[v3,5/5] syscalls/clone07: Convert to new API

Signed-off-by: zhanglianjie <zhanglianjie@uniontech.com>

--
2.20.1

Hi!
> diff --git a/testcases/kernel/syscalls/clone/clone07.c b/testcases/kernel/syscalls/clone/clone07.c
> index 4b2e04ee7..304810c40 100644
> --- a/testcases/kernel/syscalls/clone/clone07.c
> +++ b/testcases/kernel/syscalls/clone/clone07.c
> @@ -1,86 +1,58 @@
> +// SPDX-License-Identifier: GPL-2.0-only
>  /*
>   * Copyright (c) International Business Machines  Corp., 2003.
>   * Copyright (c) 2012 Wanlong Gao <gaowanlong@cn.fujitsu.com>
> - *
> - * This program is free software; you can redistribute it and/or modify it
> - * under the terms of version 2 of the GNU General Public License as
> - * published by the Free Software Foundation.
> - *
> - * 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.
> - *
> - * 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 Street, Fifth Floor, Boston, MA 02110-1301 USA.
> - *
> - */
> -/*
> - *	This is a test for a glibc bug for the clone(2) system call.
>   */
> 
> -#if defined UCLINUX && !__THROW
> -/* workaround for libc bug */
> -#define __THROW
> -#endif
> +/*\
> + * [Description]
> + * This is a test for a glibc bug for the clone(2) system call.

We can describe this a bit better.

Apparently the bug was that using return from the do_child() function
caused SIGSEGV so this should probably be:

 [Description]

 Test for a libc bug where exitting child function by returning from it
 caused SIGSEGV.

> + */
> 
> -#include <errno.h>
>  #include <sched.h>
> -#include <sys/wait.h>
> -#include "test.h"
> +#include <stdio.h>
> +#include <stdlib.h>
> +#include "tst_test.h"
> +#include "lapi/syscalls.h"
>  #include "clone_platform.h"
> 
> -#define TRUE 1
> -#define FALSE 0
> -
> -static void setup();
> -static int do_child();
> -
> -char *TCID = "clone07";
> -int TST_TOTAL = 1;
> -
> -static void sigsegv_handler(int);
> -static void sigusr2_handler(int);
>  static int child_pid;
> -static int fail = FALSE;
> +static int fail;
> +static void *child_stack;
> 
> -int main(int ac, char **av)
> +static int do_child(void *arg LTP_ATTRIBUTE_UNUSED)
>  {
> +	return 0;
> +}
> 
> -	int lc, status;
> -	void *child_stack;
> -
> -	tst_parse_opts(ac, av, NULL, NULL);
> -
> -	setup();
> +static void sigsegv_handler(int sig LTP_ATTRIBUTE_UNUSED)
> +{
> +	if (child_pid == 0) {
> +		kill(getppid(), SIGUSR2);
> +		_exit(42);
> +	}
> +}
> 
> -	for (lc = 0; TEST_LOOPING(lc); lc++) {
> -		tst_count = 0;
> -		child_stack = malloc(CHILD_STACK_SIZE);
> -		if (child_stack == NULL)
> -			tst_brkm(TBROK, NULL,
> -				 "Cannot allocate stack for child");
> +static void sigusr2_handler(int sig LTP_ATTRIBUTE_UNUSED)
> +{
> +	if (child_pid != 0)
> +		fail = 1;
> +}

I do not get why we need this complicated handler setup.

