Use C11 atomics instead of atomic_increment(_val)

Replace atomic_increment and atomic_increment_val with atomic_fetch_add_relaxed.
One case in sem_post.c uses release semantics (see comment above it).
The others are simple counters and do not protect any shared data from
concurrent accesses.

Passes regress on AArch64.

Reviewed-by: Adhemerval Zanella  <adhemerval.zanella@linaro.org>

htl/pt-create.c

@@ -228,7 +228,7 @@ __pthread_create_internal (struct __pthread **thread,
      the number of threads from within the new thread isn't an option
      since this thread might return and call `pthread_exit' before the
      new thread runs.  */
-  atomic_increment (&__pthread_total);
+  atomic_fetch_add_relaxed (&__pthread_total, 1);
 
   /* Store a pointer to this thread in the thread ID lookup table.  We
      could use __thread_setid, however, we only lock for reading as no

manual/ipc.texi

@@ -85,7 +85,7 @@ by @theglibc{}.
 
 @deftypefun int sem_wait (sem_t *@var{sem});
 @safety{@prelim{}@mtsafe{}@assafe{}@acunsafe{@acucorrupt{}}}
-@c atomic_increment (nwaiters) acucorrupt
+@c atomic_fetch_add_relaxed (nwaiters) acucorrupt
 @c
 @c Given the use atomic operations this function seems
 @c to be AS-safe.  It is AC-unsafe because there is still

manual/llio.texi

@@ -2528,7 +2528,7 @@ aiocb64}, since the LFS transparently replaces the old interface.
 @c      _dl_allocate_tls_init ok
 @c       GET_DTV ok
 @c     mmap ok
-@c     atomic_increment_val ok
+@c     atomic_fetch_add_relaxed ok
 @c     munmap ok
 @c     change_stack_perm ok
 @c      mprotect ok
@@ -2567,7 +2567,7 @@ aiocb64}, since the LFS transparently replaces the old interface.
 @c     do_clone @asulock @ascuheap @aculock @acsmem
 @c      PREPARE_CREATE ok
 @c      lll_lock (pd->lock) @asulock @aculock
-@c      atomic_increment ok
+@c      atomic_fetch_add_relaxed ok
 @c      clone ok
 @c      atomic_fetch_add_relaxed ok
 @c      atomic_exchange_acq ok

nptl/nptl_setxid.c

@@ -163,7 +163,7 @@ setxid_signal_thread (struct xid_command *cmdp, struct pthread *t)
   /* If this failed, it must have had not started yet or else exited.  */
   if (!INTERNAL_SYSCALL_ERROR_P (val))
     {
-      atomic_increment (&cmdp->cntr);
+      atomic_fetch_add_relaxed (&cmdp->cntr, 1);
       return 1;
     }
   else

nptl/pthread_create.c

@@ -759,7 +759,7 @@ __pthread_create_2_1 (pthread_t *newthread, const pthread_attr_t *attr,
      we momentarily store a false value; this doesn't matter because there
      is no kosher thing a signal handler interrupting us right here can do
      that cares whether the thread count is correct.  */
-  atomic_increment (&__nptl_nthreads);
+  atomic_fetch_add_relaxed (&__nptl_nthreads, 1);
 
   /* Our local value of stopped_start and thread_ran can be accessed at
      any time. The PD->stopped_start may only be accessed if we have

nptl/sem_post.c

@@ -91,7 +91,7 @@ __old_sem_post (sem_t *sem)
   /* We must need to synchronize with consumers of this token, so the atomic
      increment must have release MO semantics.  */
   atomic_write_barrier ();
-  (void) atomic_increment_val (futex);
+  atomic_fetch_add_release (futex, 1);
   /* We always have to assume it is a shared semaphore.  */
   futex_wake (futex, 1, LLL_SHARED);
   return 0;

nscd/cache.c

@@ -192,7 +192,7 @@ cache_add (int type, const void *key, size_t len, struct datahead *packet,
 
   /* We depend on this value being correct and at least as high as the
      real number of entries.  */
-  atomic_increment (&table->head->nentries);
+  atomic_fetch_add_relaxed (&table->head->nentries, 1);
 
   /* It does not matter that we are not loading the just increment
      value, this is just for statistics.  */

nscd/nscd_helper.c

@@ -425,7 +425,7 @@ __nscd_get_map_ref (request_type type, const char *name,
 				0))
 	    cur = NO_MAPPING;
 	  else
-	    atomic_increment (&cur->counter);
+	    atomic_fetch_add_relaxed (&cur->counter, 1);
 	}
     }
 

sysdeps/unix/sysv/linux/check_pf.c

@@ -72,8 +72,8 @@ static uint32_t nl_timestamp;
 uint32_t
 __bump_nl_timestamp (void)
 {
-  if (atomic_increment_val (&nl_timestamp) == 0)
-    atomic_increment (&nl_timestamp);
+  if (atomic_fetch_add_relaxed (&nl_timestamp, 1) + 1 == 0)
+    atomic_fetch_add_relaxed (&nl_timestamp, 1);
 
   return nl_timestamp;
 }
@@ -309,7 +309,7 @@ __check_pf (bool *seen_ipv4, bool *seen_ipv6,
   if (cache_valid_p ())
     {
       data = cache;
-      atomic_increment (&cache->usecnt);
+      atomic_fetch_add_relaxed (&cache->usecnt, 1);
     }
   else
     {