of optimizations that speed up (very common) degenerate cases. The mem2reg optimization pass is the answer to dealing with mutable variables, and we highly recommend that you depend on it. Note that mem2reg only works on variables in certain circumstances:

1. mem2reg is alloca-driven: it looks for allocas and if it can handle them, it promotes them. It does not apply to global variables or heap allocations.
2. mem2reg only looks for alloca instructions in the entry block of the function. Being in the entry block guarantees that the alloca is only executed once, which makes analysis simpler.
3. mem2reg only promotes allocas whose uses are direct loads and stores. If the address of the stack object is passed to a function, or if any funny pointer arithmetic is involved, the alloca will not be promoted.
4. mem2reg only works on allocas of first class values (such as pointers, scalars and vectors), and only if the array size of the allocation is 1 (or missing in the .ll file). mem2reg is not capable of promoting structs or arrays to registers. Note that the "scalarrepl" pass is more powerful and can promote structs, "unions", and arrays in many cases.

All of these properties are easy to satisfy for most imperative languages, and we'll illustrate it below with Kaleidoscope. The final question you may be asking is: should I bother with this nonsense for my front-end? Wouldn't it be better if I just did SSA construction directly, avoiding use of the mem2reg optimization pass? In short, we strongly recommend that you use this technique for building SSA form, unless there is an extremely good reason not to. Using this technique is:

• Proven and well tested: llvm-gcc and clang both use this technique for local mutable variables. As such, the most common clients of LLVM are using this to handle a bulk of their variables. You can be sure that bugs are found fast and fixed early.
• Extremely Fast: mem2reg has a number of special cases that make it fast in common cases as well as fully general. For example, it has fast-paths for variables that are only used in a single block, variables that only have one assignment point, good heuristics to avoid insertion of unneeded phi nodes, etc.
• Needed for debug info generation: Debug information in LLVM relies on having the address of the variable exposed so that debug info can be attached to it. This technique dovetails very naturally with this style of debug info.

If nothing else, this makes it much easier to get your front-end up and running, and is very simple to implement. Lets extend Kaleidoscope with mutable variables now!

Now that we know the sort of problem we want to tackle, lets see what this looks like in the context of our little Kaleidoscope language. We're going to add two features:

1. The ability to mutate variables with the '=' operator.
2. The ability to define new variables.

While the first item is really what this is about, we only have variables for incoming arguments as well as for induction variables, and redefining those only goes so far :). Also, the ability to define new variables is a useful thing regardless of whether you will be mutating them. Here's a motivating example that shows how we could use these:

# Define ':' for sequencing: as a low-precedence operator that ignores operands
# and just returns the RHS.
def binary : 1 (x y) y;

# Recursive fib, we could do this before.
def fib(x)
  if (x < 3) then
    1
  else
    fib(x-1)+fib(x-2);

# Iterative fib.
def fibi(x)
  var a = 1, b = 1, c in
  (for i = 3, i < x in 
     c = a + b :
     a = b :
     b = c) :
  b;

# Call it. 
fibi(10);

In order to mutate variables, we have to change our existing variables to use the "alloca trick". Once we have that, we'll add our new operator, then extend Kaleidoscope to support new variable definitions.

The symbol table in Kaleidoscope is managed at code generation time by the 'NamedValues' map. This map currently keeps track of the LLVM "Value*" that holds the double value for the named variable. In order to support mutation, we need to change this slightly, so that it NamedValues holds the memory location of the variable in question. Note that this change is a refactoring: it changes the structure of the code, but does not (by itself) change the behavior of the compiler. All of these changes are isolated in the Kaleidoscope code generator.

At this point in Kaleidoscope's development, it only supports variables for two things: incoming arguments to functions and the induction variable of 'for' loops. For consistency, we'll allow mutation of these variables in addition to other user-defined variables. This means that these will both need memory locations.

To start our transformation of Kaleidoscope, we'll change the NamedValues map so that it maps to AllocaInst* instead of Value*. Once we do this, the C++ compiler will tell us what parts of the code we need to update:

static std::map<std::string, AllocaInst*> NamedValues;

Also, since we will need to create these alloca's, we'll use a helper function that ensures that the allocas are created in the entry block of the function:

/// CreateEntryBlockAlloca - Create an alloca instruction in the entry block of
/// the function.  This is used for mutable variables etc.
static AllocaInst *CreateEntryBlockAlloca(Function *TheFunction,
                                          const std::string &VarName) {
  IRBuilder<> TmpB(&TheFunction->getEntryBlock(),
                 TheFunction->getEntryBlock().begin());
  return TmpB.CreateAlloca(Type::getDoubleTy(getGlobalContext()), 0,
                           VarName.c_str());
}

This funny looking code creates an IRBuilder object that is pointing at the first instruction (.begin()) of the entry block. It then creates an alloca with the expected name and returns it. Because all values in Kaleidoscope are doubles, there is no need to pass in a type to use.

With this in place, the first functionality change we want to make is to variable references. In our new scheme, variables live on the stack, so code generating a reference to them actually needs to produce a load from the stack slot:

Value *VariableExprAST::Codegen() {
  // Look this variable up in the function.
  Value *V = NamedValues[Name];
  if (V == 0) return ErrorV("Unknown variable name");

  // Load the value.
  return Builder.CreateLoad(V, Name.c_str());
}

As you can see, this is pretty straightforward. Now we need to update the things that define the variables to set up the alloca. We'll start with ForExprAST::Codegen (see the full code listing for the unabridged code):

  Function *TheFunction = Builder.GetInsertBlock()->getParent();

  // Create an alloca for the variable in the entry block.
  AllocaInst *Alloca = CreateEntryBlockAlloca(TheFunction, VarName);
  
    // Emit the start code first, without 'variable' in scope.
  Value *StartVal = Start->Codegen();
  if (StartVal == 0) return 0;
  
  // Store the value into the alloca.
  Builder.CreateStore(StartVal, Alloca);
  ...

  // Compute the end condition.
  Value *EndCond = End->Codegen();
  if (EndCond == 0) return EndCond;
  
  // Reload, increment, and restore the alloca.  This handles the case where
  // the body of the loop mutates the variable.
  Value *CurVar = Builder.CreateLoad(Alloca);
  Value *NextVar = Builder.CreateFAdd(uilderles or heap allocations.

mem2reg only looks for alloca instructions in the entry block of the
function.  Being in the entry block guarantees that the alloca is only executed
once, which makes analysis simpler.

mem2reg only promotes allocas whose uses are direct loads and stores.  If
the address of the stack object is passed to a function, or if any funny pointer
arithmetic is involved, the alloca will not be promoted.

mem2reg only works on allocas of first class 
values (such as pointers, scalars and vectors), and only if the array size
of the allocation is 1 (or missing in the .ll file).  mem2reg is not capable of
promoting structs or arrays to registers.  Note that the "scalarrepl" pass is
more powerful and can promote structs, "unions", and arrays in many cases.




All of these properties are easy to satisfy for most imperative languages, and
we'll illustrate it below with Kaleidoscope.  The final question you may be
asking is: should I bother with this nonsense for my front-end?  Wouldn't it be
better if I just did SSA construction directly, avoiding use of the mem2reg
optimization pass?  In short, we strongly recommend that you use this technique
for building SSA form, unless there is an extremely good reason not to.  Using
this technique is:


Proven and well tested: llvm-gcc and clang both use this technique for local
mutable variables.  As such, the most common clients of LLVM are using this to
handle a bulk of their variables.  You can be sure that bugs are found fast and
fixed early.

Extremely Fast: mem2reg has a number of special cases that make it fast in
common cases as well as fully general.  For example, it has fast-paths for
variables that are only used in a single block, variables that only have one
assignment point, good heuristics to avoid insertion of unneeded phi nodes, etc.


Needed for debug info generation: 
Debug information in LLVM relies on having the address of the variable
exposed so that debug info can be attached to it.  This technique dovetails 
very naturally with this style of debug info.


If nothing else, this makes it much easier to get your front-end up and 
running, and is very simple to implement.  Lets extend Kaleidoscope with mutable
variables now!

# Define ':' for sequencing: as a low-precedence operator that ignores operands
# and just returns the RHS.
def binary : 1 (x y) y;

# Recursive fib, we could do this before.
def fib(x)
  if (x < 3) then
    1
  else
    fib(x-1)+fib(x-2);

# Iterative fib.
def fibi(x)
  var a = 1, b = 1, c in
  (for i = 3, i < x in 
     c = a + b :
     a = b :
     b = c) :
  b;

# Call it. 
fibi(10);

static std::map<std::string, AllocaInst*> NamedValues;

/// CreateEntryBlockAlloca - Create an alloca instruction in the entry block of
/// the function.  This is used for mutable variables etc.
static AllocaInst *CreateEntryBlockAlloca(Function *TheFunction,
                                          const std::string &VarName) {
  IRBuilder<> TmpB(&TheFunction->getEntryBlock(),
                 TheFunction->getEntryBlock().begin());
  return TmpB.CreateAlloca(Type::getDoubleTy(getGlobalContext()), 0,
                           VarName.c_str());
}

Value *VariableExprAST::Codegen() {
  // Look this variable up in the function.
  Value *V = NamedValues[Name];
  if (V == 0) return ErrorV("Unknown variable name");

  // Load the value.
  return Builder.CreateLoad(V, Name.c_str());
}

  Function *TheFunction = Builder.GetInsertBlock()->getParent();

  // Create an alloca for the variable in the entry block.
  AllocaInst *Alloca = CreateEntryBlockAlloca(TheFunction, VarName);
  
    // Emit the start code first, without 'variable' in scope.
  Value *StartVal = Start->Codegen();
  if (StartVal == 0) return 0;
  
  // Store the value into the alloca.
  Builder.CreateStore(StartVal, Alloca);
  ...

  // Compute the end condition.
  Value *EndCond = End->Codegen();
  if (EndCond == 0) return EndCond;
  
  // Reload, increment, and restore the alloca.  This handles the case where
  // the body of the loop mutates the variable.
  Value *CurVar = Builder.CreateLoad(Alloca);
  Value *NextVar = Builder.CreateFAdd(uilderles or heap allocations.

mem2reg only looks for alloca instructions in the entry block of the
function.  Being in the entry block guarantees that the alloca is only executed
once, which makes analysis simpler.

mem2reg only promotes allocas whose uses are direct loads and stores.  If
the address of the stack object is passed to a function, or if any funny pointer
arithmetic is involved, the alloca will not be promoted.

mem2reg only works on allocas of first class 
values (such as pointers, scalars and vectors), and only if the array size
of the allocation is 1 (or missing in the .ll file).  mem2reg is not capable of
promoting structs or arrays to registers.  Note that the "scalarrepl" pass is
more powerful and can promote structs, "unions", and arrays in many cases.




All of these properties are easy to satisfy for most imperative languages, and
we'll illustrate it below with Kaleidoscope.  The final question you may be
asking is: should I bother with this nonsense for my front-end?  Wouldn't it be
better if I just did SSA construction directly, avoiding use of the mem2reg
optimization pass?  In short, we strongly recommend that you use this technique
for building SSA form, unless there is an extremely good reason not to.  Using
this technique is:


Proven and well tested: llvm-gcc and clang both use this technique for local
mutable variables.  As such, the most common clients of LLVM are using this to
handle a bulk of their variables.  You can be sure that bugs are found fast and
fixed early.

Extremely Fast: mem2reg has a number of special cases that make it fast in
common cases as well as fully general.  For example, it has fast-paths for
variables that are only used in a single block, variables that only have one
assignment point, good heuristics to avoid insertion of unneeded phi nodes, etc.


