Legend:
Library
Module
Module type
Parameter
Class
Class type
Basic block.
Logically block consists of a set of phi nodes, a sequence of definitions and a sequence of out-coming edges, aka jumps. A colloquial term for this three entities is a block element.
The order of Phi-nodes can be specified in any order, as they execute simultaneously . Definitions are stored in the order of execution. Jumps are specified in the order in which they should be taken, i.e., jmp_n is taken only after jmp_n-1 and if and only if the latter was not taken. For example, if block ends with N jumps, where each n-th jump have destination named t_n and condition c_n then it would have the semantics as per the following OCaml program:
if c_1 then jump t_1 else
if c_2 then jump t_2 else
if c_N then jump t_N else
stop
split_while blk ~f splits blk into two block: the first block holds all definitions for which f p is true and has the same tid as blk. The second block is freshly created and holds the rest definitions (if any). All successors of the blk become successors of the second block, which becomes the successor of the first block.
Note: if f def is true for all blocks, then the second block will not contain any definitions, i.e., the result would be the same as of split_bot function.
split_top blk returns two blocks, where first block shares the same tid as blk and has all $\Phi$-nodes of blk, but has only one destination, namely the second block. Second block has new tidentity, but inherits all definitions and jumps from the blk.
split_top blk returns two blocks, where first block shares the same tid as blk, has all $\Phi$-nodes and definitions from blk, but has only one destination, namely the second block. Second block has new tidentity, all jumps from the blk.
elts ~rev blk return all elements of the blk. if rev is false or left unspecified, then elements are returned in the following order: $\Phi$-nodes, defs (in normal order), jmps in the order in which they will be taken. If rev is true, the order will be the following: all jumps in the opposite order, then definitions in the opposite order, and finally $\Phi$-nodes.
val map_exp : ?skip:[ `phi | `def| `jmp ] list->t->f:(exp->exp)->t
map_exp b ~f applies function f for each expression in block b. By default function f will be applied to all values of type exp, including right hand sides of phi-nodes, definitions, jump conditions and targets. If skip parameter is specified, then terms of corresponding kind will be skipped, i.e., function f will not be applied to them.
map_elt ?phi ?def ?jmp blk applies provided functions to the terms of corresponding classes. All functions default to the identity function.
val substitute : ?skip:[ `phi | `def| `jmp ] list->t->exp->exp->t
substitute ?skip blk x y substitutes each occurrence of expression x with expression y in block blk. The substitution is performed deeply. If skip parameter is specified, then terms of corresponding kind will be left untouched.
val map_lhs : ?skip:[ `phi | `def ] list->t->f:(var->var)->t
map_lhs blk ~f applies f to every left hand side variable in def and phi subterms of blk. If skip parameter is specified, then terms of corresponding kind will be left untouched. E.g., map_lhs ~skip:[`phi] ~f:(substitute vars) will perform a substitution only on definitions (and will ignore phi-nodes)
free_vars blk returns a set of variables that occurs free in block blk. A variable is free, if it occurs unbound in the expression and there is no preceding definition of this variable in a block blk.
uses_var blk x true if variable x is in free_vars blk. If you need to call this function on several variables it is better to compute free_vars explicitly and use Set.mem function.