From: Cyril Hrubis <chrubis@suse.cz>
To: lufei <lufei@uniontech.com>
Cc: ltp@lists.linux.it
Subject: Re: [LTP] [PATCH] syscalls/unshare: New test: CLONE_NEWNS unshares fs info
Date: Wed, 23 Apr 2025 11:10:15 +0200 [thread overview]
Message-ID: <aAiud8vv-DHICLXY@yuki.lan> (raw)
In-Reply-To: <20250423044601.9320-1-lufei@uniontech.com>
Hi!
> diff --git a/runtest/syscalls b/runtest/syscalls
> index 844ae7a13..57338297a 100644
> --- a/runtest/syscalls
> +++ b/runtest/syscalls
> @@ -1725,6 +1725,7 @@ unlinkat01 unlinkat01
> unshare01 unshare01
> unshare02 unshare02
> unshare03 unshare03
> +unshare04 unshare04
>
> #
> # These tests require an unmounted block device
> diff --git a/testcases/kernel/syscalls/unshare/.gitignore b/testcases/kernel/syscalls/unshare/.gitignore
> index e5b5c261d..b1206e452 100644
> --- a/testcases/kernel/syscalls/unshare/.gitignore
> +++ b/testcases/kernel/syscalls/unshare/.gitignore
> @@ -1,3 +1,4 @@
> /unshare01
> /unshare02
> /unshare03
> +/unshare04
> diff --git a/testcases/kernel/syscalls/unshare/unshare04.c b/testcases/kernel/syscalls/unshare/unshare04.c
> new file mode 100644
> index 000000000..0bbb9d19d
> --- /dev/null
> +++ b/testcases/kernel/syscalls/unshare/unshare04.c
> @@ -0,0 +1,69 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +/*
> + * Copyright (c) 2025 lufei <lufei@uniontech.com>
> + */
> +
> +/*\
> + * This test case is to verify unshare(CLONE_NEWNS) also unshares filesystem
> + * information.
> + *
> + */
> +
> +#define _GNU_SOURCE
> +
> +#include "tst_test.h"
> +#include "lapi/sched.h"
> +
> +#ifdef HAVE_UNSHARE
This shouldn't be needed anymore, the minimal glibc we support at the
moment is 2.22, unshare() if I'm looking right is included in glibc 2.4
and newer. I suppose that removing the configure check for unshare() is
long overdue.
> +#define TMP "/tmp"
The test shouldn't expect or touch directories outside the test
temporary directory. If you need a diretory to chdir() into create one
in the test setup() instead.
> +static void run(void)
> +{
> + char *c_cwd;
> + char *p_cwd;
> + size_t size = 1024;
> +
> + c_cwd = SAFE_MMAP(NULL, size, PROT_READ | PROT_WRITE,
> + MAP_SHARED | MAP_ANONYMOUS,
> + -1, 0);
> +
> + p_cwd = SAFE_MALLOC(size);
These allocations should be made in test setup and freed in the test
cleanup.
> + struct tst_clone_args args = {
> + .flags = CLONE_FS,
> + .exit_signal = SIGCHLD,
> + };
> +
> + if (!SAFE_CLONE(&args)) {
> +
> + TST_EXP_PASS(unshare(CLONE_NEWNS));
> +
> + SAFE_CHDIR(TMP);
> + SAFE_GETCWD(c_cwd, size);
> + } else {
> + SAFE_WAIT(NULL);
> +
> + SAFE_GETCWD(p_cwd, size);
> +
> + tst_res(TDEBUG, "parent cwd: %s", p_cwd);
> + tst_res(TDEBUG, "child cwd: %s", c_cwd);
> +
> + //in parent, cwd stays in tst_tmpdir_path than TMP.
> + if (strcmp(p_cwd, c_cwd) == 0)
> + tst_res(TFAIL, "cwd not unshared as expected");
> + else
> + tst_res(TPASS, "cwd unshared");
We are validating that the CWD wasn't changed in parent after we have
done unshare() and chdir() in the child. Why do we even compare the
parent CWD againts the child? Rather than that we have to compare the
CWD in the parent againts the CWD at the start of the test.
What we should do is:
- do chdir() in child
- check in child that CWD was changed
- wait child in parent
- check in parent that CWD wasn't changed
For that all we have to do is to do getcwd() in the test setup and then
use it for a comparsion both in the child and parent.
> + }
> +}
> +
> +static struct tst_test test = {
> + .forks_child = 1,
> + .needs_root = 1,
> + .needs_tmpdir = 1,
> + .test_all = run,
> +};
> +
> +#else
> +TST_TEST_TCONF("unshare syscall is undefined.");
> +#endif
> --
> 2.39.3
>
>
> --
> Mailing list info: https://lists.linux.it/listinfo/ltp
--
Cyril Hrubis
chrubis@suse.cz
--
Mailing list info: https://lists.linux.it/listinfo/ltp
next prev parent reply other threads:[~2025-04-23 9:10 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-04-23 4:46 [LTP] [PATCH] syscalls/unshare: New test: CLONE_NEWNS unshares fs info lufei
2025-04-23 9:10 ` Cyril Hrubis [this message]
2025-04-23 12:05 ` [LTP] [PATCH v2] " lufei
2025-04-25 9:07 ` Cyril Hrubis
2025-04-25 9:33 ` [LTP] [PATCH v3] " lufei
2025-04-25 12:10 ` Cyril Hrubis
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=aAiud8vv-DHICLXY@yuki.lan \
--to=chrubis@suse.cz \
--cc=ltp@lists.linux.it \
--cc=lufei@uniontech.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.