Why can't we just SAFE_WAITPID() for the child, check the status and
fail the test if the status is anything else than:

WIFEXITED(status) && WEXITSTATUS(status) == 0

> -		child_pid = ltp_clone(SIGCHLD, do_child, NULL,
> -				      CHILD_STACK_SIZE, child_stack);
> +static void verify_clone(void)
> +{
> +	TST_EXP_PID_SILENT(ltp_clone(SIGCHLD, do_child, NULL, CHILD_STACK_SIZE,
> +				child_stack));
> 
> -		if (child_pid < 0)
> -			tst_brkm(TBROK | TERRNO, NULL, "clone failed");
> +	child_pid = TST_RET;
> 
> -		if ((wait(&status)) == -1)
> -			tst_brkm(TBROK | TERRNO, NULL,
> -				 "wait failed, status: %d", status);
> +	if (!TST_PASS)
> +		return;
> 
> -		free(child_stack);
> -	}
> +	tst_reap_children();
> 
> -	if (fail == FALSE)
> -		tst_resm(TPASS,
> -			 "Use of return() in child did not cause SIGSEGV");
> -	else
> -		tst_resm(TFAIL, "Use of return() in child caused SIGSEGV");
> +	TST_EXP_VAL(fail, 0, "Use of return() in child did not cause SIGSEGV");
> 
> -	tst_exit();
>  }
> 
>  static void setup(void)
> @@ -88,41 +60,26 @@ static void setup(void)
>  	struct sigaction def_act;
>  	struct sigaction act;
> 
> -	TEST_PAUSE;
> -
>  	act.sa_handler = sigsegv_handler;
>  	act.sa_flags = SA_RESTART;
> -	sigemptyset(&act.sa_mask);
> -	if ((sigaction(SIGSEGV, &act, NULL)) == -1)
> -		tst_resm(TWARN | TERRNO,
> -			 "sigaction() for SIGSEGV failed in test_setup()");
> +	SAFE_SIGEMPTYSET(&act.sa_mask);
> +	SAFE_SIGACTION(SIGSEGV, &act, NULL);
> 
> -	/* Setup signal handler for SIGUSR2 */
>  	def_act.sa_handler = sigusr2_handler;
>  	def_act.sa_flags = SA_RESTART | SA_RESETHAND;
> -	sigemptyset(&def_act.sa_mask);
> -
> -	if ((sigaction(SIGUSR2, &def_act, NULL)) == -1)
> -		tst_resm(TWARN | TERRNO,
> -			 "sigaction() for SIGUSR2 failed in test_setup()");
> -}
> +	SAFE_SIGEMPTYSET(&def_act.sa_mask);
> +	SAFE_SIGACTION(SIGUSR2, &def_act, NULL);
> 
> -static int do_child(void)
> -{
> -	return 0;
> +	child_stack = SAFE_MALLOC(CHILD_STACK_SIZE);
>  }
> 
> -static void sigsegv_handler(int sig)
> +static void cleanup(void)
>  {
> -	if (child_pid == 0) {
> -		kill(getppid(), SIGUSR2);
> -		_exit(42);
> -	}
> +	free(child_stack);
>  }
> 
> -/* sig_default_handler() - Default handler for parent */
> -static void sigusr2_handler(int sig)
> -{
> -	if (child_pid != 0)
> -		fail = TRUE;
> -}
> +static struct tst_test test = {
> +	.setup = setup,
> +	.test_all = verify_clone,
> +	.cleanup = cleanup,
> +};
> --
> 2.20.1
> 
> 
> 
> 
> -- 
> Mailing list info: https://lists.linux.it/listinfo/ltp

