nptl: Fix tst-cancel4 sendto tests

Now that send might be implemented calling sendto syscall on Linux,
I am seeing some issue in some kernel configurations where tst-cancel4
sendto do not block as expected.

The socket used to force the syscall blocking is used with default
system configuration for buffer sending size, which might not be
suffice to force blocking.  This patch fixes it by explicit setting
buffer socket lower than the buffer size used.  It also enables sendto
cancellation tests to work in both ways (since internally send is
implemented routing to sendto on Linux kernel).

The patch also removes unrequired make rules on some archictures
for send/recv. The generic nptl Makefile already set the compiler flags
required on some architectures for correct unwinding and libc object
are not strictly required to support unwind (since pthread_cancel
requires linking against libpthread).

Checked on aarch64-linux-gnu and x86_64-linux-gnu. I also did a
sniff test with tst-cancel{4,5} on a simulated mips64-linux-gnu.

	* nptl/tst-cancel4-common.h (set_socket_buffer): New function.
	* nptl/tst-cancel4-common.c (do_test): Call set_socket_buffer
	for socketpair endpoint.
	* nptl/tst-cancel4.c (tf_send): Call set_socket_buffer and use
	WRITE_BUFFER_SIZE as buffer size for sending socket.
	(tf_sendto): Use SOCK_STREAM instead of SOCK_DGRAM and fix an
	issue on system where send is implemented with sendto syscall.
	* sysdeps/unix/sysv/linux/mips/mips64/Makefile [$(subdir) = socket]
	(CFLAGS-recv.c, CFLAGS-send.c): Remove rules.
	[$(subdir) = nptl] (CFLAGS-recv.c, CFLAGS-send.c): Likewise.
	* sysdeps/unix/sysv/linux/riscv/rv64/Makefile: Remove file.

3 files changed, 35 insertions, 34 deletions

diff --git a/nptl/tst-cancel4-common.c b/nptl/tst-cancel4-common.c
index 5bc7e44..c6eee73 100644
--- a/nptl/tst-cancel4-common.c
+++ b/nptl/tst-cancel4-common.c
@@ -20,29 +20,13 @@
 static int
 do_test (void)
 {
-  int val;
-  socklen_t len;
-
   if (socketpair (AF_UNIX, SOCK_STREAM, PF_UNIX, fds) != 0)
     {
       perror ("socketpair");
       exit (1);
     }
 
-  val = 1;
-  len = sizeof(val);
-  setsockopt (fds[1], SOL_SOCKET, SO_SNDBUF, &val, sizeof(val));
-  if (getsockopt (fds[1], SOL_SOCKET, SO_SNDBUF, &val, &len) < 0)
-    {
-      perror ("getsockopt");
-      exit (1);
-    }
-  if (val >= WRITE_BUFFER_SIZE)
-    {
-      puts ("minimum write buffer size too large");
-      exit (1);
-    }
-  setsockopt (fds[1], SOL_SOCKET, SO_SNDBUF, &val, sizeof(val));
+  set_socket_buffer (fds[1]);
 
   if (mktemp (fifoname) == NULL)
     {
diff --git a/nptl/tst-cancel4-common.h b/nptl/tst-cancel4-common.h
index a526c0c..10cc4f3 100644
--- a/nptl/tst-cancel4-common.h
+++ b/nptl/tst-cancel4-common.h
@@ -62,6 +62,20 @@ static pthread_barrier_t b2;
 
