From: Cyril Hrubis <chrubis@suse.cz>
To: zhanglianjie <zhanglianjie@uniontech.com>
Cc: ltp@lists.linux.it
Subject: Re: [LTP] [PATCH v3 1/5] syscalls/clone02: Convert to new API
Date: Mon, 11 Oct 2021 17:40:17 +0200 [thread overview]
Message-ID: <YWRa4VvL33YclVX3@yuki> (raw)
In-Reply-To: <20210923085224.868-1-zhanglianjie@uniontech.com>
Hi!
> -/*
> - * test_SIG() - This function changes the signal handler for SIGUSR2
> - * signal for child. If CLONE_SIGHAND flag is set, this
> - * affects parent also.
> - */
> -static int test_SIG(void)
> +static void verify_clone(void)
> {
> + TST_EXP_PID_SILENT(ltp_clone(tcases[tst_variant].flags, child_fn, NULL,
> + CHILD_STACK_SIZE, child_stack));
>
> - struct sigaction new_act;
> + if (!TST_PASS)
> + return;
>
> - new_act.sa_handler = sig_child_defined_handler;
> - new_act.sa_flags = SA_RESTART;
> - sigemptyset(&new_act.sa_mask);
> -
> - /* Set signal handler to sig_child_defined_handler */
> - if (sigaction(SIGUSR2, &new_act, NULL) == -1) {
> - tst_resm(TWARN | TERRNO, "signal failed in test_SIG");
> - return -1;
> - }
> -
> - /* Send SIGUSR2 signal to parent */
> - if (kill(getppid(), SIGUSR2) == -1) {
> - tst_resm(TWARN | TERRNO, "kill failed in test_SIG");
> - return -1;
> - }
> + tst_reap_children();
>
> - return 0;
> + TST_EXP_PASS(tcases[tst_variant].parent_fn(), "%s", tcases[tst_variant].desc);
Can we, instead of this, print PASS/FAIL for each check we do, so that
if something fails the log explains what exactly has failed?
> }
>
> -/*
> - * modified_VM() - This function is called by parent process to check
> - * whether child's modification to parent_variable
> - * is visible to parent
> - */
> -
> -static int modified_VM(void)
> -{
> -
> - if (parent_variable == CHILD_VALUE)
> - /* child has modified parent_variable */
> - return 1;
> -
> - return 0;
> -}
>
> -/*
> - * modified_FILES() - This function checks for file descriptor table
> - * modifications done by child
> - */
> -static int modified_FILES(void)
> +static void cleanup(void)
> {
> - char buff[20];
> + SAFE_CHDIR(cwd_parent);
> + SAFE_UNLINK(TESTFILE);
> + SAFE_RMDIR(cwd_child);
>
> - if (((read(fd_parent, buff, sizeof(buff))) == -1) && (errno == EBADF))
> - /* Child has closed this file descriptor */
> - return 1;
> -
> - /* close fd_parent */
> - if ((close(fd_parent)) == -1)
> - tst_resm(TWARN | TERRNO, "close() failed in modified_FILES()");
> -
> - return 0;
> + free(cwd_parent);
> + free(child_stack);
> }
>
> -/*
> - * modified_FS() - This function checks parent's current working directory
> - * to see whether its modified by child or not.
