problem about SEMAPHORE at cygwin 1.5.7

problem about SEMAPHORE at cygwin 1.5.7

[xuzhd]

[-- Attachment #1: Type: text/plain, Size: 2017 bytes --]



(See attached file: lcctestsem.c)

problem about SEMAPHORE  at cygwin 1.5.7

cygwin 1.5.7
cygserver as ipc server

a process has gotten semaphore using SEM_UNDO, when the  process aborts, it
should release corresponding semaphore at Linux.But at Cygwin 1.5.7,1.5.4,
1.3.22, I cannot prove it. is it a bug to cygwin ? thank you


the following is my program's result.




Administrator@node1 ~
$ ./a.exe  ( the first running)
before Get semaphore exclusively...
         semval[0] = 8
after Get semaphore ...
         semval[0] = 7
before release semaphore exclusively...
         semval[0] = 7
before release semaphore exclusively...
         semval[0] = 8

Administrator@node1 ~
$ ./a.exe                ( the second running, gets semaphore and releases
semaphore successfully)
Test semaphore has existed.
Get test semaphore...
         semval[0] = 8
Get test semaphore successfully.
before Get semaphore exclusively...
         semval[0] = 8
after Get semaphore ...
         semval[0] = 7
before release semaphore exclusively...
         semval[0] = 7
before release semaphore exclusively...
         semval[0] = 8

Administrator@node1 ~
$ ./a.exe                ( the third running, gets semaphore
successfully,but fails to  release semaphore )
Test semaphore has existed.
Get test semaphore...
         semval[0] = 8
Get test semaphore successfully.
before Get semaphore exclusively...
         semval[0] = 8
after Get semaphore ...
         semval[0] = 7
before release semaphore exclusively...

Administrator@node1 ~
$ ./a.exe           ( the fourth running, proves that the third running
didnot release semephore when aborting , a bug ????)
Test semaphore has existed.
Get test semaphore...
         semval[0] = 7
Get test semaphore successfully.
before Get semaphore exclusively...
         semval[0] = 7
after Get semaphore ...
         semval[0] = 6
before release semaphore exclusively...
         semval[0] = 6
before release semaphore exclusively...
         semval[0] = 7

Administrator@node1 ~
$ */


[-- Attachment #2: lcctestsem.c --]
[-- Type: application/octet-stream, Size: 3218 bytes --]

#include <stdio.h>
#include <errno.h>
#include <unistd.h>

#include <net/if.h>
#include <sys/ipc.h>
#include <sys/sem.h>


#define TESTSEMID	100
#define	SEMSIZE		1

#if defined(__GNU_LIBRARY__) && !defined(_SEM_SEMUN_UNDEFINED)
/* union semun is defined by including <sys/sem.h> */
#else
/* according to X/OPEN we have to define it ourselves */
union semun {
        int16_t val;                    /* value for SETVAL */
        struct semid_ds *buf;       /* buffer for IPC_STAT, IPC_SET */
        unsigned short int *array;  /* array for GETALL, SETALL */
        struct seminfo *__buf;      /* buffer for IPC_INFO */
};
#endif


int	test_semid;

void 
to_show(){

	union 	semun arg;
	unsigned short int allval[SEMSIZE];
	int i;
	
	arg.array = allval;

	if (semctl(test_semid, 0 , GETALL, arg) < 0) {
            printf("semctl failed for[%s].\n", strerror(errno));
            return ;
	}

	for( i=0; i < SEMSIZE; i++)
		printf("\t semval[%d] = %d \n", i , allval[i]);
	return; 
}

int
create_test_semaphore(void)
{
		int     key, i;
        char    fn[100];
		union 	semun arg;
		unsigned short int allval[SEMSIZE];

        sprintf(fn, ".");
        if ((key = ftok(fn, TESTSEMID)) < 0) {
                printf("ftok failed for[%s].\n", strerror(errno));
                return -1;
        }

        if ((test_semid = semget(key, SEMSIZE, IPC_CREAT | IPC_EXCL | 0666)) < 0) {
			if (errno == EEXIST) {
				printf("Test semaphore has existed.\n");
				printf("Get test semaphore...\n");
				if ((test_semid = semget(key, 0, 0666)) < 0) {
					printf("semget failed for[%s].\n", strerror(errno));
					return -1;
				} else {
					to_show();
					printf("Get test semaphore successfully.\n");
					return 0;
				}
			}
			printf("semget failed[%s].\n", strerror(errno));
			return -1;
        }
	
	for (i = 0; i< SEMSIZE; i++)
		allval[i] = 8;
	arg.array = allval;

	if (semctl(test_semid, SEMSIZE, SETALL, arg) < 0) {
            printf("semctl failed for[%s].\n", strerror(errno));
            return -1;
	}
	
	return 0;
}

int
mutex_op_semaphore(int op)
{
	int	i;
	struct sembuf oparray[SEMSIZE];
	
	for (i = 0; i < SEMSIZE; i++) {
		oparray[i].sem_num = i;
		oparray[i].sem_op = op;
		oparray[i].sem_flg = SEM_UNDO;
	}

	if (semop(test_semid, oparray, SEMSIZE) < 0) {
		printf("semop failed for[%s].\n", strerror(errno));
		return -1;
	} else
		return 0;
}

int
main(void)
{
	int	ret;

//	printf("Create test semaphore...\n");
	if (create_test_semaphore() < 0) {
		printf("Create test semaphore failed.\n");
		return -1;
	} //else
		//printf("Create test semaphore successfully.\n");

	printf("before Get semaphore exclusively...\n");
	to_show();
	if (mutex_op_semaphore(-1) < 0) {
		printf("Get test semaphore failed.\n");
		return -1;
	} else
		printf("after Get semaphore ...\n");

	to_show();

	sleep(5);	

	printf("before release semaphore exclusively...\n");
	to_show();

	if (mutex_op_semaphore(1) < 0) {
		printf("Release test semaphore failed.\n");
		return -1;
	} else 
		printf("before release semaphore exclusively...\n");
	to_show();
	return 0;
}



[-- Attachment #3: Type: text/plain, Size: 218 bytes --]

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: problem about SEMAPHORE at cygwin 1.5.7

[Corinna Vinschen]

On Mar  1 12:36, xuzhd@Lenovo.com wrote:
> (See attached file: lcctestsem.c)
> 
> problem about SEMAPHORE  at cygwin 1.5.7
> 
> cygwin 1.5.7
> cygserver as ipc server
> 
> a process has gotten semaphore using SEM_UNDO, when the  process aborts, it
> should release corresponding semaphore at Linux.But at Cygwin 1.5.7,1.5.4,
> 1.3.22, I cannot prove it. is it a bug to cygwin ? thank you

Yes, it's a bug in Cygserver.  Thanks for the testcase!  It helped to
track down the problem.  I've checked in a fix to Cygserver.  It will
show up in the next developers snapshot (http://cygwin.com/snapshots/).


Thanks again,
Corinna


-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Developer                                mailto:cygwin@cygwin.com
Red Hat, Inc.

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/