 #define WRITE_BUFFER_SIZE 16384
 
+/* Set the send buffer of socket S to 1 byte so any send operation
+   done with WRITE_BUFFER_SIZE bytes will force syscall blocking.  */
+static void
+set_socket_buffer (int s)
+{
+  int val = 1;
+  socklen_t len = sizeof(val);
+
+  TEST_VERIFY_EXIT (setsockopt (s, SOL_SOCKET, SO_SNDBUF, &val,
+		    sizeof(val)) == 0);
+  TEST_VERIFY_EXIT (getsockopt (s, SOL_SOCKET, SO_SNDBUF, &val, &len) == 0);
+  TEST_VERIFY_EXIT (val < WRITE_BUFFER_SIZE);
+}
+
 /* Cleanup handling test.  */
 static int cl_called;
 
diff --git a/nptl/tst-cancel4.c b/nptl/tst-cancel4.c
index 92a3d80..0532538 100644
--- a/nptl/tst-cancel4.c
+++ b/nptl/tst-cancel4.c
@@ -726,6 +726,8 @@ tf_send (void *arg)
   if (tempfd2 == -1)
     FAIL_EXIT1 ("socket (AF_UNIX, SOCK_STREAM, 0): %m");
 
+  set_socket_buffer (tempfd2);
+
   if (connect (tempfd2, (struct sockaddr *) &sun, sizeof (sun)) != 0)
     FAIL_EXIT1 ("connect: %m");
 
@@ -738,8 +740,7 @@ tf_send (void *arg)
 
   pthread_cleanup_push (cl, NULL);
 
-  /* Very large block, so that the send call blocks.  */
-  char mem[700000];
+  char mem[WRITE_BUFFER_SIZE];
 
   send (tempfd2, mem, arg == NULL ? sizeof (mem) : 1, 0);
 
@@ -1230,16 +1231,11 @@ tf_msync (void *arg)
 static void *
 tf_sendto (void *arg)
 {
-  if (arg == NULL)
-    // XXX If somebody can provide a portable test case in which sendto()
-    // blocks we can enable this test to run in both rounds.
-    abort ();
-
   struct sockaddr_un sun;
 
-  tempfd = socket (AF_UNIX, SOCK_DGRAM, 0);
+  tempfd = socket (AF_UNIX, SOCK_STREAM, 0);
   if (tempfd == -1)
-    FAIL_EXIT1 ("socket (AF_UNIX, SOCK_DGRAM, 0): %m");
+    FAIL_EXIT1 ("socket (AF_UNIX, SOCK_STREAM, 0): %m");
 
   int tries = 0;
   do
@@ -1254,23 +1250,30 @@ tf_sendto (void *arg)
   while (bind (tempfd, (struct sockaddr *) &sun,
 	       offsetof (struct sockaddr_un, sun_path)
 	       + strlen (sun.sun_path) + 1) != 0);
-  tempfname = strdup (sun.sun_path);
 
-  tempfd2 = socket (AF_UNIX, SOCK_DGRAM, 0);
+  listen (tempfd, 5);
+
+  tempfd2 = socket (AF_UNIX, SOCK_STREAM, 0);
   if (tempfd2 == -1)
-    FAIL_EXIT1 ("socket (AF_UNIX, SOCK_DGRAM, 0): %m");
+    FAIL_EXIT1 ("socket (AF_UNIX, SOCK_STREAM, 0): %m");
 
-  xpthread_barrier_wait (&b2);
+  set_socket_buffer (tempfd2);
+
+  if (connect (tempfd2, (struct sockaddr *) &sun, sizeof (sun)) != 0)
+    FAIL_EXIT1 ("connect: %m");
+
+  unlink (sun.sun_path);
 
   xpthread_barrier_wait (&b2);
 
+  if (arg != NULL)
+    xpthread_barrier_wait (&b2);
+
   pthread_cleanup_push (cl, NULL);
 
-  char mem[1];
+  char mem[WRITE_BUFFER_SIZE];
 
-  sendto (tempfd2, mem, arg == NULL ? sizeof (mem) : 1, 0,
-	  (struct sockaddr *) &sun,
-	  offsetof (struct sockaddr_un, sun_path) + strlen (sun.sun_path) + 1);
+  sendto (tempfd2, mem, arg == NULL ? sizeof (mem) : 1, 0, NULL, 0);
 
   pthread_cleanup_pop (0);