runtime: make GC see object as allocated after it is initialized

When the GC is scanning some memory (possibly conservatively),
finding a pointer, while concurrently another goroutine is
allocating an object at the same address as the found pointer, the
GC may see the pointer before the object and/or the heap bits are
initialized. This may cause the GC to see bad pointers and
possibly crash.

To prevent this, we make it that the scanner can only see the
object as allocated after the object and the heap bits are
initialized. Currently the allocator uses freeindex to find the
next available slot, and that code is coupled with updating the
free index to a new slot past it. The scanner also uses the
freeindex to determine if an object is allocated. This is somewhat
racy. This CL makes the scanner use a different field, which is
only updated after the object initialization (and a memory
barrier).

Fixes #54596.

Change-Id: I2a57a226369926e7192c253dd0d21d3faf22297c
Reviewed-on: https://go-review.googlesource.com/c/go/+/449017
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

src/runtime/arena.go

@@ -995,6 +995,8 @@ func (h *mheap) allocUserArenaChunk() *mspan {
 	memclrNoHeapPointers(unsafe.Pointer(s.base()), s.elemsize)
 	s.needzero = 0
 
+	s.freeIndexForScan = 1
+
 	// Set up the range for allocation.
 	s.userArenaChunkFree = makeAddrRange(base, s.limit)
 	return s

src/runtime/malloc.go

@@ -1092,6 +1092,16 @@ func mallocgc(size uintptr, typ *_type, needzero bool) unsafe.Pointer {
 	// the garbage collector could follow a pointer to x,
 	// but see uninitialized memory or stale heap bits.
 	publicationBarrier()
+	// As x and the heap bits are initialized, update
+	// freeIndexForScan now so x is seen by the GC
+	// (including convervative scan) as an allocated object.
+	// While this pointer can't escape into user code as a
+	// _live_ pointer until we return, conservative scanning
+	// may find a dead pointer that happens to point into this
+	// object. Delaying this update until now ensures that
+	// conservative scanning considers this pointer dead until
+	// this point.
+	span.freeIndexForScan = span.freeindex
 
 	// Allocate black during GC.
 	// All slots hold nil so no scanning is needed.

src/runtime/mbitmap.go

@@ -191,7 +191,7 @@ func (s *mspan) nextFreeIndex() uintptr {
 // been no preemption points since ensuring this (which could allow a
 // GC transition, which would allow the state to change).
 func (s *mspan) isFree(index uintptr) bool {
-	if index < s.freeindex {
+	if index < s.freeIndexForScan {
 		return false
 	}
 	bytep, mask := s.allocBits.bitp(index)

src/runtime/mgcsweep.go

@@ -648,6 +648,7 @@ func (sl *sweepLocked) sweep(preserve bool) bool {
 
 	s.allocCount = nalloc
 	s.freeindex = 0 // reset allocation index to start of span.
+	s.freeIndexForScan = 0
 	if trace.enabled {
 		getg().m.p.ptr().traceReclaimed += uintptr(nfreed) * s.elemsize
 	}

src/runtime/mheap.go

@@ -487,6 +487,14 @@ type mspan struct {
 	speciallock           mutex         // guards specials list
 	specials              *special      // linked list of special records sorted by offset.
 	userArenaChunkFree    addrRange     // interval for managing chunk allocation
+
+	// freeIndexForScan is like freeindex, except that freeindex is
+	// used by the allocator whereas freeIndexForScan is used by the
+	// GC scanner. They are two fields so that the GC sees the object
+	// is allocated only when the object and the heap bits are
+	// initialized (see also the assignment of freeIndexForScan in
+	// mallocgc, and issue 54596).
+	freeIndexForScan uintptr
 }
 
 func (s *mspan) base() uintptr {
@@ -1386,6 +1394,7 @@ func (h *mheap) initSpan(s *mspan, typ spanAllocType, spanclass spanClass, base,
 
 		// Initialize mark and allocation structures.
 		s.freeindex = 0
+		s.freeIndexForScan = 0
 		s.allocCache = ^uint64(0) // all 1s indicating all free.
 		s.gcmarkBits = newMarkBits(s.nelems)
 		s.allocBits = newAllocBits(s.nelems)
@@ -1657,6 +1666,7 @@ func (span *mspan) init(base uintptr, npages uintptr) {
 	span.specials = nil
 	span.needzero = 0
 	span.freeindex = 0
+	span.freeIndexForScan = 0
 	span.allocBits = nil
 	span.gcmarkBits = nil
 	span.state.set(mSpanDead)