X-Git-Url: https://git.openpandora.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=ipc%2Fsem.c;h=bf1bc36cb7ee62bc95021ba62253d5b667802952;hb=f367422794315b7632f2a78c93e15b3c757d4be8;hp=d5ce4000ca17bebe0a16f8d158fd6dfe6ccdcb37;hpb=2c0c29d414087f3b021059673c20a7088f5f1fff;p=pandora-kernel.git

diff --git a/ipc/sem.c b/ipc/sem.c
index d5ce4000ca17..bf1bc36cb7ee 100644
--- a/ipc/sem.c
+++ b/ipc/sem.c
@@ -272,8 +272,7 @@ static int newary(struct ipc_namespace *ns, struct ipc_params *params)
 	ns->used_sems += nsems;
 
 	sma->sem_base = (struct sem *) &sma[1];
-	/* sma->sem_pending = NULL; */
-	sma->sem_pending_last = &sma->sem_pending;
+	INIT_LIST_HEAD(&sma->sem_pending);
 	INIT_LIST_HEAD(&sma->list_id);
 	sma->sem_nsems = nsems;
 	sma->sem_ctime = get_seconds();
@@ -331,38 +330,6 @@ asmlinkage long sys_semget(key_t key, int nsems, int semflg)
 	return ipcget(ns, &sem_ids(ns), &sem_ops, &sem_params);
 }
 
-/* Manage the doubly linked list sma->sem_pending as a FIFO:
- * insert new queue elements at the tail sma->sem_pending_last.
- */
-static inline void append_to_queue (struct sem_array * sma,
-				    struct sem_queue * q)
-{
-	*(q->prev = sma->sem_pending_last) = q;
-	*(sma->sem_pending_last = &q->next) = NULL;
-}
-
-static inline void prepend_to_queue (struct sem_array * sma,
-				     struct sem_queue * q)
-{
-	q->next = sma->sem_pending;
-	*(q->prev = &sma->sem_pending) = q;
-	if (q->next)
-		q->next->prev = &q->next;
-	else /* sma->sem_pending_last == &sma->sem_pending */
-		sma->sem_pending_last = &q->next;
-}
-
-static inline void remove_from_queue (struct sem_array * sma,
-				      struct sem_queue * q)
-{
-	*(q->prev) = q->next;
-	if (q->next)
-		q->next->prev = q->prev;
-	else /* sma->sem_pending_last == &q->next */
-		sma->sem_pending_last = q->prev;
-	q->prev = NULL; /* mark as removed */
-}
-
 /*
  * Determine whether a sequence of semaphore operations would succeed
  * all at once. Return 0 if yes, 1 if need to sleep, else return error code.
@@ -438,16 +405,15 @@ static void update_queue (struct sem_array * sma)
 	int error;
 	struct sem_queue * q;
 
-	q = sma->sem_pending;
-	while(q) {
+	q = list_entry(sma->sem_pending.next, struct sem_queue, list);
+	while (&q->list != &sma->sem_pending) {
 		error = try_atomic_semop(sma, q->sops, q->nsops,
 					 q->undo, q->pid);
 
 		/* Does q->sleeper still need to sleep? */
 		if (error <= 0) {
 			struct sem_queue *n;
-			remove_from_queue(sma,q);
-			q->status = IN_WAKEUP;
+
 			/*
 			 * Continue scanning. The next operation
 			 * that must be checked depends on the type of the
@@ -458,11 +424,26 @@ static void update_queue (struct sem_array * sma)
 			 *   for semaphore values to become 0.
 			 * - if the operation didn't modify the array,
 			 *   then just continue.
+			 * The order of list_del() and reading ->next
+			 * is crucial: In the former case, the list_del()
+			 * must be done first [because we might be the
+			 * first entry in ->sem_pending], in the latter
+			 * case the list_del() must be done last
+			 * [because the list is invalid after the list_del()]
 			 */
-			if (q->alter)
-				n = sma->sem_pending;
-			else
-				n = q->next;
+			if (q->alter) {
+				list_del(&q->list);
+				n = list_entry(sma->sem_pending.next,
+						struct sem_queue, list);
+			} else {
+				n = list_entry(q->list.next, struct sem_queue,
+						list);
+				list_del(&q->list);
+			}
+
+			/* wake up the waiting thread */
+			q->status = IN_WAKEUP;
+
 			wake_up_process(q->sleeper);
 			/* hands-off: q will disappear immediately after
 			 * writing q->status.
@@ -471,7 +452,7 @@ static void update_queue (struct sem_array * sma)
 			q->status = error;
 			q = n;
 		} else {
-			q = q->next;
+			q = list_entry(q->list.next, struct sem_queue, list);
 		}
 	}
 }
@@ -491,7 +472,7 @@ static int count_semncnt (struct sem_array * sma, ushort semnum)
 	struct sem_queue * q;
 