Hi,

On 2021-10-12 17:37, Cyril Hrubis wrote:
> Hi!
>> diff --git a/testcases/kernel/syscalls/clone/clone07.c b/testcases/kernel/syscalls/clone/clone07.c
>> index 4b2e04ee7..304810c40 100644
>> --- a/testcases/kernel/syscalls/clone/clone07.c
>> +++ b/testcases/kernel/syscalls/clone/clone07.c
>> @@ -1,86 +1,58 @@
>> +// SPDX-License-Identifier: GPL-2.0-only
>>   /*
>>    * Copyright (c) International Business Machines  Corp., 2003.
>>    * Copyright (c) 2012 Wanlong Gao <gaowanlong@cn.fujitsu.com>
>> - *
>> - * This program is free software; you can redistribute it and/or modify it
>> - * under the terms of version 2 of the GNU General Public License as
>> - * published by the Free Software Foundation.
>> - *
>> - * 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.
>> - *
>> - * 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 Street, Fifth Floor, Boston, MA 02110-1301 USA.
>> - *
>> - */
>> -/*
>> - *	This is a test for a glibc bug for the clone(2) system call.
>>    */
>>
>> -#if defined UCLINUX && !__THROW
>> -/* workaround for libc bug */
>> -#define __THROW
>> -#endif
>> +/*\
>> + * [Description]
>> + * This is a test for a glibc bug for the clone(2) system call.
> 
> We can describe this a bit better.
> 
> Apparently the bug was that using return from the do_child() function
> caused SIGSEGV so this should probably be:
> 
>   [Description]
> 
>   Test for a libc bug where exitting child function by returning from it
>   caused SIGSEGV.
> 
>> + */
>>
>> -#include <errno.h>
>>   #include <sched.h>
>> -#include <sys/wait.h>
>> -#include "test.h"
>> +#include <stdio.h>
>> +#include <stdlib.h>
>> +#include "tst_test.h"
>> +#include "lapi/syscalls.h"
>>   #include "clone_platform.h"
>>
>> -#define TRUE 1
>> -#define FALSE 0
>> -
>> -static void setup();
>> -static int do_child();
>> -
>> -char *TCID = "clone07";
>> -int TST_TOTAL = 1;
>> -
>> -static void sigsegv_handler(int);
>> -static void sigusr2_handler(int);
>>   static int child_pid;
>> -static int fail = FALSE;
>> +static int fail;
>> +static void *child_stack;
>>
>> -int main(int ac, char **av)
>> +static int do_child(void *arg LTP_ATTRIBUTE_UNUSED)
>>   {
>> +	return 0;
>> +}
>>
>> -	int lc, status;
>> -	void *child_stack;
>> -
>> -	tst_parse_opts(ac, av, NULL, NULL);
>> -
>> -	setup();
>> +static void sigsegv_handler(int sig LTP_ATTRIBUTE_UNUSED)
>> +{
>> +	if (child_pid == 0) {
>> +		kill(getppid(), SIGUSR2);
>> +		_exit(42);
>> +	}
>> +}
>>
>> -	for (lc = 0; TEST_LOOPING(lc); lc++) {
>> -		tst_count = 0;
>> -		child_stack = malloc(CHILD_STACK_SIZE);
>> -		if (child_stack == NULL)
>> -			tst_brkm(TBROK, NULL,
>> -				 "Cannot allocate stack for child");
>> +static void sigusr2_handler(int sig LTP_ATTRIBUTE_UNUSED)
>> +{
>> +	if (child_pid != 0)
>> +		fail = 1;
>> +}
> 
> I do not get why we need this complicated handler setup.
> 
> Why can't we just SAFE_WAITPID() for the child, check the status and
> fail the test if the status is anything else than:
> 
> WIFEXITED(status) && WEXITSTATUS(status) == 0
> 
The processing here is to accurately determine whether a segment fault 
signal has occurred in the child process. Of course, can also use the 
method you provide, but you can only judge that the child process exits 
abnormally.