Needed for debug info generation: 
Debug information in LLVM relies on having the address of the variable
exposed so that debug info can be attached to it.  This technique dovetails 
very naturally with this style of debug info.


If nothing else, this makes it much easier to get your front-end up and 
running, and is very simple to implement.  Lets extend Kaleidoscope with mutable
variables now!

# Define ':' for sequencing: as a low-precedence operator that ignores operands
# and just returns the RHS.
def binary : 1 (x y) y;

# Recursive fib, we could do this before.
def fib(x)
  if (x < 3) then
    1
  else
    fib(x-1)+fib(x-2);

# Iterative fib.
def fibi(x)
  var a = 1, b = 1, c in
  (for i = 3, i < x in 
     c = a + b :
     a = b :
     b = c) :
  b;

# Call it. 
fibi(10);

static std::map<std::string, AllocaInst*> NamedValues;

/// CreateEntryBlockAlloca - Create an alloca instruction in the entry block of
/// the function.  This is used for mutable variables etc.
static AllocaInst *CreateEntryBlockAlloca(Function *TheFunction,
                                          const std::string &VarName) {
  IRBuilder<> TmpB(&TheFunction->getEntryBlock(),
                 TheFunction->getEntryBlock().begin());
  return TmpB.CreateAlloca(Type::getDoubleTy(getGlobalContext()), 0,
                           VarName.c_str());
}

Value *VariableExprAST::Codegen() {
  // Look this variable up in the function.
  Value *V = NamedValues[Name];
  if (V == 0) return ErrorV("Unknown variable name");

  // Load the value.
  return Builder.CreateLoad(V, Name.c_str());
}

  Function *TheFunction = Builder.GetInsertBlock()->getParent();

  // Create an alloca for the variable in the entry block.
  AllocaInst *Alloca = CreateEntryBlockAlloca(TheFunction, VarName);
  
    // Emit the start code first, without 'variable' in scope.
  Value *StartVal = Start->Codegen();
  if (StartVal == 0) return 0;
  
  // Store the value into the alloca.
  Builder.CreateStore(StartVal, Alloca);
  ...

  // Compute the end condition.
  Value *EndCond = End->Codegen();
  if (EndCond == 0) return EndCond;
  
  // Reload, increment, and restore the alloca.  This handles the case where
  // the body of the loop mutates the variable.
  Value *CurVar = Builder.CreateLoad(Alloca);
  Value *NextVar = Builder.CreateFAdd(uilderles or heap allocations.

mem2reg only looks for alloca instructions in the entry block of the
function.  Being in the entry block guarantees that the alloca is only executed
once, which makes analysis simpler.

mem2reg only promotes allocas whose uses are direct loads and stores.  If
the address of the stack object is passed to a function, or if any funny pointer
arithmetic is involved, the alloca will not be promoted.

mem2reg only works on allocas of first class 
values (such as pointers, scalars and vectors), and only if the array size
of the allocation is 1 (or missing in the .ll file).  mem2reg is not capable of
promoting structs or arrays to registers.  Note that the "scalarrepl" pass is
more powerful and can promote structs, "unions", and arrays in many cases.




All of these properties are easy to satisfy for most imperative languages, and
we'll illustrate it below with Kaleidoscope.  The final question you may be
asking is: should I bother with this nonsense for my front-end?  Wouldn't it be
better if I just did SSA construction directly, avoiding use of the mem2reg
optimization pass?  In short, we strongly recommend that you use this technique
for building SSA form, unless there is an extremely good reason not to.  Using
this technique is:


Proven and well tested: llvm-gcc and clang both use this technique for local
mutable variables.  As such, the most common clients of LLVM are using this to
handle a bulk of their variables.  You can be sure that bugs are found fast and
fixed early.

