samba.git/lib/tsocket, branch talloc-2.4.0

lib/tsocket: fix a typo in the tsocket guide doc

2023-01-17T18:23:18+00:00

Signed-off-by: Björn Baumbach 
Reviewed-by: Ralph Boehme 

Autobuild-User(master): Björn Baumbach 
Autobuild-Date(master): Tue Jan 17 18:23:18 UTC 2023 on sn-devel-184

tsocket: Fix the build on FreeBSD

2022-12-12T21:16:33+00:00

FreeBSD does not have TCP_USER_TIMEOUT

Signed-off-by: Volker Lendecke 
Reviewed-by: Jeremy Allison

lib/tsocket: avoid endless cpu-spinning in tstream_bsd_fde_handler()

2022-10-19T16:14:36+00:00

There were some reports that strace output an LDAP server socket is in
CLOSE_WAIT state, returning EAGAIN for writev over and over (after a call to
epoll() each time).

In the tstream_bsd code the problem happens when we have a pending
writev_send, while there's no readv_send pending. In that case
we still ask for TEVENT_FD_READ in order to notice connection errors
early, so we try to call writev even if the socket doesn't report TEVENT_FD_WRITE.
And there are situations where we do that over and over again.

It happens like this with a Linux kernel:

    tcp_fin() has this:
        struct tcp_sock *tp = tcp_sk(sk);

        inet_csk_schedule_ack(sk);

        sk->sk_shutdown |= RCV_SHUTDOWN;
        sock_set_flag(sk, SOCK_DONE);

        switch (sk->sk_state) {
        case TCP_SYN_RECV:
        case TCP_ESTABLISHED:
                /* Move to CLOSE_WAIT */
                tcp_set_state(sk, TCP_CLOSE_WAIT);
                inet_csk_enter_pingpong_mode(sk);
                break;

It means RCV_SHUTDOWN gets set as well as TCP_CLOSE_WAIT, but
sk->sk_err is not changed to indicate an error.

    tcp_sendmsg_locked has this:
    ...
        err = -EPIPE;
        if (sk->sk_err || (sk->sk_shutdown & SEND_SHUTDOWN))
                goto do_error;

        while (msg_data_left(msg)) {
                int copy = 0;

                skb = tcp_write_queue_tail(sk);
                if (skb)
                        copy = size_goal - skb->len;

                if (copy <= 0 || !tcp_skb_can_collapse_to(skb)) {
                        bool first_skb;

    new_segment:
                        if (!sk_stream_memory_free(sk))
                                goto wait_for_space;

    ...

    wait_for_space:
                set_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
                if (copied)
                        tcp_push(sk, flags & ~MSG_MORE, mss_now,
                                 TCP_NAGLE_PUSH, size_goal);

                err = sk_stream_wait_memory(sk, &timeo);
                if (err != 0)
                        goto do_error;

It means if (sk->sk_err || (sk->sk_shutdown & SEND_SHUTDOWN)) doesn't
hit as we only have RCV_SHUTDOWN and sk_stream_wait_memory returns
-EAGAIN.

    tcp_poll has this:

        if (sk->sk_shutdown & RCV_SHUTDOWN)
                mask |= EPOLLIN | EPOLLRDNORM | EPOLLRDHUP;

So we'll get EPOLLIN | EPOLLRDNORM | EPOLLRDHUP triggering
TEVENT_FD_READ and writev/sendmsg keeps getting EAGAIN.

So we need to always clear TEVENT_FD_READ if we don't
have readable handler in order to avoid burning cpu.
But we turn it on again after a timeout of 1 second
in order to monitor the error state of the connection.

And now that our tsocket_bsd_error() helper checks for POLLRDHUP,
we can check if the socket is in an error state before calling the
writable handler when TEVENT_FD_READ was reported.
Only on error we'll call the writable handler, which will pick
the error without calling writev().

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15202

Signed-off-by: Stefan Metzmacher 
Reviewed-by: Ralph Boehme

lib/tsocket: remember the first error as tstream_bsd->error

2022-10-19T16:14:36+00:00

If we found that the connection is broken, there's no point
in trying to use it anymore, so just return the first error we detected.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15202

Signed-off-by: Stefan Metzmacher 
Reviewed-by: Ralph Boehme

lib/tsocket: check for errors indicated by poll() before getsockopt(fd, SOL_SOCKET, SO_ERROR)

2022-10-19T16:14:36+00:00

This also returns an error if we got TCP_FIN from the peer,
which is only reported by an explicit POLLRDHUP check.

Also on FreeBSD getsockopt(fd, SOL_SOCKET, SO_ERROR) fetches
and resets the error, so a 2nd call no longer returns an error.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15202

Signed-off-by: Stefan Metzmacher 
Reviewed-by: Ralph Boehme

lib/tsocket: split out tsocket_bsd_error() from tsocket_bsd_pending()

2022-10-19T16:14:36+00:00

This will be used on its own soon.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15202

Signed-off-by: Stefan Metzmacher 
Reviewed-by: Ralph Boehme

lib/tsocket: Add tests for loop on EAGAIN

2022-10-19T16:14:36+00:00

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15202

Pair-Programmed-With: Stefan Metzmacher 

Signed-off-by: Andrew Bartlett 
Signed-off-by: Stefan Metzmacher

selftest: test tsocket_address_inet_from_hostport_strings

2021-09-28T10:34:12+00:00

Signed-off-by: Uri Simchoni 
Reviewed-by: Andrew Bartlett 

Autobuild-User(master): Andrew Bartlett 
Autobuild-Date(master): Tue Sep 28 10:34:12 UTC 2021 on sn-devel-184

selftest: add more tests for test_address_inet_from_strings

2021-09-28T09:44:35+00:00

Test the case of NULL address as input

Signed-off-by: Uri Simchoni 
Reviewed-by: Andrew Bartlett

lib/tsocket: new function to parse host port strs.

2021-09-28T09:44:35+00:00

tsocket_address_inet_from_hostport_strings() on top of
tsocket_address_inet_from_strings(), implementing the ability to parse a
port number appended to an IPv6 or IPv4 address. IPv6 addresses can also
optionally have square brackets around them, but these are needed to
specify the port number as colon is used to delimit port from the IP
address in the string.

Note that this code just recognises and parses the strings with port
given, or just IPv6 with square brackets.  The rest of the parsing is
passed on to tsocket_address_inet_from strings(), and errors from there
passed back up the stack.

Signed-off-by: Matthew Grant 
Reviewed-by: Uri Simchoni 
Reviewed-by: Andrew Bartlett