selftest: packetdrill: Import opt34/reset-* tests.

This imports the non-experimental version of opt34/reset-*.pkt.

                                   |  Child  |              RST              | sk_err  |
  ---------------------------------+---------+-------------------------------+---------+
  reset-after-accept.pkt           |   TFO   |   after accept(), SYN_RECV    |  read() |
  reset-close-with-unread-data.pkt |   TFO   |   after accept(), SYN_RECV    | write() |
  reset-before-accept.pkt          |   TFO   |  before accept(), SYN_RECV    |  read() |
  reset-non-tfo-socket.pkt         | non-TFO |  before accept(), ESTABLISHED | write() |

The first 3 files test scenarios where a SYN_RECV socket receives RST
before/after accept() and data in SYN must be read() without error,
but the following read() or fist write() will return ECONNRESET.

The last test is similar but with non-TFO socket.

Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://patch.msgid.link/20250927213022.1850048-10-kuniyu@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Author: q2ven

tools/testing/selftests/net/packetdrill/tcp_fastopen_server_reset-after-accept.pkt

@@ -0,0 +1,37 @@
+// SPDX-License-Identifier: GPL-2.0
+//
+// Send a RST to a TFO socket after it has been accepted.
+//
+// First read() will return all the data and this is consistent
+// with the non-TFO case. Second read will return -1
+
+`./defaults.sh`
+
+    0 socket(..., SOCK_STREAM|SOCK_NONBLOCK, IPPROTO_TCP) = 3
+   +0 setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0
+   +0 bind(3, ..., ...) = 0
+   +0 listen(3, 1) = 0
+   +0 setsockopt(3, SOL_TCP, TCP_FASTOPEN, [1], 4) = 0
+
+   +0 < S 0:10(10) win 32792 <mss 1460,sackOK,nop,nop,FO TFO_COOKIE,nop,nop>
+   +0 > S. 0:0(0) ack 11 <mss 1460,nop,nop,sackOK>
+
+   +0 accept(3, ..., ...) = 4
+   +0 %{ assert (tcpi_options & TCPI_OPT_SYN_DATA) != 0, tcpi_options }%
+   +0 %{ assert tcpi_state == TCP_SYN_RECV, tcpi_state }%
+
+// 1st read will return the data from SYN.
+// tcp_reset() sets sk->sk_err to ECONNRESET for SYN_RECV.
+   +0 < R. 11:11(0) win 32792
+   +0 %{ assert tcpi_state == TCP_CLOSE, tcpi_state }%
+
+// This one w/o ACK bit will cause the same effect.
+// +0 < R 11:11(0) win 32792
+// See Step 2 in tcp_validate_incoming().
+
+// found_ok_skb in tcp_recvmsg_locked()
+   +0 read(4, ..., 512) = 10
+
+// !copied && sk->sk_err -> sock_error(sk)
+   +0 read(4, ..., 512) = -1 ECONNRESET (Connection reset by peer)
+   +0 close(4) = 0

tools/testing/selftests/net/packetdrill/tcp_fastopen_server_reset-before-accept.pkt

@@ -0,0 +1,32 @@
+// SPDX-License-Identifier: GPL-2.0
+//
+// Send a RST to a TFO socket before it is accepted.
+//
+// The socket won't go away and after it's accepted the data
+// in the SYN pkt can still be read. But that's about all that
+// the acceptor can do with the socket.
+
+`./defaults.sh`
+
+    0 socket(..., SOCK_STREAM|SOCK_NONBLOCK, IPPROTO_TCP) = 3
+   +0 setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0
+   +0 bind(3, ..., ...) = 0
+   +0 listen(3, 1) = 0
+   +0 setsockopt(3, SOL_TCP, TCP_FASTOPEN, [1], 4) = 0
+
+   +0 < S 0:10(10) win 32792 <mss 1460,sackOK,nop,nop,nop,wscale 7,FO TFO_COOKIE,nop,nop>
+   +0 > S. 0:0(0) ack 11 <mss 1460,nop,nop,sackOK,nop,wscale 8>
+
+// 1st read will return the data from SYN.
+   +0 < R. 11:11(0) win 257
+
+// This one w/o ACK bit will cause the same effect.
+// +0 < R 11:11(0) win 257
+
+   +0 accept(3, ..., ...) = 4
+   +0 %{ assert (tcpi_options & TCPI_OPT_SYN_DATA) != 0, tcpi_options }%
+   +0 %{ assert tcpi_state == TCP_CLOSE, tcpi_state }%
+
+   +0 read(4, ..., 512) = 10
+   +0 read(4, ..., 512) = -1 ECONNRESET (Connection reset by peer)
+   +0 close(4) = 0

tools/testing/selftests/net/packetdrill/tcp_fastopen_server_reset-close-with-unread-data.pkt

