valgrind sucks

[Kuciwalker]

#include stdio.h
#include sys/types.h
#include sys/ipc.h
#include sys/sem.h

#define KEY 1942

main ()
{
        struct sembuf sbuf;
        int semid;

        semid = semget(KEY, 1, IPC_CREAT | IPC_EXCL | 0777);
        printf("semid = %d\n", semid);

        semctl(semid, 0, IPC_RMID);
        puts("destroyed semid");
}

==19854== Syscall param semctl(arg) points to uninitialised byte(s)
==19854==    at 0xB8AA08: semctl@@GLIBC_2.2 (in /lib/tls/libc-2.3.4.so)
==19854==    by 0x8048477: main (in ~/test/semtest)
==19854==  Address 0xBEE053D4 is on thread 1's stack

[Kuciwalker]

Alternatively, glibc sucks.

[Krill]

I laugh at you for taking a stupid degree

[BlackCat]

The semget() function returns the semaphore identifier associated with key.

A semaphore identifier with its associated semid_ds data structure and its associated set of nsems semaphores, see , are created for key if one of the following is true:

* The argument key is equal to IPC_PRIVATE .

* The argument key does not already have a semaphore identifier associated with it and (semflg & IPC_CREAT) is non-zero.

Upon creation, the semid_ds data structure associated with the new semaphore identifier is initialised as follows:

* In the operation permissions structure sem_perm.cuid, sem_perm.uid, sem_perm.cgid and sem_perm.gid are set equal to the effective user ID and effective group ID, respectively, of the calling process.

* The low-order 9 bits of sem_perm.mode are set equal to the low-order 9 bits of semflg.

* The variable sem_nsems is set equal to the value of nsems.

* The variable sem_otime is set equal to 0 and sem_ctime is set equal to the current time.

* The data structure associated with each semaphore in the set is not initialised. The semctl() function with the command SETVAL or SETALL can be used to initialise each semaphore.

[Kuciwalker]

That's not a good excuse for throwing errors on delete.

[BlackCat]

Ehrm, I think that the programmers of valgrind didn't expect someone to create a sem just to delete it immediatedly without initialization - who would do that ?

To be honest, I really don't think valgrind cares a bit about the point that you try to delete it - it just reports that uninitialized data is accessed.

[Asher]

Bad code. Both the C/C++ Standards demand main() have a return type.

There are two and only two ways to declare a main function:

int main (void);

int main (int argc, char **argv);

I am very proud to say I worked on a compiler that would not compile the code in the OP and ***** at you to learn how to program the language properly.

Sidenote: This thread reminds me of one of my favourite interview questions: How should you handle a destructor that fails?

I can provide a solution if someone wants it.

[BlackCat]

Asher, you missed a thing - main without a return value isn't acceptable either.

Bet that Kuci hasn't -Wall set

[Asher]

Asher, you missed a thing - main without a return value isn't acceptable either.

Bet that Kuci hasn't -Wall set

That's kind of implicit in the requirement. If you declare a return type, you must return a value.

[BlackCat]

Yeah, but if you don't ask for warns, it still compiles without further notice

[Asher]

Only on ****ty compilers.

On IBM's compilers, if you write bad code like this you need to add options to tell the compiler to let you be stupid.

[BlackCat]

AIX would allow it.

[Asher]

AIX is not a compiler.

[Asher]

So everyone knows the answer to the destructor question?

[BlackCat]

cc on an AIX platform would accept it

And don't ask me to verify it - we burned our last one in the previous century.