af_iucv: Return -EAGAIN if iucv msg limit is exceeded
Hendrik Brueckner [Wed, 17 Jun 2009 21:54:48 +0000 (21:54 +0000)]
If the iucv message limit for a communication path is exceeded,
sendmsg() returns -EAGAIN instead of -EPIPE.
The calling application can then handle this error situtation,
e.g. to try again after waiting some time.

For blocking sockets, sendmsg() waits up to the socket timeout
before returning -EAGAIN. For the new wait condition, a macro
has been introduced and the iucv_sock_wait_state() has been
refactored to this macro.

Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

include/net/iucv/af_iucv.h
net/iucv/af_iucv.c

index 21ee49f..f82a1e8 100644 (file)
@@ -94,8 +94,6 @@ unsigned int iucv_sock_poll(struct file *file, struct socket *sock,
                            poll_table *wait);
 void iucv_sock_link(struct iucv_sock_list *l, struct sock *s);
 void iucv_sock_unlink(struct iucv_sock_list *l, struct sock *s);
-int  iucv_sock_wait_state(struct sock *sk, int state, int state2,
-                         unsigned long timeo);
 int  iucv_sock_wait_cnt(struct sock *sk, unsigned long timeo);
 void iucv_accept_enqueue(struct sock *parent, struct sock *sk);
 void iucv_accept_unlink(struct sock *sk);
index 42b7198..abadb4a 100644 (file)
@@ -53,6 +53,38 @@ static const u8 iprm_shutdown[8] =
 #define CB_TRGCLS(skb) ((skb)->cb + CB_TAG_LEN) /* iucv msg target class */
 #define CB_TRGCLS_LEN  (TRGCLS_SIZE)
 
+#define __iucv_sock_wait(sk, condition, timeo, ret)                    \
+do {                                                                   \
+       DEFINE_WAIT(__wait);                                            \
+       long __timeo = timeo;                                           \
+       ret = 0;                                                        \
+       while (!(condition)) {                                          \
+               prepare_to_wait(sk->sk_sleep, &__wait, TASK_INTERRUPTIBLE); \
+               if (!__timeo) {                                         \
+                       ret = -EAGAIN;                                  \
+                       break;                                          \
+               }                                                       \
+               if (signal_pending(current)) {                          \
+                       ret = sock_intr_errno(__timeo);                 \
+                       break;                                          \
+               }                                                       \
+               release_sock(sk);                                       \
+               __timeo = schedule_timeout(__timeo);                    \
+               lock_sock(sk);                                          \
+               ret = sock_error(sk);                                   \
+               if (ret)                                                \
+                       break;                                          \
+       }                                                               \
+       finish_wait(sk->sk_sleep, &__wait);                             \
+} while (0)
+
+#define iucv_sock_wait(sk, condition, timeo)                           \
+({                                                                     \
+       int __ret = 0;                                                  \
+       if (!(condition))                                               \
+               __iucv_sock_wait(sk, condition, timeo, __ret);          \
+       __ret;                                                          \
+})
 
 static void iucv_sock_kill(struct sock *sk);
 static void iucv_sock_close(struct sock *sk);
@@ -121,6 +153,48 @@ static inline size_t iucv_msg_length(struct iucv_message *msg)
        return msg->length;
 }
 
+/**
+ * iucv_sock_in_state() - check for specific states
+ * @sk:                sock structure
+ * @state:     first iucv sk state
+ * @state:     second iucv sk state
+ *
+ * Returns true if the socket in either in the first or second state.
+ */
+static int iucv_sock_in_state(struct sock *sk, int state, int state2)
+{
+       return (sk->sk_state == state || sk->sk_state == state2);
+}
+
+/**
+ * iucv_below_msglim() - function to check if messages can be sent
+ * @sk:                sock structure
+ *
+ * Returns true if the send queue length is lower than the message limit.
+ * Always returns true if the socket is not connected (no iucv path for
+ * checking the message limit).
+ */
+static inline int iucv_below_msglim(struct sock *sk)
+{
+       struct iucv_sock *iucv = iucv_sk(sk);
+
+       if (sk->sk_state != IUCV_CONNECTED)
+               return 1;
+       return (skb_queue_len(&iucv->send_skb_q) < iucv->path->msglim);
+}
+
+/**
+ * iucv_sock_wake_msglim() - Wake up thread waiting on msg limit
+ */
+static void iucv_sock_wake_msglim(struct sock *sk)
+{
+       read_lock(&sk->sk_callback_lock);
+       if (sk->sk_sleep && waitqueue_active(sk->sk_sleep))
+               wake_up_interruptible_all(sk->sk_sleep);
+       sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
+       read_unlock(&sk->sk_callback_lock);
+}
+
 /* Timers */
 static void iucv_sock_timeout(unsigned long arg)
 {
@@ -212,7 +286,9 @@ static void iucv_sock_close(struct sock *sk)
                                timeo = sk->sk_lingertime;
                        else
                                timeo = IUCV_DISCONN_TIMEOUT;
-                       err = iucv_sock_wait_state(sk, IUCV_CLOSED, 0, timeo);
+                       err = iucv_sock_wait(sk,
+                                       iucv_sock_in_state(sk, IUCV_CLOSED, 0),
+                                       timeo);
                }
 
        case IUCV_CLOSING:   /* fall through */
@@ -393,39 +469,6 @@ struct sock *iucv_accept_dequeue(struct sock *parent, struct socket *newsock)
        return NULL;
 }
 
-int iucv_sock_wait_state(struct sock *sk, int state, int state2,
-                        unsigned long timeo)
-{
-       DECLARE_WAITQUEUE(wait, current);
-       int err = 0;
-
-       add_wait_queue(sk->sk_sleep, &wait);
-       while (sk->sk_state != state && sk->sk_state != state2) {
-               set_current_state(TASK_INTERRUPTIBLE);
-
-               if (!timeo) {
-                       err = -EAGAIN;
-                       break;
-               }
-
-               if (signal_pending(current)) {
-                       err = sock_intr_errno(timeo);
-                       break;
-               }
-
-               release_sock(sk);
-               timeo = schedule_timeout(timeo);
-               lock_sock(sk);
-
-               err = sock_error(sk);
-               if (err)
-                       break;
-       }
-       set_current_state(TASK_RUNNING);
-       remove_wait_queue(sk->sk_sleep, &wait);
-       return err;
-}
-
 /* Bind an unbound socket */
 static int iucv_sock_bind(struct socket *sock, struct sockaddr *addr,
                          int addr_len)
@@ -570,8 +613,9 @@ static int iucv_sock_connect(struct socket *sock, struct sockaddr *addr,
        }
 
        if (sk->sk_state != IUCV_CONNECTED) {
-               err = iucv_sock_wait_state(sk, IUCV_CONNECTED, IUCV_DISCONN,
-                               sock_sndtimeo(sk, flags & O_NONBLOCK));
+               err = iucv_sock_wait(sk, iucv_sock_in_state(sk, IUCV_CONNECTED,
+                                                           IUCV_DISCONN),
+                                    sock_sndtimeo(sk, flags & O_NONBLOCK));
        }
 
        if (sk->sk_state == IUCV_DISCONN) {
@@ -725,9 +769,11 @@ static int iucv_sock_sendmsg(struct kiocb *iocb, struct socket *sock,
        struct iucv_message txmsg;
        struct cmsghdr *cmsg;
        int cmsg_done;
+       long timeo;
        char user_id[9];
        char appl_id[9];
        int err;
+       int noblock = msg->msg_flags & MSG_DONTWAIT;
 
        err = sock_error(sk);
        if (err)
@@ -799,8 +845,7 @@ static int iucv_sock_sendmsg(struct kiocb *iocb, struct socket *sock,
         * this is fine for SOCK_SEQPACKET (unless we want to support
         * segmented records using the MSG_EOR flag), but
         * for SOCK_STREAM we might want to improve it in future */
-       skb = sock_alloc_send_skb(sk, len, msg->msg_flags & MSG_DONTWAIT,
-                                 &err);
+       skb = sock_alloc_send_skb(sk, len, noblock, &err);
        if (!skb)
                goto out;
        if (memcpy_fromiovec(skb_put(skb, len), msg->msg_iov, len)) {
@@ -808,6 +853,18 @@ static int iucv_sock_sendmsg(struct kiocb *iocb, struct socket *sock,
                goto fail;
        }
 
+       /* wait if outstanding messages for iucv path has reached */
+       timeo = sock_sndtimeo(sk, noblock);
+       err = iucv_sock_wait(sk, iucv_below_msglim(sk), timeo);
+       if (err)
+               goto fail;
+
+       /* return -ECONNRESET if the socket is no longer connected */
+       if (sk->sk_state != IUCV_CONNECTED) {
+               err = -ECONNRESET;
+               goto fail;
+       }
+
        /* increment and save iucv message tag for msg_completion cbk */
        txmsg.tag = iucv->send_tag++;
        memcpy(CB_TAG(skb), &txmsg.tag, CB_TAG_LEN);
@@ -844,9 +901,10 @@ static int iucv_sock_sendmsg(struct kiocb *iocb, struct socket *sock,
                        pr_err("Application %s on z/VM guest %s"
                                " exceeds message limit\n",
                                appl_id, user_id);
-               }
+                       err = -EAGAIN;
+               } else
+                       err = -EPIPE;
                skb_unlink(skb, &iucv->send_skb_q);
-               err = -EPIPE;
                goto fail;
        }
 
@@ -1463,7 +1521,11 @@ static void iucv_callback_txdone(struct iucv_path *path,
 
                spin_unlock_irqrestore(&list->lock, flags);
 
-               kfree_skb(this);
+               if (this) {
+                       kfree_skb(this);
+                       /* wake up any process waiting for sending */
+                       iucv_sock_wake_msglim(sk);
+               }
        }
        BUG_ON(!this);