}

type partsByVarOffset struct {
	slotIDs []SlotID
	slots   []LocalSlot
}

func (a partsByVarOffset) Len() int { return len(a.slotIDs) }
func (a partsByVarOffset) Less(i, j int) bool {
	return varOffset(a.slots[a.slotIDs[i]]) < varOffset(a.slots[a.slotIDs[j]])
}
func (a partsByVarOffset) Swap(i, j int) { a.slotIDs[i], a.slotIDs[j] = a.slotIDs[j], a.slotIDs[i] }

In Go 1.23, overhead should be in the single digit percentages.

<!-- https://go.dev/issue/62737 , https://golang.org/cl/576681,  https://golang.org/cl/577615 -->
The compiler in Go 1.23 can now overlap the stack frame slots of local variables
accessed in disjoint regions of a function, which reduces stack usage
for Go applications.

<!-- https://go.dev/cl/577935 -->
For 386 and amd64, the compiler will use information from PGO to align certain

   * number of CPUS. Table slots remain empty (null) until they are
   * needed.
   *
   * A single spinlock ("busy") is used for initializing and
   * resizing the table, as well as populating slots with new Cells.
   * There is no need for a blocking lock; when the lock is not
   * available, threads try other slots (or the base).  During these

   * number of CPUS. Table slots remain empty (null) until they are
   * needed.
   *
   * A single spinlock ("busy") is used for initializing and
   * resizing the table, as well as populating slots with new Cells.
   * There is no need for a blocking lock; when the lock is not
   * available, threads try other slots (or the base).  During these

// the slot doesn't concurrently change color without the mutator
// noticing seems prohibitive.
//
// Consider the following example where the mutator writes into
// a slot and then loads the slot's mark bit while the GC thread
// writes to the slot's mark bit and then as part of scanning reads
// the slot.
//
// Initially both [slot] and [slotmark] are 0 (nil)
// Mutator thread          GC thread
// st [slot], ptr          st [slotmark], 1

   * number of CPUS. Table slots remain empty (null) until they are
   * needed.
   *
   * A single spinlock ("busy") is used for initializing and
   * resizing the table, as well as populating slots with new Cells.
   * There is no need for a blocking lock; when the lock is not
   * available, threads try other slots (or the base).  During these

   * number of CPUS. Table slots remain empty (null) until they are
   * needed.
   *
   * A single spinlock ("busy") is used for initializing and
   * resizing the table, as well as populating slots with new Cells.
   * There is no need for a blocking lock; when the lock is not
   * available, threads try other slots (or the base).  During these

//
//   - <StorageConcurrentTasksLimit>
//     Setting reduces the parallel Repository Task slots that offload or write data to object storage. The same user interface
//     setting overwrites the storage-defined setting.
//     Optional value, default 0, range: 0-unlimited (0 equals unlimited, which means the maximum configured repository task
//     slots are used for object offloading or writing)
//
//   - <KbBlockSize>

	}

	// Sort normal variables before open-coded-defer slots, so that the
	// latter are grouped together and near the top of the frame (to
	// minimize varint encoding of their varp offset).
	if a.OpenDeferSlot() != b.OpenDeferSlot() {
		return a.OpenDeferSlot()
	}

	// If a and b are both open-coded defer slots, then order them by
	// index in descending order, so they'll be laid out in the frame in

var Meta struct {
	// List contains the list of currently registered meta-data
	// blobs for the running program.
	List []CovMetaBlob

	// PkgMap records mappings from hard-coded package IDs to
	// slots in the List above.
	PkgMap map[int]int

	// Set to true if we discover a package mapping glitch.
	hardCodedListNeedsUpdating bool
}

// AddMeta is invoked during package "init" functions by the