[LTP] [PATCH v4 1/2] syscalls/prctl08: New test for prctl() with PR_{SET, GET}_TIMERSLACK

Yang Xu xuyang2018.jy@cn.fujitsu.com
Fri Nov 8 12:03:52 CET 2019


on 2019/11/07 21:05, Cyril Hrubis wrote:

> Hi!
>> ------
>> v3->v4:
>> split it into two cases, prctl08 test read/set timer slack value and
>> prctl09 is a timer samle test.
>> Also, in prctl08, I split into 5 point(0,1,middle,max and inherit).
>> -----
>>
>> Signed-off-by: Yang Xu <xuyang2018.jy@cn.fujitsu.com>
>> ---
>>   include/lapi/prctl.h                       |   5 +
>>   runtest/syscalls                           |   1 +
>>   testcases/kernel/syscalls/prctl/.gitignore |   1 +
>>   testcases/kernel/syscalls/prctl/prctl08.c  | 135 +++++++++++++++++++++
>>   4 files changed, 142 insertions(+)
>>   create mode 100644 testcases/kernel/syscalls/prctl/prctl08.c
>>
>> diff --git a/include/lapi/prctl.h b/include/lapi/prctl.h
>> index 8ee492259..0b4e196c3 100644
>> --- a/include/lapi/prctl.h
>> +++ b/include/lapi/prctl.h
>> @@ -19,6 +19,11 @@
>>   # define PR_SET_SECCOMP  22
>>   #endif
>>   
>> +#ifndef PR_SET_TIMERSLACK
>> +# define PR_SET_TIMERSLACK 29
>> +# define PR_GET_TIMERSLACK 30
>> +#endif
>> +
>>   #ifndef PR_SET_CHILD_SUBREAPER
>>   # define PR_SET_CHILD_SUBREAPER	36
>>   # define PR_GET_CHILD_SUBREAPER	37
>> diff --git a/runtest/syscalls b/runtest/syscalls
>> index 4e6310193..76961a684 100644
>> --- a/runtest/syscalls
>> +++ b/runtest/syscalls
>> @@ -878,6 +878,7 @@ prctl04 prctl04
>>   prctl05 prctl05
>>   prctl06 prctl06
>>   prctl07 prctl07
>> +prctl08 prctl08
>>   
>>   pread01 pread01
>>   pread01_64 pread01_64
>> diff --git a/testcases/kernel/syscalls/prctl/.gitignore b/testcases/kernel/syscalls/prctl/.gitignore
>> index 2178db366..fe36a8e0f 100644
>> --- a/testcases/kernel/syscalls/prctl/.gitignore
>> +++ b/testcases/kernel/syscalls/prctl/.gitignore
>> @@ -6,3 +6,4 @@
>>   /prctl06
>>   /prctl06_execve
>>   /prctl07
>> +/prctl08
>> diff --git a/testcases/kernel/syscalls/prctl/prctl08.c b/testcases/kernel/syscalls/prctl/prctl08.c
>> new file mode 100644
>> index 000000000..9aa15b06d
>> --- /dev/null
>> +++ b/testcases/kernel/syscalls/prctl/prctl08.c
>> @@ -0,0 +1,135 @@
>> +// SPDX-License-Identifier: GPL-2.0-or-later
>> +/*
>> + * Copyright (c) 2019 FUJITSU LIMITED. All rights reserved.
>> + * Author: Yang Xu <xuyang2018.jy@cn.fujitsu.com>
>> + *
>> + * Test PR_GET_TIMERSLACK and PR_SET_TIMERSLACK of prctl(2).
>> + * 1)Each thread has two associated timer slack values: a "default"
>> + *   value, and a "current" value. PR_SET_TIMERSLACK sets the "current"
>> + *   timer slack value for the calling thread.
>> + * 2)When a new thread is created, the two timer slack values are made
>> + *   the same as the "current" value of the creating thread.
>> + * 3)The maximum timer slack value is ULONG_MAX. On 32bit machines, it
>> + *   is a valid value(about 4s). On 64bit machines, it is about 500 years
>> + *   and no person will set this over 4s.  prctl return value is int, so
>> + *   we test themaximum value is INT_MAX.
>> + * 4)we also check current value via /proc/self/timerslack_ns if it is
>> + *  supported.
>> + */
>> +
>> +#include <sys/prctl.h>
>> +#include <string.h>
>> +#include <stdio.h>
>> +#include <stdlib.h>
>> +#include <linux/limits.h>
>> +#include "lapi/syscalls.h"
>> +#include "lapi/prctl.h"
>> +#include "tst_test.h"
>> +
>> +#define PROC_TIMERSLACK_PATH "/proc/self/timerslack_ns"
>> +
>> +static void check_get_timerslack(char *message, unsigned long value);
>> +static void check_inherit_timerslack(char *message, unsigned long value);
>> +
>> +static struct tcase {
>> +	void (*func_check)();