>> -		child_pid = ltp_clone(SIGCHLD, do_child, NULL,
>> -				      CHILD_STACK_SIZE, child_stack);
>> +static void verify_clone(void)
>> +{
>> +	TST_EXP_PID_SILENT(ltp_clone(SIGCHLD, do_child, NULL, CHILD_STACK_SIZE,
>> +				child_stack));
>>
>> -		if (child_pid < 0)
>> -			tst_brkm(TBROK | TERRNO, NULL, "clone failed");
>> +	child_pid = TST_RET;
>>
>> -		if ((wait(&status)) == -1)
>> -			tst_brkm(TBROK | TERRNO, NULL,
>> -				 "wait failed, status: %d", status);
>> +	if (!TST_PASS)
>> +		return;
>>
>> -		free(child_stack);
>> -	}
>> +	tst_reap_children();
>>
>> -	if (fail == FALSE)
>> -		tst_resm(TPASS,
>> -			 "Use of return() in child did not cause SIGSEGV");
>> -	else
>> -		tst_resm(TFAIL, "Use of return() in child caused SIGSEGV");
>> +	TST_EXP_VAL(fail, 0, "Use of return() in child did not cause SIGSEGV");
>>
>> -	tst_exit();
>>   }
>>
>>   static void setup(void)
>> @@ -88,41 +60,26 @@ static void setup(void)
>>   	struct sigaction def_act;
>>   	struct sigaction act;
>>
>> -	TEST_PAUSE;
>> -
>>   	act.sa_handler = sigsegv_handler;
>>   	act.sa_flags = SA_RESTART;
>> -	sigemptyset(&act.sa_mask);
>> -	if ((sigaction(SIGSEGV, &act, NULL)) == -1)
>> -		tst_resm(TWARN | TERRNO,
>> -			 "sigaction() for SIGSEGV failed in test_setup()");
>> +	SAFE_SIGEMPTYSET(&act.sa_mask);
>> +	SAFE_SIGACTION(SIGSEGV, &act, NULL);
>>
>> -	/* Setup signal handler for SIGUSR2 */
>>   	def_act.sa_handler = sigusr2_handler;
>>   	def_act.sa_flags = SA_RESTART | SA_RESETHAND;
>> -	sigemptyset(&def_act.sa_mask);
>> -
>> -	if ((sigaction(SIGUSR2, &def_act, NULL)) == -1)
>> -		tst_resm(TWARN | TERRNO,
>> -			 "sigaction() for SIGUSR2 failed in test_setup()");
>> -}
>> +	SAFE_SIGEMPTYSET(&def_act.sa_mask);
>> +	SAFE_SIGACTION(SIGUSR2, &def_act, NULL);
>>
>> -static int do_child(void)
>> -{
>> -	return 0;
>> +	child_stack = SAFE_MALLOC(CHILD_STACK_SIZE);
>>   }
>>
>> -static void sigsegv_handler(int sig)
>> +static void cleanup(void)
>>   {
>> -	if (child_pid == 0) {
>> -		kill(getppid(), SIGUSR2);
>> -		_exit(42);
>> -	}
>> +	free(child_stack);
>>   }
>>
>> -/* sig_default_handler() - Default handler for parent */
>> -static void sigusr2_handler(int sig)
>> -{
>> -	if (child_pid != 0)
>> -		fail = TRUE;
>> -}
>> +static struct tst_test test = {
>> +	.setup = setup,
>> +	.test_all = verify_clone,
>> +	.cleanup = cleanup,
>> +};
>> --
>> 2.20.1
>>
>>
>>
>>
>> -- 
>> Mailing list info: https://lists.linux.it/listinfo/ltp
>

Hi!
> > I do not get why we need this complicated handler setup.
> > 
> > Why can't we just SAFE_WAITPID() for the child, check the status and
> > fail the test if the status is anything else than:
> > 
> > WIFEXITED(status) && WEXITSTATUS(status) == 0
> > 
> The processing here is to accurately determine whether a segment fault 
> signal has occurred in the child process. Of course, can also use the 
> method you provide, but you can only judge that the child process exits 
> abnormally.

Well I do not think that the SEGFAULT needs to have special treatement
here, as long as the child does anything else than exit with 0 it's a
bug, that's why I think that we should fail the test in all other cases.

I guess that we can add special check for the regression with
WIFSIGNALED() && WTERMSIG() == SIGSEGV and produce slightly different
TFAIL message, but we should really fail all the other cases as well.