diff --git a/src/runtime/mprof.go b/src/runtime/mprof.go
index 09cea65bd91..c232b15424e 100644
--- a/src/runtime/mprof.go
+++ b/src/runtime/mprof.go
@@ -1131,12 +1131,15 @@ func (p *goroutineProfileStateHolder) CompareAndSwap(old, new goroutineProfileSt
 	return (*atomic.Uint32)(p).CompareAndSwap(uint32(old), uint32(new))
 }
 
-//go:linkname runtime_gcount runtime/pprof.runtime_gcount
-func runtime_gcount() int {
-	return int(gcount())
-}
-
 func goroutineProfileWithLabelsConcurrent(p []StackRecord, labels []unsafe.Pointer) (n int, ok bool) {
+	if len(p) == 0 {
+		// An empty slice is obviously too small. Return a rough
+		// allocation estimate without bothering to STW. As long as
+		// this is close, then we'll only need to STW once (on the next
+		// call).
+		return int(gcount()), false
+	}
+
 	semacquire(&goroutineProfile.sema)
 
 	ourg := getg()
diff --git a/src/runtime/pprof/pprof.go b/src/runtime/pprof/pprof.go
index 79ca11c6b4c..a8422181cca 100644
--- a/src/runtime/pprof/pprof.go
+++ b/src/runtime/pprof/pprof.go
@@ -755,9 +755,6 @@ func writeGoroutineStacks(w io.Writer) error {
 	return err
 }
 
-// runtime_gcount is defined in runtime/mprof.go
-func runtime_gcount() (n int)
-
 func writeRuntimeProfile(w io.Writer, debug int, name string, fetch func([]runtime.StackRecord, []unsafe.Pointer) (int, bool)) error {
 	// Find out how many records there are (fetch(nil)),
 	// allocate that many records, and get the data.
@@ -767,12 +764,7 @@ func writeRuntimeProfile(w io.Writer, debug int, name string, fetch func([]runti
 	// The loop should only execute one iteration in the common case.
 	var p []runtime.StackRecord
 	var labels []unsafe.Pointer
-	var n, ok = 0, false
-	if name == "goroutine" {
-		n = runtime_gcount()
-	} else {
-		n, ok = fetch(nil, nil)
-	}
+	n, ok := fetch(nil, nil)
 
 	for {
 		// Allocate room for a slightly bigger profile,