MapInitToVar is the inverse of VarToMapInit; it maintains a mapping from a compiler-generated init function to the map the function is initializing.
var MapInitToVar map[*ir.Func]*ir.Name
func AddKeepRelocations()
AddKeepRelocations adds a dummy "R_KEEP" relocation from each global map variable V to its associated outlined init function. These relocation ensure that if the map var itself is determined to be reachable at link time, we also mark the init function as reachable.
func AnySideEffects(n ir.Node) bool
AnySideEffects reports whether n contains any operations that could have observable side effects.
func OutlineMapInits(fn *ir.Func)
OutlineMapInits replaces global map initializers with outlined calls to separate "map init" functions (where possible and profitable), to facilitate better dead-code elimination by the linker.
func StaticLoc(n ir.Node) (name *ir.Name, offset int64, ok bool)
StaticLoc returns the static address of n, if n has one, or else nil.
func StaticName(t *types.Type) *ir.Name
StaticName returns a name backed by a (writable) static data symbol. Use readonlystaticname for read-only node.
type Entry struct {
Xoffset int64 // struct, array only
Expr ir.Node // bytes of run-time computed expressions
}
type Plan struct {
E []Entry
}
An Schedule is used to decompose assignment statements into static and dynamic initialization parts. Static initializations are handled by populating variables' linker symbol data, while dynamic initializations are accumulated to be executed in order.
type Schedule struct {
// Out is the ordered list of dynamic initialization
// statements.
Out []ir.Node
Plans map[ir.Node]*Plan
Temps map[ir.Node]*ir.Name
// contains filtered or unexported fields
}
func (s *Schedule) StaticAssign(l *ir.Name, loff int64, r ir.Node, typ *types.Type) bool
func (s *Schedule) StaticInit(n ir.Node)
StaticInit adds an initialization statement n to the schedule.