Extremely Fast: mem2reg has a number of special cases that make it fast in
common cases as well as fully general.  For example, it has fast-paths for
variables that are only used in a single block, variables that only have one
assignment point, good heuristics to avoid insertion of unneeded phi nodes, etc.


Needed for debug info generation: 
Debug information in LLVM relies on having the address of the variable
exposed so that debug info can be attached to it.  This technique dovetails 
very naturally with this style of debug info.


If nothing else, this makes it much easier to get your front-end up and 
running, and is very simple to implement.  Lets extend Kaleidoscope with mutable
variables now!

# Define ':' for sequencing: as a low-precedence operator that ignores operands
# and just returns the RHS.
def binary : 1 (x y) y;

# Recursive fib, we could do this before.
def fib(x)
  if (x < 3) then
    1
  else
    fib(x-1)+fib(x-2);

# Iterative fib.
def fibi(x)
  var a = 1, b = 1, c in
  (for i = 3, i < x in 
     c = a + b :
     a = b :
     b = c) :
  b;

# Call it. 
fibi(10);

static std::map<std::string, AllocaInst*> NamedValues;

/// CreateEntryBlockAlloca - Create an alloca instruction in the entry block of
/// the function.  This is used for mutable variables etc.
static AllocaInst *CreateEntryBlockAlloca(Function *TheFunction,
                                          const std::string &VarName) {
  IRBuilder<> TmpB(&TheFunction->getEntryBlock(),
                 TheFunction->getEntryBlock().begin());
  return TmpB.CreateAlloca(Type::getDoubleTy(getGlobalContext()), 0,
                           VarName.c_str());
}

Value *VariableExprAST::Codegen() {
  // Look this variable up in the function.
  Value *V = NamedValues[Name];
  if (V == 0) return ErrorV("Unknown variable name");

  // Load the value.
  return Builder.CreateLoad(V, Name.c_str());
}

  Function *TheFunction = Builder.GetInsertBlock()->getParent();

  // Create an alloca for the variable in the entry block.
  AllocaInst *Alloca = CreateEntryBlockAlloca(TheFunction, VarName);
  
    // Emit the start code first, without 'variable' in scope.
  Value *StartVal = Start->Codegen();
  if (StartVal == 0) return 0;
  
  // Store the value into the alloca.
  Builder.CreateStore(StartVal, Alloca);
  ...

  // Compute the end condition.
  Value *EndCond = End->Codegen();
  if (EndCond == 0) return EndCond;
  
  // Reload, increment, and restore the alloca.  This handles the case where
  // the body of the loop mutates the variable.
  Value *CurVar = Builder.CreateLoad(Alloca);
  Value *NextVar = Builder.CreateFAdd(uilderles or heap allocations.

mem2reg only looks for alloca instructions in the entry block of the
function.  Being in the entry block guarantees that the alloca is only executed
once, which makes analysis simpler.

mem2reg only promotes allocas whose uses are direct loads and stores.  If
the address of the stack object is passed to a function, or if any funny pointer
arithmetic is involved, the alloca will not be promoted.

mem2reg only works on allocas of first class 
values (such as pointers, scalars and vectors), and only if the array size
of the allocation is 1 (or missing in the .ll file).  mem2reg is not capable of
promoting structs or arrays to registers.  Note that the "scalarrepl" pass is
more powerful and can promote structs, "unions", and arrays in many cases.




All of these properties are easy to satisfy for most imperative languages, and
we'll illustrate it below with Kaleidoscope.  The final question you may be
asking is: should I bother with this nonsense for my front-end?  Wouldn't it be
better if I just did SSA construction directly, avoiding use of the mem2reg
optimization pass?  In short, we strongly recommend that you use this technique
for building SSA form, unless there is an extremely good reason not to.  Using
this technique is:


Proven and well tested: llvm-gcc and clang both use this technique for local
mutable variables.  As such, the most common clients of LLVM are using this to
handle a bulk of their variables.  You can be sure that bugs are found fast and
fixed early.