 	semncnt = 0;
-	for (q = sma->sem_pending; q; q = q->next) {
+	list_for_each_entry(q, &sma->sem_pending, list) {
 		struct sembuf * sops = q->sops;
 		int nsops = q->nsops;
 		int i;
@@ -503,13 +484,14 @@ static int count_semncnt (struct sem_array * sma, ushort semnum)
 	}
 	return semncnt;
 }
+
 static int count_semzcnt (struct sem_array * sma, ushort semnum)
 {
 	int semzcnt;
 	struct sem_queue * q;
 
 	semzcnt = 0;
-	for (q = sma->sem_pending; q; q = q->next) {
+	list_for_each_entry(q, &sma->sem_pending, list) {
 		struct sembuf * sops = q->sops;
 		int nsops = q->nsops;
 		int i;
@@ -522,36 +504,41 @@ static int count_semzcnt (struct sem_array * sma, ushort semnum)
 	return semzcnt;
 }
 
+void free_un(struct rcu_head *head)
+{
+	struct sem_undo *un = container_of(head, struct sem_undo, rcu);
+	kfree(un);
+}
+
 /* Free a semaphore set. freeary() is called with sem_ids.rw_mutex locked
  * as a writer and the spinlock for this semaphore set hold. sem_ids.rw_mutex
  * remains locked on exit.
  */
 static void freeary(struct ipc_namespace *ns, struct kern_ipc_perm *ipcp)
 {
-	struct sem_undo *un;
-	struct sem_queue *q;
+	struct sem_undo *un, *tu;
+	struct sem_queue *q, *tq;
 	struct sem_array *sma = container_of(ipcp, struct sem_array, sem_perm);
 
-	/* Invalidate the existing undo structures for this semaphore set.
-	 * (They will be freed without any further action in exit_sem()
-	 * or during the next semop.)
-	 */
+	/* Free the existing undo structures for this semaphore set.  */
 	assert_spin_locked(&sma->sem_perm.lock);
-	list_for_each_entry(un, &sma->list_id, list_id)
+	list_for_each_entry_safe(un, tu, &sma->list_id, list_id) {
+		list_del(&un->list_id);
+		spin_lock(&un->ulp->lock);
 		un->semid = -1;
+		list_del_rcu(&un->list_proc);
+		spin_unlock(&un->ulp->lock);
+		call_rcu(&un->rcu, free_un);
+	}
 
 	/* Wake up all pending processes and let them fail with EIDRM. */
-	q = sma->sem_pending;
-	while(q) {
-		struct sem_queue *n;
-		/* lazy remove_from_queue: we are killing the whole queue */
-		q->prev = NULL;
-		n = q->next;
+	list_for_each_entry_safe(q, tq, &sma->sem_pending, list) {
+		list_del(&q->list);
+
 		q->status = IN_WAKEUP;
 		wake_up_process(q->sleeper); /* doesn't sleep */
 		smp_wmb();
 		q->status = -EIDRM;	/* hands-off q */
-		q = n;
 	}
 
 	/* Remove the semaphore set from the IDR */
@@ -969,16 +956,11 @@ static inline int get_undo_list(struct sem_undo_list **undo_listp)
 
 static struct sem_undo *lookup_undo(struct sem_undo_list *ulp, int semid)
 {
-	struct sem_undo *walk, *tmp;
+	struct sem_undo *walk;
 
