|Version 73 (modified by simonpj, 18 months ago) (diff)|
The stack-layout phase decides where to spill variables. The important goals are to avoid memory traffic and to minimize the size of the stack frame. Both of these goals are accomplished by reusing stack slots.
Representing Stack Slots
For each stack slot, we introduce a new name, then treat the name as the addressing expression for the slot. At the end of the pipeline, we choose a stack layout, then replace each stack slot with its offset from the stack pointer. The benefit is that we break the phase-ordering problem: any phase of the compiler can name a stack slot.
For example, for a variable x, the expression SS(x) is the address of the stack slot where we can spill x. (I don't think we output any C-- that uses SS anymore, but the new code generator marks its stack slots prior to layout with young<k> + 4, etc. -- Edward) The stack is assumed to grow down, and we assume that the address SS(x) points to the old end of the slot. Therefore, to address the low address of a 4-byte slot, we would use the expression SS(x + 4). And we would spill x using the following instruction:
m[SS(x + 4)] := x;
where m[e] refers to an address e in memory.
But what about parameter passing? We use a similar technique, but this time we describe the slot for each location as an offset within the area where the parameters are passed. For example, we lower a function call
x, y = f(a, b, c);
into approximately the following C--:
sp := SS(k + 16); m[SS(k + 4)] := k_info_table; m[SS(k + 8)] := a; m[SS(k + 12)] := b; m[SS(k + 16)] := c; call f returns to k; k: // on entry to k, sp == stack<k+12> x := m[SS(k + 8)] y := m[SS(k + 12)]
We use the following types to represent stack slots and parameter-passing areas:
data Area = RegSlot LocalReg | CallArea AreaId deriving (Eq, Ord) data AreaId = Old | Young BlockId deriving (Eq, Ord) data CmmExpr = CmmLit CmmLit ... | CmmStackSlot Area Int deriving Eq
An Area represents space on the stack; it may use either the RegSlot constructor to represent a single stack slot for a register or the CallArea constructor to represent parameters passed to/from a function call/return. In a young CallArea, the BlockId is the label of the function call's continuation, and it passes parameters to the call.
Area layout and addressing
- Each Area grows down, towards lower machine addresses.
- Offsets are always positive. The low-offset end is also called the "old end" of the area, the high-offset end is also called the "young end".
- Offset 0 (if we allowed it) would address the byte one beyond the high-address end of the Area.
- Larger offsets (from the beginning of the Area) correspond to lower machine addresses.
- Hence, to address a 4-byte object at the old end of Area a, we use the offset +4, thus (CmmStackSlot a 4).
The Old call area is the initial state of the stack on entry to the function (the overflow parameters and the return address) as well as any arguments that will be passed to a tail call. Note that RegSlot areas are very small (since they only need to store a single register), while CallArea are contiguous chunks of arguments.
To name a specific location on the stack, we represent its address with a new kind of CmmExpr: the CmmStackSlot.
A CmmStackSlot is just an integer offset into an Area.
Notice that a CmmStackSlot is an address, so we can say
Sp = SS(a+4)
to make Sp point to a particular stack slot. Use a CmmLoad to load from the stack slot.
The following figure shows the layout of a CallArea for both the outgoing parameters (function call) and incoming results (continuation after returning from the function call). Note that the incoming and outgoing parameters may be different, and they may overlap.
A RegSlot is laid out in the same fashion, with the offset 0 pointing off the high byte of the stack slot. To address an 8-byte double-word, we would use the offset 8. To address only the high word of the same stack slot, we would use the offset 4.
Currently, the intermediate code does not explicitly use a virtual frame pointer, but when we talk about offsets into the stack, we implicitly assume that there is a virtual frame pointer that points just off the oldest byte of the return address on entry to the procedures. Therefore, on entry to the procedure, the offset of the (4-byte) return address is 4.
Laying out the stack
The business of the stack-layout pass is to construct a mapping (fixed across a single procedure)
Area |-> VirtualOffset
which assigns a virtual stack slot (i.e. offset in bytes, relative to the virtual frame pointer) to each Area.
A naive approach to laying out the stack would be to give each variable its own stack slot for spilling, and allocate only the ends of the stack frame for parameter-passing areas. But this approach misses two opportunities for optimization:
- Stack slots can be reused by variables that are never on the stack at the same time
- If a function returns a variable on the stack, we might be able to use the return location as the variable's stack slot.
As it turns out, it is quite common in GHC that the first definition of a variable comes when its value is returned from a function call. If the value is returned on the stack, then an important optimization is to avoid copying that value to some other location on the stack. How is that achieved? By making sure the location where the value is returned is also its spill slot.
A greedy algorithm
We rewrite the stack slots in two passes:
- Walk over the graph and choose an offset for each Area.
- Walk over the graph, keeping track of the stack pointer, and rewrite each address of a stack slot with an offset from the stack pointer. Also, insert adjustments to the stack pointer before and after proc points.
The details are in cmm/CmmProcPointZ.hs (they have not yet been committed, but will be soon - Aug 4, 2008).