Fail to allocate semaphore/netconn Tiva C TM4C1294 - c

Question

I am using tm4c1294+lwip1.4.1+FreeRTOS.
As netconn_alloc() is called for socket communication, it allocates an unused semaphore. the number of semaphore is defined as SYS_SEM_MAX, so it can not be over SYS_SEM_MAX.
However, as semaphores are allocated continuously it reaches SYS_SEM_MAX and stop working since I guess sys_sem_free() does not deallocate it properly
Here is function that creates a semaphore implemented in sys_arch.c
err_t
sys_sem_new(sys_sem_t *sem, u8_t count)
{
void *temp;
u32_t i;
/* Find a semaphore that is not in use. */
for(i = 0; i < SYS_SEM_MAX; i++) {
if(sems[i].queue == 0) {
break;
}
}
if(i == SYS_SEM_MAX) {
#if SYS_STATS
STATS_INC(sys.sem.err);
#endif /* SYS_STATS */
return ERR_MEM;
}
/* Create a single-entry queue to act as a semaphore. */
#if RTOS_FREERTOS
sem->queue = xQueueCreate(1, sizeof(void *));
if(sem->queue == NULL) {
#endif /* RTOS_FREERTOS */
#if SYS_STATS
STATS_INC(sys.sem.err);
#endif /* SYS_STATS */
return ERR_MEM;
}
/* Acquired the semaphore if necessary. */
if(count == 0) {
temp = 0;
xQueueSend(sem->queue, &temp, 0);
}
/* Update the semaphore statistics. */
#if SYS_STATS
STATS_INC(sys.sem.used);
#if LWIP_STATS
if(lwip_stats.sys.sem.max < lwip_stats.sys.sem.used) {
lwip_stats.sys.sem.max = lwip_stats.sys.sem.used;
}
#endif
#endif /* SYS_STATS */
/* Save the queue handle. */
sems[i].queue = sem->queue;
/* Return this semaphore. */
return (ERR_OK);
}
Here is another function that frees semaphore implemented in sys_arch.c
void
sys_sem_free(sys_sem_t *sem)
{
/* Delete Sem , By Jin */
vQueueDelete(sem->queue);
/* Clear the queue handle. */
sem->queue = 0;
/* Update the semaphore statistics. */
#if SYS_STATS
STATS_DEC(sys.sem.used);
#endif /* SYS_STATS */
}
Whenever netconn_free() is called sys_sem_free() deallocates the semaphore, but it does not free the semaphore assigned in sem[] array.
I added vQueueDelete(sem->queue); that was suggested by someone, but still all same.
Not only functions creates/frees semaphore but also functions handling mbox are same as functions above, so functions handling mbox could be wrong as well.
Someone already reported this issue to TI, but it seems they have not solved the problems yet.
Therefore, I may need to implement my own functions handling semaphore/mbox in sys_arch.c, but I don't have any clues so far.
Can anyone give me any ideas? or anything?
Thanks,
Jin

Answer 1

The sys_arch.txt file in /doc is somewhat helpful. Apparently, looking at that document, and at what lwip 1.3.2 used to do, it looks like the ports/tiva-tm4c129/sys_arch.c is incorrect and incomplete.
sys_sem_free() should indeed be doing the vQueueDelete() as you discovered. It should not be doing "sem->queue = 0". If you look at netconn_free() over in src/api/api_msg.c, you can see it calls sys_sem_free() and then sys_sem_set_invalid(). The queue handle will be needed in the second function, and should not be clobbered in the first.
sys_sem_set_invalid() should make a sweep over the sems[] array and if it finds a match to the sem->queue, it should zero out that copy in sems[]. Once that is done, it should set sem->queue to 0.
This, I think, best matches what's in Dunkel's sys_arch.txt document, and fixed the resource leak on my system.
I concur that the mailboxes are in the same shape. Fwiw, I went ahead and modified those in similar fashion to that just described for the sems.
btw, the lwip files I was working with were from the TivaWare_C_Series 2.1.0.12573 third_party folder.