-	assert_spin_locked(&ulp->lock);
-	list_for_each_entry_safe(walk, tmp, &ulp->list_proc, list_proc) {
+	list_for_each_entry_rcu(walk, &ulp->list_proc, list_proc) {
 		if (walk->semid == semid)
 			return walk;
-		if (walk->semid == -1) {
-			list_del(&walk->list_proc);
-			kfree(walk);
-		}
 	}
 	return NULL;
 }
@@ -991,6 +973,8 @@ static struct sem_undo *lookup_undo(struct sem_undo_list *ulp, int semid)
  * The function looks up (and if not present creates) the undo structure.
  * The size of the undo structure depends on the size of the semaphore
  * array, thus the alloc path is not that straightforward.
+ * Lifetime-rules: sem_undo is rcu-protected, on success, the function
+ * performs a rcu_read_lock().
  */
 static struct sem_undo *find_alloc_undo(struct ipc_namespace *ns, int semid)
 {
@@ -1004,11 +988,13 @@ static struct sem_undo *find_alloc_undo(struct ipc_namespace *ns, int semid)
 	if (error)
 		return ERR_PTR(error);
 
+	rcu_read_lock();
 	spin_lock(&ulp->lock);
 	un = lookup_undo(ulp, semid);
 	spin_unlock(&ulp->lock);
 	if (likely(un!=NULL))
 		goto out;
+	rcu_read_unlock();
 
 	/* no undo structure around - allocate one. */
 	/* step 1: figure out the size of the semaphore array */
@@ -1026,38 +1012,38 @@ static struct sem_undo *find_alloc_undo(struct ipc_namespace *ns, int semid)
 		return ERR_PTR(-ENOMEM);
 	}
 
-	/* step 3: Acquire the lock on the undo list pointer */
-	spin_lock(&ulp->lock);
-
-	/* step 4: check for races: someone else allocated the undo struct,
-	 *         semaphore array was destroyed.
-	 */
-	un = lookup_undo(ulp, semid);
-	if (un) {
-		spin_unlock(&ulp->lock);
-		kfree(new);
-		sem_putref(sma);
-		goto out;
-	}
+	/* step 3: Acquire the lock on semaphore array */
 	sem_lock_and_putref(sma);
 	if (sma->sem_perm.deleted) {
 		sem_unlock(sma);
-		spin_unlock(&ulp->lock);
 		kfree(new);
 		un = ERR_PTR(-EIDRM);
 		goto out;
 	}
+	spin_lock(&ulp->lock);
+
+	/*
+	 * step 4: check for races: did someone else allocate the undo struct?
+	 */
+	un = lookup_undo(ulp, semid);
+	if (un) {
+		kfree(new);
+		goto success;
+	}
 	/* step 5: initialize & link new undo structure */
 	new->semadj = (short *) &new[1];
+	new->ulp = ulp;
 	new->semid = semid;
 	assert_spin_locked(&ulp->lock);
-	list_add(&new->list_proc, &ulp->list_proc);
+	list_add_rcu(&new->list_proc, &ulp->list_proc);
 	assert_spin_locked(&sma->sem_perm.lock);
 	list_add(&new->list_id, &sma->list_id);
+	un = new;
 
-	sem_unlock(sma);
+success:
 	spin_unlock(&ulp->lock);
-	un = new;
+	rcu_read_lock();
+	sem_unlock(sma);
 out:
 	return un;
 }
@@ -1124,6 +1110,8 @@ asmlinkage long sys_semtimedop(int semid, struct sembuf __user *tsops,
 
 	sma = sem_lock_check(ns, semid);
 	if (IS_ERR(sma)) {
+		if (un)
+			rcu_read_unlock();
 		error = PTR_ERR(sma);
 		goto out_free;
 	}
@@ -1132,10 +1120,26 @@ asmlinkage long sys_semtimedop(int semid, struct sembuf __user *tsops,
 	 * semid identifiers are not unique - find_alloc_undo may have
 	 * allocated an undo structure, it was invalidated by an RMID
 	 * and now a new array with received the same id. Check and fail.
+	 * This case can be detected checking un->semid. The existance of
+	 * "un" itself is guaranteed by rcu.
 	 */
 	error = -EIDRM;
-	if (un && un->semid == -1)
-		goto out_unlock_free;
+	if (un) {
+		if (un->semid == -1) {
+			rcu_read_unlock();
+			goto out_unlock_free;
+		} else {
+			/*
+			 * rcu lock can be released, "un" cannot disappear:
+			 * - sem_lock is acquired, thus IPC_RMID is
+			 *   impossible.
+			 * - exit_sem is impossible, it always operates on
+			 *   current (or a dead task).
+			 */
+
+			rcu_read_unlock();
+		}
+	}
 