@@ -0,0 +1,32 @@
+// SPDX-License-Identifier: GPL-2.0
+//
+// Send a RST to a TFO socket after it is accepted.
+//
+// The socket will change to TCP_CLOSE state with pending data so
+// write() will fail. Pending data can be still be read and close()
+// won't trigger RST if data is not read
+//
+// 565b7b2d2e63 ("tcp: do not send reset to already closed sockets")
+// https://lore.kernel.org/netdev/4C1A2502.1030502@openvz.org/
+
+`./defaults.sh`
+
+    0 socket(..., SOCK_STREAM|SOCK_NONBLOCK, IPPROTO_TCP) = 3
+   +0 setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0
+   +0 bind(3, ..., ...) = 0
+   +0 listen(3, 1) = 0
+   +0 setsockopt(3, SOL_TCP, TCP_FASTOPEN, [1], 4) = 0
+
+   +0 < S 0:10(10) win 32792 <mss 1460,sackOK,nop,nop, FO TFO_COOKIE,nop,nop>
+   +0 > S. 0:0(0) ack 11 <mss 1460,nop,nop,sackOK>
+
+   +0 accept(3, ..., ...) = 4
+   +0 %{ assert (tcpi_options & TCPI_OPT_SYN_DATA) != 0, tcpi_options }%
+   +0 %{ assert tcpi_state == TCP_SYN_RECV, tcpi_state }%
+
+// tcp_done() sets sk->sk_state to TCP_CLOSE and clears tp->fastopen_rsk
+   +0 < R. 11:11(0) win 32792
+   +0 %{ assert tcpi_state == TCP_CLOSE, tcpi_state }%
+
+   +0 write(4, ..., 100) = -1 ECONNRESET(Connection reset by peer)
+   +0 close(4) = 0

tools/testing/selftests/net/packetdrill/tcp_fastopen_server_reset-non-tfo-socket.pkt

@@ -0,0 +1,37 @@
+// SPDX-License-Identifier: GPL-2.0
+//
+// Send a RST to a fully established socket with pending data before
+// it is accepted.
+//
+// The socket with pending data won't go away and can still be accepted
+// with data read. But it will be in TCP_CLOSE state.
+
+`./defaults.sh`
+
+    0 socket(..., SOCK_STREAM|SOCK_NONBLOCK, IPPROTO_TCP) = 3
+   +0 setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0
+   +0 bind(3, ..., ...) = 0
+   +0 listen(3, 1) = 0
+   +0 setsockopt(3, SOL_TCP, TCP_FASTOPEN, [1], 4) = 0
+
+// Invalid cookie, so accept() fails.
+   +0 < S 0:10(10) win 32792 <mss 1460,sackOK,nop,nop,FO aaaaaaaaaaaaaaaa,nop,nop>
+   +0 > S. 0:0(0) ack 1 <mss 1460,nop,nop,sackOK, FO TFO_COOKIE,nop,nop>
+
+   +0 accept(3, ..., ...) = -1 EAGAIN (Resource temporarily unavailable)
+
+// Complete 3WHS and send data and RST
+   +0 < . 1:1(0) ack 1 win 32792
+   +0 < . 1:11(10) ack 1 win 32792
+   +0 < R. 11:11(0) win 32792
+
+// A valid reset won't make the fully-established socket go away.
+// It's just that the acceptor will get a dead, unusable socket
+// in TCP_CLOSE state.
+   +0 accept(3, ..., ...) = 4
+   +0 %{ assert (tcpi_options & TCPI_OPT_SYN_DATA) == 0, tcpi_options }%
+   +0 %{ assert tcpi_state == TCP_CLOSE, tcpi_state }%
+
+   +0 write(4, ..., 100) = -1 ECONNRESET(Connection reset by peer)
+   +0 read(4, ..., 512) = 10
+   +0 read(4, ..., 512) = 0