> - */
> -static int modified_FS(void)
> +static void setup(void)
> {
> - char cwd[FILENAME_MAX];
> -
> - if ((getcwd(cwd, sizeof(cwd))) == NULL)
> - tst_resm(TWARN | TERRNO, "getcwd() failed");
> + struct sigaction def_act;
>
> - if (!(strcmp(cwd, cwd_parent)))
> - /* cwd hasn't changed */
> - return 0;
> + /* Save current working directory of parent */
> + cwd_parent = tst_get_tmpdir();
>
> - return 1;
> -}
> + /*
> + * Set value for parent_variable in parent, which will be
> + * changed by child for testing CLONE_VM flag
> + */
> + parent_variable = PARENT_VALUE;
>
> -/*
> - * modified_SIG() - This function checks whether child has changed
> - * parent's signal handler for signal, SIGUSR2
> - */
> -static int modified_SIG(void)
> -{
> + /*
> + * Open file from parent, which will be closed by
> + * child, used for testing CLONE_FILES flag
> + */
> + fd_parent = SAFE_OPEN(TESTFILE, O_CREAT | O_RDWR, 0777);
>
> - if (parent_got_signal)
> - /*
> - * parent came through sig_child_defined_handler()
> - * this means child has changed parent's handler
> - */
> - return 1;
> + /*
> + * set parent_got_signal to 0, used for testing
> + * CLONE_SIGHAND flag
> + */
> + parent_got_signal = 0;
We have to make sure we reset the $PWD, variable, got_signal flag and
open() the file before each test iteration otherwise the test will fail
on subsequent iterations with -i 2 command line parameter.
> - return 0;
> -}
> + def_act.sa_handler = sig_parent_default_handler;
> + def_act.sa_flags = SA_RESTART;
> + SAFE_SIGEMPTYSET(&def_act.sa_mask);
> + SAFE_SIGACTION(SIGUSR2, &def_act, NULL);
>
> -/*
> - * sig_child_defined_handler() - Signal handler installed by child
> - */
> -static void sig_child_defined_handler(int pid)
> -{
> - if ((syscall(__NR_gettid)) == child_pid)
> - /* Child got signal, give warning */
> - tst_resm(TWARN, "Child got SIGUSR2 signal");
> - else
> - parent_got_signal = TRUE;
> + SAFE_MKDIR(TESTDIR, 0777);
> + sprintf(cwd_child, "%s/%s", cwd_parent, TESTDIR);
> + child_stack = SAFE_MALLOC(CHILD_STACK_SIZE);
Can we use the guarded buffer instead of MALLOC in this test as well?
Just as we do in clone01.c now.
> }
>
> -/* sig_default_handler() - Default handler for parent */
> -static void sig_default_handler(void)
> -{
> -}
> +static struct tst_test test = {
> + .setup = setup,
> + .cleanup = cleanup,
> + .test_variants = ARRAY_SIZE(tcases),
This should rather be .tcnt and .test = verify_clone instead of
variants.
Test variants are usually used when the whole test is exactly same but
the TEST_*() function calls different variant of the syscall instead.
> + .test_all = verify_clone,
> + .needs_tmpdir = 1,
> +};
--
Cyril Hrubis
chrubis@suse.cz
--
Mailing list info: https://lists.linux.it/listinfo/ltp
next prev parent reply other threads:[~2021-10-11 15:39 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-09-23 8:52 [LTP] [PATCH v3 1/5] syscalls/clone02: Convert to new API zhanglianjie
2021-09-23 8:52 ` [LTP] [PATCH v3 2/5] syscalls/clone03: " zhanglianjie
2021-09-24 9:19 ` Richard Palethorpe
2021-10-11 15:02 ` Cyril Hrubis
2021-09-23 8:52 ` [LTP] [PATCH v3 3/5] syscalls/clone05: " zhanglianjie
2021-10-11 15:59 ` Cyril Hrubis
2021-09-23 8:52 ` [LTP] [PATCH v3 4/5] syscalls/clone06: " zhanglianjie
2021-10-12 9:21 ` Cyril Hrubis
2021-10-13 2:49 ` zhanglianjie
2021-09-23 8:52 ` [LTP] [PATCH v3 5/5] syscalls/clone07: " zhanglianjie
2021-10-12 9:37 ` Cyril Hrubis
2021-10-13 5:14 ` zhanglianjie
2021-10-13 10:00 ` Cyril Hrubis
2021-10-11 15:40 ` Cyril Hrubis [this message]
2021-10-12 6:39 ` [LTP] [PATCH v3 1/5] syscalls/clone02: " zhanglianjie
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=YWRa4VvL33YclVX3@yuki \
--to=chrubis@suse.cz \
--cc=ltp@lists.linux.it \
--cc=zhanglianjie@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox