diff --git a/src/runtime/mgc.go b/src/runtime/mgc.go
index 94a6e00685..8c50e9fb79 100644
--- a/src/runtime/mgc.go
+++ b/src/runtime/mgc.go
@@ -1461,14 +1461,25 @@ func gcBgMarkWorker(_p_ *p) {
 			throw("work.nwait was > work.nproc")
 		}
 
-		switch _p_.gcMarkWorkerMode {
-		default:
-			throw("gcBgMarkWorker: unexpected gcMarkWorkerMode")
-		case gcMarkWorkerDedicatedMode:
-			gcDrain(&_p_.gcw, gcDrainNoBlock|gcDrainFlushBgCredit)
-		case gcMarkWorkerFractionalMode, gcMarkWorkerIdleMode:
-			gcDrain(&_p_.gcw, gcDrainUntilPreempt|gcDrainFlushBgCredit)
-		}
+		systemstack(func() {
+			// Mark our goroutine preemptible so its stack
+			// can be scanned. This lets two mark workers
+			// scan each other (otherwise, they would
+			// deadlock). We must not modify anything on
+			// the G stack. However, stack shrinking is
+			// disabled for mark workers, so it is safe to
+			// read from the G stack.
+			casgstatus(gp, _Grunning, _Gwaiting)
+			switch _p_.gcMarkWorkerMode {
+			default:
+				throw("gcBgMarkWorker: unexpected gcMarkWorkerMode")
+			case gcMarkWorkerDedicatedMode:
+				gcDrain(&_p_.gcw, gcDrainNoBlock|gcDrainFlushBgCredit)
+			case gcMarkWorkerFractionalMode, gcMarkWorkerIdleMode:
+				gcDrain(&_p_.gcw, gcDrainUntilPreempt|gcDrainFlushBgCredit)
+			}
+			casgstatus(gp, _Gwaiting, _Grunning)
+		})
 
 		// If we are nearing the end of mark, dispose
 		// of the cache promptly. We must do this
diff --git a/src/runtime/mgcmark.go b/src/runtime/mgcmark.go
index e62e470217..9489a0a344 100644
--- a/src/runtime/mgcmark.go
+++ b/src/runtime/mgcmark.go
@@ -221,24 +221,13 @@ func markroot(gcw *gcWork, i uint32) {
 			gp.waitsince = work.tstart
 		}
 
-		if gcphase != _GCmarktermination && gp.startpc == gcBgMarkWorkerPC && readgstatus(gp) != _Gdead {
-			// GC background workers may be
-			// non-preemptible, so we may deadlock if we
-			// try to scan them during a concurrent phase.
-			// They also have tiny stacks, so just ignore
-			// them until mark termination.
-			gp.gcscandone = true
-			queueRescan(gp)
-			break
-		}
-
 		// scang must be done on the system stack in case
 		// we're trying to scan our own stack.
 		systemstack(func() {
 			// If this is a self-scan, put the user G in
 			// _Gwaiting to prevent self-deadlock. It may
-			// already be in _Gwaiting if this is mark
-			// termination.
+			// already be in _Gwaiting if this is a mark
+			// worker or we're in mark termination.
 			userG := getg().m.curg
 			selfScan := gp == userG && readgstatus(userG) == _Grunning
 			if selfScan {
diff --git a/src/runtime/stack.go b/src/runtime/stack.go
index e803dc17a0..dfc71b41c3 100644
--- a/src/runtime/stack.go
+++ b/src/runtime/stack.go
@@ -1122,6 +1122,11 @@ func shrinkstack(gp *g) {
 	if debug.gcshrinkstackoff > 0 {
 		return
 	}
+	if gp.startpc == gcBgMarkWorkerPC {
+		// We're not allowed to shrink the gcBgMarkWorker
+		// stack (see gcBgMarkWorker for explanation).
+		return
+	}
 
 	oldsize := gp.stackAlloc
 	newsize := oldsize / 2