Extremely Fast: mem2reg has a number of special cases that make it fast in
common cases as well as fully general.  For example, it has fast-paths for
variables that are only used in a single block, variables that only have one
assignment point, good heuristics to avoid insertion of unneeded phi nodes, etc.


Needed for debug info generation: 
Debug information in LLVM relies on having the address of the variable
exposed so that debug info can be attached to it.  This technique dovetails 
very naturally with this style of debug info.


If nothing else, this makes it much easier to get your front-end up and 
running, and is very simple to implement.  Lets extend Kaleidoscope with mutable
variables now!

# Define ':' for sequencing: as a low-precedence operator that ignores operands
# and just returns the RHS.
def binary : 1 (x y) y;

# Recursive fib, we could do this before.
def fib(x)
  if (x < 3) then
    1
  else
    fib(x-1)+fib(x-2);

# Iterative fib.
def fibi(x)
  var a = 1, b = 1, c in
  (for i = 3, i < x in 
     c = a + b :
     a = b :
     b = c) :
  b;

# Call it. 
fibi(10);

static std::map<std::string, AllocaInst*> NamedValues;

/// CreateEntryBlockAlloca - Create an alloca instruction in the entry block of
/// the function.  This is used for mutable variables etc.
static AllocaInst *CreateEntryBlockAlloca(Function *TheFunction,
                                          const std::string &VarName) {
  IRBuilder<> TmpB(&TheFunction->getEntryBlock(),
                 TheFunction->getEntryBlock().begin());
  return TmpB.CreateAlloca(Type::getDoubleTy(getGlobalContext()), 0,
                           VarName.c_str());
}

Value *VariableExprAST::Codegen() {
  // Look this variable up in the function.
  Value *V = NamedValues[Name];
  if (V == 0) return ErrorV("Unknown variable name");

  // Load the value.
  return Builder.CreateLoad(V, Name.c_str());
}

  Function *TheFunction = Builder.GetInsertBlock()->getParent();

  // Create an alloca for the variable in the entry block.
  AllocaInst *Alloca = CreateEntryBlockAlloca(TheFunction, VarName);
  
    // Emit the start code first, without 'variable' in scope.
  Value *StartVal = Start->Codegen();
  if (StartVal == 0) return 0;
  
  // Store the value into the alloca.
  Builder.CreateStore(StartVal, Alloca);
  ...

  // Compute the end condition.
  Value *EndCond = End->Codegen();
  if (EndCond == 0) return EndCond;
  
  // Reload, increment, and restore the alloca.  This handles the case where
  // the body of the loop mutates the variable.
  Value *CurVar = Builder.CreateLoad(Alloca);
  Value *NextVar = Builder.CreateFAdd(uilderles or heap allocations.

mem2reg only looks for alloca instructions in the entry block of the
function.  Being in the entry block guarantees that the alloca is only executed
once, which makes analysis simpler.

mem2reg only promotes allocas whose uses are direct loads and stores.  If
the address of the stack object is passed to a function, or if any funny pointer
arithmetic is involved, the alloca will not be promoted.

mem2reg only works on allocas of first class 
values (such as pointers, scalars and vectors), and only if the array size
of the allocation is 1 (or missing in the .ll file).  mem2reg is not capable of
promoting structs or arrays to registers.  Note that the "scalarrepl" pass is
more powerful and can promote structs, "unions", and arrays in many cases.




All of these properties are easy to satisfy for most imperative languages, and
we'll illustrate it below with Kaleidoscope.  The final question you may be
asking is: should I bother with this nonsense for my front-end?  Wouldn't it be
better if I just did SSA construction directly, avoiding use of the mem2reg
optimization pass?  In short, we strongly recommend that you use this technique
for building SSA form, unless there is an extremely good reason not to.  Using
this technique is:


Proven and well tested: llvm-gcc and clang both use this technique for local
mutable variables.  As such, the most common clients of LLVM are using this to
handle a bulk of their variables.  You can be sure that bugs are found fast and
fixed early.