 	error = -EFBIG;
 	if (max >= sma->sem_nsems)
@@ -1166,9 +1170,9 @@ asmlinkage long sys_semtimedop(int semid, struct sembuf __user *tsops,
 	queue.pid = task_tgid_vnr(current);
 	queue.alter = alter;
 	if (alter)
-		append_to_queue(sma ,&queue);
+		list_add_tail(&queue.list, &sma->sem_pending);
 	else
-		prepend_to_queue(sma ,&queue);
+		list_add(&queue.list, &sma->sem_pending);
 
 	queue.status = -EINTR;
 	queue.sleeper = current;
@@ -1194,7 +1198,6 @@ asmlinkage long sys_semtimedop(int semid, struct sembuf __user *tsops,
 
 	sma = sem_lock(ns, semid);
 	if (IS_ERR(sma)) {
-		BUG_ON(queue.prev != NULL);
 		error = -EIDRM;
 		goto out_free;
 	}
@@ -1212,7 +1215,7 @@ asmlinkage long sys_semtimedop(int semid, struct sembuf __user *tsops,
 	 */
 	if (timeout && jiffies_left == 0)
 		error = -EAGAIN;
-	remove_from_queue(sma,&queue);
+	list_del(&queue.list);
 	goto out_unlock_free;
 
 out_unlock_free:
@@ -1264,7 +1267,6 @@ int copy_semundo(unsigned long clone_flags, struct task_struct *tsk)
 void exit_sem(struct task_struct *tsk)
 {
 	struct sem_undo_list *ulp;
-	struct sem_undo *un, *tmp;
 
 	ulp = tsk->sysvsem.undo_list;
 	if (!ulp)
@@ -1274,28 +1276,47 @@ void exit_sem(struct task_struct *tsk)
 	if (!atomic_dec_and_test(&ulp->refcnt))
 		return;
 
-	spin_lock(&ulp->lock);
-
-	list_for_each_entry_safe(un, tmp, &ulp->list_proc, list_proc) {
+	for (;;) {
 		struct sem_array *sma;
+		struct sem_undo *un;
+		int semid;
 		int i;
 
-		if (un->semid == -1)
-			goto free;
+		rcu_read_lock();
+		un = list_entry(rcu_dereference(ulp->list_proc.next),
+					struct sem_undo, list_proc);
+		if (&un->list_proc == &ulp->list_proc)
+			semid = -1;
+		 else
+			semid = un->semid;
+		rcu_read_unlock();
 
-		sma = sem_lock(tsk->nsproxy->ipc_ns, un->semid);
-		if (IS_ERR(sma))
-			goto free;
+		if (semid == -1)
+			break;
 
-		if (un->semid == -1)
-			goto unlock_free;
+		sma = sem_lock_check(tsk->nsproxy->ipc_ns, un->semid);
 
-		BUG_ON(sem_checkid(sma, un->semid));
+		/* exit_sem raced with IPC_RMID, nothing to do */
+		if (IS_ERR(sma))
+			continue;
 
-		/* remove un from sma->list_id */
+		un = lookup_undo(ulp, semid);
+		if (un == NULL) {
+			/* exit_sem raced with IPC_RMID+semget() that created
+			 * exactly the same semid. Nothing to do.
+			 */
+			sem_unlock(sma);
+			continue;
+		}
+
+		/* remove un from the linked lists */
 		assert_spin_locked(&sma->sem_perm.lock);
 		list_del(&un->list_id);
 
+		spin_lock(&ulp->lock);
+		list_del_rcu(&un->list_proc);
+		spin_unlock(&ulp->lock);
+
 		/* perform adjustments registered in un */
 		for (i = 0; i < sma->sem_nsems; i++) {
 			struct sem * semaphore = &sma->sem_base[i];
@@ -1324,14 +1345,10 @@ void exit_sem(struct task_struct *tsk)
 		sma->sem_otime = get_seconds();
 		/* maybe some queued-up processes were waiting for this */
 		update_queue(sma);
-unlock_free:
 		sem_unlock(sma);
-free:
-		assert_spin_locked(&ulp->lock);
-		list_del(&un->list_proc);
-		kfree(un);
+
+		call_rcu(&un->rcu, free_un);
 	}
-	spin_unlock(&ulp->lock);
 	kfree(ulp);
 }