diff --git a/compiler-rt/lib/tsan/rtl/tsan_rtl_thread.cpp b/compiler-rt/lib/tsan/rtl/tsan_rtl_thread.cpp
--- a/compiler-rt/lib/tsan/rtl/tsan_rtl_thread.cpp
+++ b/compiler-rt/lib/tsan/rtl/tsan_rtl_thread.cpp
@@ -138,6 +138,8 @@
   creation_stack_id = CurrentStackId(args->thr, args->pc);
 }
 
+extern "C" void __tsan_stack_initialization() {}
+
 struct OnStartedArgs {
   ThreadState *thr;
   uptr stk_addr;
@@ -173,9 +175,15 @@
 #endif
 
 #if !SANITIZER_GO
+  // Don't imitate stack/TLS writes for the main thread,
+  // because its initialization is synchronized with all
+  // subsequent threads anyway.
   if (tid != kMainTid) {
-    if (stk_addr && stk_size)
-      MemoryRangeImitateWrite(thr, /*pc=*/1, stk_addr, stk_size);
+    if (stk_addr && stk_size) {
+      const uptr pc = StackTrace::GetNextInstructionPc(
+          reinterpret_cast<uptr>(__tsan_stack_initialization));
+      MemoryRangeImitateWrite(thr, pc, stk_addr, stk_size);
+    }
 
     if (tls_addr && tls_size)
       ImitateTlsWrite(thr, tls_addr, tls_size);
diff --git a/compiler-rt/test/tsan/stack_race3.cpp b/compiler-rt/test/tsan/stack_race3.cpp
new file mode 100644
--- /dev/null
+++ b/compiler-rt/test/tsan/stack_race3.cpp
@@ -0,0 +1,34 @@
+// RUN: %clangxx_tsan -O1 %s -o %t && %deflake %run %t | FileCheck %s
+
+// Race with initial stack initialization:
+// there is no explicit second write,
+// but the stack variable is published unsafely.
+#include "test.h"
+
+long *P;
+
+void *Thread(void *a) {
+  long X;
+  __atomic_store_n(&P, &X, __ATOMIC_RELAXED);
+  barrier_wait(&barrier);
+  barrier_wait(&barrier);
+  return 0;
+}
+
+int main() {
+  barrier_init(&barrier, 2);
+  pthread_t t;
+  pthread_create(&t, NULL, Thread, NULL);
+  barrier_wait(&barrier);
+  long *p = __atomic_load_n(&P, __ATOMIC_RELAXED);
+  *p = 42;
+  barrier_wait(&barrier);
+  pthread_join(t, 0);
+}
+
+// CHECK: WARNING: ThreadSanitizer: data race
+// CHECK:   Write of size 8 at {{.*}} by main thread:
+// CHECK:     #0 main
+// CHECK:   Previous write of size 8 at {{.*}} by thread T1:
+// CHECK:     #0 __tsan_stack_initialization
+// CHECK:   Location is stack of thread T1