Extremely Fast: mem2reg has a number of special cases that make it fast in
common cases as well as fully general.  For example, it has fast-paths for
variables that are only used in a single block, variables that only have one
assignment point, good heuristics to avoid insertion of unneeded phi nodes, etc.


Needed for debug info generation: 
Debug information in LLVM relies on having the address of the variable
exposed so that debug info can be attached to it.  This technique dovetails 
very naturally with this style of debug info.


If nothing else, this makes it much easier to get your front-end up and 
running, and is very simple to implement.  Lets extend Kaleidoscope with mutable
variables now!

# Define ':' for sequencing: as a low-precedence operator that ignores operands
# and just returns the RHS.
def binary : 1 (x y) y;

# Recursive fib, we could do this before.
def fib(x)
  if (x < 3) then
    1
  else
    fib(x-1)+fib(x-2);

# Iterative fib.
def fibi(x)
  var a = 1, b = 1, c in
  (for i = 3, i < x in 
     c = a + b :
     a = b :
     b = c) :
  b;

# Call it. 
fibi(10);

static std::map<std::string, AllocaInst*> NamedValues;

/// CreateEntryBlockAlloca - Create an alloca instruction in the entry block of
/// the function.  This is used for mutable variables etc.
static AllocaInst *CreateEntryBlockAlloca(Function *TheFunction,
                                          const std::string &VarName) {
  IRBuilder<> TmpB(&TheFunction->getEntryBlock(),
                 TheFunction->getEntryBlock().begin());
  return TmpB.CreateAlloca(Type::getDoubleTy(getGlobalContext()), 0,
                           VarName.c_str());
}

Value *VariableExprAST::Codegen() {
  // Look this variable up in the function.
  Value *V = NamedValues[Name];
  if (V == 0) return ErrorV("Unknown variable name");

  // Load the value.
  return Builder.CreateLoad(V, Name.c_str());
}

  Function *TheFunction = Builder.GetInsertBlock()->getParent();

  // Create an alloca for the variable in the entry block.
  AllocaInst *Alloca = CreateEntryBlockAlloca(TheFunction, VarName);
  
    // Emit the start code first, without 'variable' in scope.
  Value *StartVal = Start->Codegen();
  if (StartVal == 0) return 0;
  
  // Store the value into the alloca.
  Builder.CreateStore(StartVal, Alloca);
  ...

  // Compute the end condition.
  Value *EndCond = End->Codegen();
  if (EndCond == 0) return EndCond;
  
  // Reload, increment, and restore the alloca.  This handles the case where
  // the body of the loop mutates the variable.
  Value *CurVar = Builder.CreateLoad(Alloca);
  Value *NextVar = Builder.CreateFAdd(uilderles or heap allocations.

mem2reg only looks for alloca instructions in the entry block of the
function.  Being in the entry block guarantees that the alloca is only executed
once, which makes analysis simpler.

mem2reg only promotes allocas whose uses are direct loads and stores.  If
the address of the stack object is passed to a function, or if any funny pointer
arithmetic is involved, the alloca will not be promoted.

mem2reg only works on allocas of first class 
values (such as pointers, scalars and vectors), and only if the array size
of the allocation is 1 (or missing in the .ll file).  mem2reg is not capable of
promoting structs or arrays to registers.  Note that the "scalarrepl" pass is
more powerful and can promote structs, "unions", and arrays in many cases.




All of these properties are easy to satisfy for most imperative languages, and
we'll illustrate it below with Kaleidoscope.  The final question you may be
asking is: should I bother with this nonsense for my front-end?  Wouldn't it be
better if I just did SSA construction directly, avoiding use of the mem2reg
optimization pass?  In short, we strongly recommend that you use this technique
for building SSA form, unless there is an extremely good reason not to.  Using
this technique is:


Proven and well tested: llvm-gcc and clang both use this technique for local
mutable variables.  As such, the most common clients of LLVM are using this to
handle a bulk of their variables.  You can be sure that bugs are found fast and
fixed early.

