diff --git a/src/internal/poll/fd_unix.go b/src/internal/poll/fd_unix.go
index 7d95c8d68f8977bcaa10660aeb4727af98fb8e98..2bbf60216f24dd62dc62f23fdd6f5ba122a73e78 100644
--- a/src/internal/poll/fd_unix.go
+++ b/src/internal/poll/fd_unix.go
@@ -40,6 +40,9 @@ type FD struct {
 
 	// Whether this is a file rather than a network socket.
 	isFile bool
+
+	// Whether this file has been set to blocking mode.
+	isBlocking bool
 }
 
 // Init initializes the FD. The Sysfd field should already be set.
@@ -76,18 +79,26 @@ func (fd *FD) Close() error {
 	if !fd.fdmu.increfAndClose() {
 		return errClosing(fd.isFile)
 	}
+
 	// Unblock any I/O.  Once it all unblocks and returns,
 	// so that it cannot be referring to fd.sysfd anymore,
 	// the final decref will close fd.sysfd. This should happen
 	// fairly quickly, since all the I/O is non-blocking, and any
 	// attempts to block in the pollDesc will return errClosing(fd.isFile).
 	fd.pd.evict()
+
 	// The call to decref will call destroy if there are no other
 	// references.
 	err := fd.decref()
+
 	// Wait until the descriptor is closed. If this was the only
-	// reference, it is already closed.
-	runtime_Semacquire(&fd.csema)
+	// reference, it is already closed. Only wait if the file has
+	// not been set to blocking mode, as otherwise any current I/O
+	// may be blocking, and that would block the Close.
+	if !fd.isBlocking {
+		runtime_Semacquire(&fd.csema)
+	}
+
 	return err
 }
 
@@ -100,6 +111,16 @@ func (fd *FD) Shutdown(how int) error {
 	return syscall.Shutdown(fd.Sysfd, how)
 }
 
+// SetBlocking puts the file into blocking mode.
+func (fd *FD) SetBlocking() error {
+	if err := fd.incref(); err != nil {
+		return err
+	}
+	defer fd.decref()
+	fd.isBlocking = true
+	return syscall.SetNonblock(fd.Sysfd, false)
+}
+
 // Darwin and FreeBSD can't read or write 2GB+ files at a time,
 // even on 64-bit systems.
 // The same is true of socket implementations on many systems.
diff --git a/src/net/fd_unix.go b/src/net/fd_unix.go
index 352010c17d31c34c78500f26a787ae04b4861ba4..87486578af46d71800f9500f3c0302ab25a2d251 100644
--- a/src/net/fd_unix.go
+++ b/src/net/fd_unix.go
@@ -313,7 +313,7 @@ func (fd *netFD) dup() (f *os.File, err error) {
 	// This also puts the old fd into blocking mode, meaning that
 	// I/O will block the thread instead of letting us use the epoll server.
 	// Everything will still work, just with more threads.
-	if err = syscall.SetNonblock(ns, false); err != nil {
+	if err = fd.pfd.SetBlocking(); err != nil {
 		return nil, os.NewSyscallError("setnonblock", err)
 	}
 
diff --git a/src/os/file_unix.go b/src/os/file_unix.go
index 39171cdc3514d2352b1d81edc10bc6be9060a33e..8c95f49daee32562180f7f76339660252a86bf59 100644
--- a/src/os/file_unix.go
+++ b/src/os/file_unix.go
@@ -66,7 +66,7 @@ func (f *File) Fd() uintptr {
 	// opened in blocking mode. The File will continue to work,
 	// but any blocking operation will tie up a thread.
 	if f.nonblock {
-		syscall.SetNonblock(f.pfd.Sysfd, false)
+		f.pfd.SetBlocking()
 	}
 
 	return uintptr(f.pfd.Sysfd)
diff --git a/src/os/pipe_test.go b/src/os/pipe_test.go
index 9d79d84575d40e8c2894ab535fb5b1f5afc95c33..50e11f1de5ae44a95a755a582c438b0d0fc3d384 100644
--- a/src/os/pipe_test.go
+++ b/src/os/pipe_test.go
@@ -17,6 +17,7 @@ import (
 	"runtime"
 	"strconv"
 	"strings"
+	"sync"
 	"syscall"
 	"testing"
 	"time"
@@ -220,3 +221,70 @@ func TestReadNonblockingFd(t *testing.T) {
 		t.Errorf("child process failed: %v", err)
 	}
 }
+
+// Test that we don't let a blocking read prevent a close.
+func TestCloseWithBlockingRead(t *testing.T) {
+	r, w, err := os.Pipe()
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer r.Close()
+	defer w.Close()
+
+	c1, c2 := make(chan bool), make(chan bool)
+	var wg sync.WaitGroup
+
+	wg.Add(1)
+	go func(c chan bool) {
+		defer wg.Done()
+		// Give the other goroutine a chance to enter the Read
+		// or Write call. This is sloppy but the test will
+		// pass even if we close before the read/write.
+		time.Sleep(20 * time.Millisecond)
+
+		if err := r.Close(); err != nil {
+			t.Error(err)
+		}
+		close(c)
+	}(c1)
+
+	// Calling Fd will put the file into blocking mode.
+	_ = r.Fd()
+
+	wg.Add(1)
+	go func(c chan bool) {
+		defer wg.Done()
+		var b [1]byte
+		_, err = r.Read(b[:])
+		close(c)
+		if err == nil {
+			t.Error("I/O on closed pipe unexpectedly succeeded")
+		}
+	}(c2)
+
+	for c1 != nil || c2 != nil {
+		select {
+		case <-c1:
+			c1 = nil
+			// r.Close has completed, but the blocking Read
+			// is hanging. Close the writer to unblock it.
+			w.Close()
+		case <-c2:
+			c2 = nil
+		case <-time.After(1 * time.Second):
+			switch {
+			case c1 != nil && c2 != nil:
+				t.Error("timed out waiting for Read and Close")
+				w.Close()
+			case c1 != nil:
+				t.Error("timed out waiting for Close")
+			case c2 != nil:
+				t.Error("timed out waiting for Read")
+			default:
+				t.Error("impossible case")
+			}
+		}
+	}
+
+	wg.Wait()
+}