>> +	unsigned long setvalue;
>> +	unsigned long expvalue;
>> +	char message[50];
>> +} tcases[] = {
>> +	{check_get_timerslack, 0, 50000, "Reset"},
>> +	{check_get_timerslack, 1, 1, "Min"},
>> +	{check_get_timerslack, 70000, 70000, "Middle"},
>> +	{check_get_timerslack, INT_MAX, INT_MAX, "Max"},
>> +	{check_inherit_timerslack, 70000, 70000, "Child process"},
>> +};
>> +
>> +static int proc_flag = 1;
>> +
>> +static void check_proc_timerslack(char *message, unsigned long value)
>> +{
>> +	unsigned long proc_value;
>> +
>> +	SAFE_FILE_SCANF(PROC_TIMERSLACK_PATH, "%lu", &proc_value);
>> +	if (proc_value == value)
>> +		tst_res(TPASS, "%s %s  got %lu expectedly",
>> +				message, PROC_TIMERSLACK_PATH, proc_value);
>> +	else
>> +		tst_res(TFAIL, "%s %s expected %lu got %lu",
>> +				message, PROC_TIMERSLACK_PATH, value, proc_value);
>> +}
>> +
>> +static void check_get_timerslack(char *message, unsigned long value)
>> +{
>> +	TEST(prctl(PR_GET_TIMERSLACK));
>> +	if ((unsigned long)TST_RET == value)
>> +		tst_res(TPASS, "%s prctl(PR_GET_TIMERSLACK) got %lu expectedly",
>> +				message, value);
>> +	else
>> +		tst_res(TFAIL, "%s prctl(PR_GET_TIMERSLACK) expected %lu got %lu",
>> +				message, value, TST_RET);
>> +
>> +	if (proc_flag)
>> +		check_proc_timerslack(message, value);
>> +
>> +}
>> +
>> +static void check_inherit_timerslack(char *message, unsigned long value)
>> +{
>> +	int pid;
>> +	unsigned long current_value;
>> +	unsigned long default_value;
>> +
>> +	pid = SAFE_FORK();
>> +	if (pid == 0) {
>> +		current_value = prctl(PR_GET_TIMERSLACK);
>> +		prctl(PR_SET_TIMERSLACK, 0);
>> +		default_value = prctl(PR_GET_TIMERSLACK);
>> +		if (current_value == value && default_value == value)
>> +			tst_res(TPASS,
>> +				"%s two timer slack values are made the same as the current value(%lu) of the creating thread.",
>> +				message, value);
>> +		else
>> +			tst_res(TFAIL,
>> +				"%s current_value is %lu, default value is %lu, the parent current value is %lu",
>> +				message, current_value, default_value, value);
>> +	}
>> +
>> +}
>> +
>> +static void verify_prctl(unsigned int n)
>> +{
>> +	struct tcase *tc = &tcases[n];
>> +
>> +	TEST(prctl(PR_SET_TIMERSLACK, tc->setvalue));
>> +	if (TST_RET == -1) {
>> +		tst_res(TFAIL | TTERRNO, "prctl(PR_SET_TIMERSLACK, %lu) failed",
>> +					  tc->setvalue);
>> +		return;
>> +	}
>> +
>> +	tst_res(TPASS, "prctl(PR_SET_TIMERSLACK, %lu) succeed", tc->setvalue);
>> +	tc->func_check(tc->message, tc->expvalue);
>> +}
>> +
>> +static void setup(void)
>> +{
>> +	if (access(PROC_TIMERSLACK_PATH, F_OK) == -1) {
>> +		tst_res(TCONF, "proc doesn't support timerslack_ns interface");
>> +		proc_flag = 0;
>> +	}
>> +
>> +	TEST(prctl(PR_GET_TIMERSLACK));
>> +	if (TST_RET == 50000)
>> +		tst_res(TINFO, "current timerslack value is 50us");
>> +	else
>> +		tst_brk(TCONF,
>> +			"current environment doesn't meet test, please set timerslack to 50us");
> Why can't we just store the result here into a global and use it later
> on in the reset case?
>
> Given that we have check function already all we need to do is to add
> one more function that passes the global to the check_get_timerslack()
> instead of the expected value.

Hi Cyril

It sounds reasonable. We should not limit it in 50us. I will follow your advise.

Thanks
Yang Xu

>> +}
>> +
>> +static struct tst_test test = {
>> +	.setup = setup,
>> +	.test = verify_prctl,
>> +	.tcnt = ARRAY_SIZE(tcases),
>> +	.forks_child = 1,
>> +};
>> -- 
>> 2.18.1
>>
>>
>>


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20191108/ff1610fb/attachment.htm>


More information about the ltp mailing list