Extremely Fast: mem2reg has a number of special cases that make it fast in
common cases as well as fully general.  For example, it has fast-paths for
variables that are only used in a single block, variables that only have one
assignment point, good heuristics to avoid insertion of unneeded phi nodes, etc.


Needed for debug info generation: 
Debug information in LLVM relies on having the address of the variable
exposed so that debug info can be attached to it.  This technique dovetails 
very naturally with this style of debug info.


If nothing else, this makes it much easier to get your front-end up and 
running, and is very simple to implement.  Lets extend Kaleidoscope with mutable
variables now!

# Define ':' for sequencing: as a low-precedence operator that ignores operands
# and just returns the RHS.
def binary : 1 (x y) y;

# Recursive fib, we could do this before.
def fib(x)
  if (x < 3) then
    1
  else
    fib(x-1)+fib(x-2);

# Iterative fib.
def fibi(x)
  var a = 1, b = 1, c in
  (for i = 3, i < x in 
     c = a + b :
     a = b :
     b = c) :
  b;

# Call it. 
fibi(10);

static std::map<std::string, AllocaInst*> NamedValues;

/// CreateEntryBlockAlloca - Create an alloca instruction in the entry block of
/// the function.  This is used for mutable variables etc.
static AllocaInst *CreateEntryBlockAlloca(Function *TheFunction,
                                          const std::string &VarName) {
  IRBuilder<> TmpB(&TheFunction->getEntryBlock(),
                 TheFunction->getEntryBlock().begin());
  return TmpB.CreateAlloca(Type::getDoubleTy(getGlobalContext()), 0,
                           VarName.c_str());
}

Value *VariableExprAST::Codegen() {
  // Look this variable up in the function.
  Value *V = NamedValues[Name];
  if (V == 0) return ErrorV("Unknown variable name");

  // Load the value.
  return Builder.CreateLoad(V, Name.c_str());
}

  Function *TheFunction = Builder.GetInsertBlock()->getParent();

  // Create an alloca for the variable in the entry block.
  AllocaInst *Alloca = CreateEntryBlockAlloca(TheFunction, VarName);
  
    // Emit the start code first, without 'variable' in scope.
  Value *StartVal = Start->Codegen();
  if (StartVal == 0) return 0;
  
  // Store the value into the alloca.
  Builder.CreateStore(StartVal, Alloca);
  ...

  // Compute the end condition.
  Value *EndCond = End->Codegen();
  if (EndCond == 0) return EndCond;
  
  // Reload, increment, and restore the alloca.  This handles the case where
  // the body of the loop mutates the variable.
  Value *CurVar = Builder.CreateLoad(Alloca);
  Value *NextVar = Builder.CreateFAdd(uilderles or heap allocations.

mem2reg only looks for alloca instructions in the entry block of the
function.  Being in the entry block guarantees that the alloca is only executed
once, which makes analysis simpler.

mem2reg only promotes allocas whose uses are direct loads and stores.  If
the address of the stack object is passed to a function, or if any funny pointer
arithmetic is involved, the alloca will not be promoted.

mem2reg only works on allocas of first class 
values (such as pointers, scalars and vectors), and only if the array size
of the allocation is 1 (or missing in the .ll file).  mem2reg is not capable of
promoting structs or arrays to registers.  Note that the "scalarrepl" pass is
more powerful and can promote structs, "unions", and arrays in many cases.




All of these properties are easy to satisfy for most imperative languages, and
we'll illustrate it below with Kaleidoscope.  The final question you may be
asking is: should I bother with this nonsense for my front-end?  Wouldn't it be
better if I just did SSA construction directly, avoiding use of the mem2reg
optimization pass?  In short, we strongly recommend that you use this technique
for building SSA form, unless there is an extremely good reason not to.