From: Alejandro Colomar <alx@kernel.org>
To: "Tomáš Golembiovský" <tgolembi@redhat.com>
Cc: linux-man@vger.kernel.org
Subject: Re: [PATCH] close.2: Suggest shutdown(2) when closing socket from another thread
Date: Fri, 25 Aug 2023 20:05:01 +0200 [thread overview]
Message-ID: <e0d5e4a4-82d4-68ae-a2c0-9c3e0b779806@kernel.org> (raw)
In-Reply-To: <717947ba515ec7b179574ca3a3577cf849199872.1692717581.git.tgolembi@redhat.com>
[-- Attachment #1.1: Type: text/plain, Size: 6432 bytes --]
Hi Tomáš,
On 2023-08-22 17:19, Tomáš Golembiovský wrote:
> This is another take on the ancient saga of closing sockets from one
> thread while another thread is blocked on recv(2). It all started with
> [1,2] and continued by [3]. It was established that this is expected and
> long term behavior or Linux and the issue was closed by Michael Kerrisk
> by commit c2f15a1349a7271f6c1d81361ec8b256266e1c09.
>
> This is all fine and the patch covered the issue in general terms.
> However, it does not mention the specific case of sockets and shutdown,
> where the issue can be (at least for the read case) mitigated by proper
> shutdown. While one may argue that such information may be implied by
> other man pages (perhaps return value of recv(2)) and thus is redundant,
> it seems only fair to mention shutdown(2) here as it is only rarely
> noted in Linux documentation that properly shutting down both side of
> the socket is a good programming practice when dealing with sockets.
>
> As a test program I am attaching the program originally written by Lukas
> Czerner. Only with small fixes here and there.
>
> [1] https://lore.kernel.org/linux-man/1314620800-15587-1-git-send-email-lczerner@redhat.com/
> [2] https://bugzilla.redhat.com/show_bug.cgi?id=650985
> [3] https://bugzilla.kernel.org/show_bug.cgi?id=53781
>
> ```c
> /**
> * Copyright 2011 (C) Red Hat, Inc., Lukas Czerner <lczerner@redhat.com>
> * Copyright 2023 (C) Red Hat, Inc., Tomas Golembiovsky <tgolembi@redhat.com>
> *
> * This program is free software; you can redistribute it and/or
> * modify it under the terms of the GNU General Public License as
> * published by the Free Software Foundation.
No version of the GPL may be understood as GPL-1.0-or-later. I suggest
specifying. But rather, we prefer SPDX license tags in this project,
so could you please specify the license as something SPDX?
Thanks,
Alex
> *
> * This program is distributed in the hope that it would be useful,
> * but WITHOUT ANY WARRANTY; without even the implied warranty of
> * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> * GNU General Public License for more details.
> *
> * You should have received a copy of the GNU General Public License
> * along with this program; if not, write the Free Software Foundation,
> * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
> */
>
> void *close_socket(void *arg) {
> int sockfd = *(int *)arg;
>
> sleep(3);
> printf("Thread: closing socket %d\n", sockfd);
> // shutdown(sockfd, SHUT_RDWR);
> close(sockfd);
> return NULL;
> }
>
> int client(void) {
> int sockfd;
> int len;
> struct sockaddr_un address;
> int ret;
> char *buffer=malloc(BUFSIZE);
> pthread_t thread;
>
> sockfd = socket(AF_UNIX, SOCK_STREAM, 0);
>
> address.sun_family = AF_UNIX;
> strcpy(address.sun_path, "server_socket");
> len = sizeof(address);
>
> ret = connect(sockfd, (struct sockaddr *)&address, len);
> if (ret == -1) {
> perror("connect");
> return 1;
> }
> printf("client connected\n");
>
> ret = pthread_create(&thread, NULL, close_socket, (void *)&sockfd);
> if (ret != 0) {
> perror("Creating thread failed");
> return 1;
> }
>
> while (1) {
> ret = recv(sockfd,buffer,BUFSIZE,0);
> if (ret < 0) {
> perror("recv");
> return 1;
> }
> printf("Data received: %s\n", buffer);
> sleep(1);
> }
>
> close(sockfd);
> return 0;
> }
>
> int server(void) {
> char *message="This is the message I am sending to you";
> struct sockaddr_un server_addr, client_addr;
> int server_sockfd, client_sockfd;
> socklen_t server_len, client_len;
> int ret;
>
> unlink("server_socket");
> server_sockfd = socket(AF_UNIX, SOCK_STREAM, 0);
>
> server_addr.sun_family = AF_UNIX;
> strcpy(server_addr.sun_path, "server_socket");
> server_len = sizeof(server_addr);
> bind(server_sockfd, (struct sockaddr *)&server_addr, server_len);
>
> listen(server_sockfd, 5);
>
> client_len = sizeof(client_addr);
> client_sockfd = accept(server_sockfd,
> (struct sockaddr *)&client_addr, &client_len);
>
> printf("Server: sending data...\n");
> ret = send(client_sockfd ,message,strlen(message),0);
> if (ret < 0) {
> perror("send");
> return 1;
> }
>
> /* simulate running server by not closing the client_socket socket */
> return 0;
> }
>
> int main() {
> pid_t pid;
>
> pid = fork();
> if (pid < 0) {
> perror("fork failed");
> exit(1);
> }
> if (pid > 0) {
> printf(" - starting server\n");
> server();
> printf(" - exiting server\n");
> wait(NULL);
> } else {
> sleep(1);
> printf(" - starting client\n");
> client();
> printf(" - exiting client\n");
> }
> }
> ```
>
> Signed-off-by: Tomáš Golembiovský <tgolembi@redhat.com>
> ---
> man2/close.2 | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/man2/close.2 b/man2/close.2
> index 68211bc58..08c6a0839 100644
> --- a/man2/close.2
> +++ b/man2/close.2
> @@ -145,6 +145,11 @@ Thus, the blocking system call in the first thread may successfully
> complete after the
> .BR close ()
> in the second thread.
> +.PP
> +When dealing with sockets,
> +blocking forever in another thread may be prevented by using
> +.BR shutdown (2)
> +to shut down both parts of the connection before closing the socket.
LGTM.
> .\"
> .SS Dealing with error returns from close()
> A careful programmer will check the return value of
--
<http://www.alejandro-colomar.es/>
GPG key fingerprint: A9348594CE31283A826FBDD8D57633D441E25BB5
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
next prev parent reply other threads:[~2023-08-25 18:05 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-08-22 15:19 [PATCH] close.2: Suggest shutdown(2) when closing socket from another thread Tomáš Golembiovský
2023-08-25 18:05 ` Alejandro Colomar [this message]
2023-09-14 7:39 ` Tomáš Golembiovský
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=e0d5e4a4-82d4-68ae-a2c0-9c3e0b779806@kernel.org \
--to=alx@kernel.org \
--cc=linux-man@vger.kernel.org \
--cc=tgolembi@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox