aboutsummaryrefslogtreecommitdiff
path: root/net/unix
diff options
context:
space:
mode:
Diffstat (limited to 'net/unix')
-rw-r--r--net/unix/af_unix.c148
-rw-r--r--net/unix/garbage.c325
2 files changed, 279 insertions, 194 deletions
diff --git a/net/unix/af_unix.c b/net/unix/af_unix.c
index fc12ba51c1fc..65ebccc0a698 100644
--- a/net/unix/af_unix.c
+++ b/net/unix/af_unix.c
@@ -174,11 +174,11 @@ static struct sock *unix_peer_get(struct sock *s)
{
struct sock *peer;
- unix_state_rlock(s);
+ unix_state_lock(s);
peer = unix_peer(s);
if (peer)
sock_hold(peer);
- unix_state_runlock(s);
+ unix_state_unlock(s);
return peer;
}
@@ -369,7 +369,7 @@ static int unix_release_sock (struct sock *sk, int embrion)
unix_remove_socket(sk);
/* Clear state */
- unix_state_wlock(sk);
+ unix_state_lock(sk);
sock_orphan(sk);
sk->sk_shutdown = SHUTDOWN_MASK;
dentry = u->dentry;
@@ -378,7 +378,7 @@ static int unix_release_sock (struct sock *sk, int embrion)
u->mnt = NULL;
state = sk->sk_state;
sk->sk_state = TCP_CLOSE;
- unix_state_wunlock(sk);
+ unix_state_unlock(sk);
wake_up_interruptible_all(&u->peer_wait);
@@ -386,12 +386,12 @@ static int unix_release_sock (struct sock *sk, int embrion)
if (skpair!=NULL) {
if (sk->sk_type == SOCK_STREAM || sk->sk_type == SOCK_SEQPACKET) {
- unix_state_wlock(skpair);
+ unix_state_lock(skpair);
/* No more writes */
skpair->sk_shutdown = SHUTDOWN_MASK;
if (!skb_queue_empty(&sk->sk_receive_queue) || embrion)
skpair->sk_err = ECONNRESET;
- unix_state_wunlock(skpair);
+ unix_state_unlock(skpair);
skpair->sk_state_change(skpair);
read_lock(&skpair->sk_callback_lock);
sk_wake_async(skpair,1,POLL_HUP);
@@ -448,7 +448,7 @@ static int unix_listen(struct socket *sock, int backlog)
err = -EINVAL;
if (!u->addr)
goto out; /* No listens on an unbound socket */
- unix_state_wlock(sk);
+ unix_state_lock(sk);
if (sk->sk_state != TCP_CLOSE && sk->sk_state != TCP_LISTEN)
goto out_unlock;
if (backlog > sk->sk_max_ack_backlog)
@@ -462,7 +462,7 @@ static int unix_listen(struct socket *sock, int backlog)
err = 0;
out_unlock:
- unix_state_wunlock(sk);
+ unix_state_unlock(sk);
out:
return err;
}
@@ -592,7 +592,8 @@ static struct sock * unix_create1(struct socket *sock)
u->dentry = NULL;
u->mnt = NULL;
spin_lock_init(&u->lock);
- atomic_set(&u->inflight, sock ? 0 : -1);
+ atomic_set(&u->inflight, 0);
+ INIT_LIST_HEAD(&u->link);
mutex_init(&u->readlock); /* single task reading lock */
init_waitqueue_head(&u->peer_wait);
unix_insert_socket(unix_sockets_unbound, sk);
@@ -858,6 +859,31 @@ out_mknod_parent:
goto out_up;
}
+static void unix_state_double_lock(struct sock *sk1, struct sock *sk2)
+{
+ if (unlikely(sk1 == sk2) || !sk2) {
+ unix_state_lock(sk1);
+ return;
+ }
+ if (sk1 < sk2) {
+ unix_state_lock(sk1);
+ unix_state_lock_nested(sk2);
+ } else {
+ unix_state_lock(sk2);
+ unix_state_lock_nested(sk1);
+ }
+}
+
+static void unix_state_double_unlock(struct sock *sk1, struct sock *sk2)
+{
+ if (unlikely(sk1 == sk2) || !sk2) {
+ unix_state_unlock(sk1);
+ return;
+ }
+ unix_state_unlock(sk1);
+ unix_state_unlock(sk2);
+}
+
static int unix_dgram_connect(struct socket *sock, struct sockaddr *addr,
int alen, int flags)
{
@@ -877,11 +903,19 @@ static int unix_dgram_connect(struct socket *sock, struct sockaddr *addr,
!unix_sk(sk)->addr && (err = unix_autobind(sock)) != 0)
goto out;
+restart:
other=unix_find_other(sunaddr, alen, sock->type, hash, &err);
if (!other)
goto out;
- unix_state_wlock(sk);
+ unix_state_double_lock(sk, other);
+
+ /* Apparently VFS overslept socket death. Retry. */
+ if (sock_flag(other, SOCK_DEAD)) {
+ unix_state_double_unlock(sk, other);
+ sock_put(other);
+ goto restart;
+ }
err = -EPERM;
if (!unix_may_send(sk, other))
@@ -896,7 +930,7 @@ static int unix_dgram_connect(struct socket *sock, struct sockaddr *addr,
* 1003.1g breaking connected state with AF_UNSPEC
*/
other = NULL;
- unix_state_wlock(sk);
+ unix_state_double_lock(sk, other);
}
/*
@@ -905,19 +939,19 @@ static int unix_dgram_connect(struct socket *sock, struct sockaddr *addr,
if (unix_peer(sk)) {
struct sock *old_peer = unix_peer(sk);
unix_peer(sk)=other;
- unix_state_wunlock(sk);
+ unix_state_double_unlock(sk, other);
if (other != old_peer)
unix_dgram_disconnected(sk, old_peer);
sock_put(old_peer);
} else {
unix_peer(sk)=other;
- unix_state_wunlock(sk);
+ unix_state_double_unlock(sk, other);
}
return 0;
out_unlock:
- unix_state_wunlock(sk);
+ unix_state_double_unlock(sk, other);
sock_put(other);
out:
return err;
@@ -936,7 +970,7 @@ static long unix_wait_for_peer(struct sock *other, long timeo)
(skb_queue_len(&other->sk_receive_queue) >
other->sk_max_ack_backlog);
- unix_state_runlock(other);
+ unix_state_unlock(other);
if (sched)
timeo = schedule_timeout(timeo);
@@ -994,11 +1028,11 @@ restart:
goto out;
/* Latch state of peer */
- unix_state_rlock(other);
+ unix_state_lock(other);
/* Apparently VFS overslept socket death. Retry. */
if (sock_flag(other, SOCK_DEAD)) {
- unix_state_runlock(other);
+ unix_state_unlock(other);
sock_put(other);
goto restart;
}
@@ -1048,18 +1082,18 @@ restart:
goto out_unlock;
}
- unix_state_wlock_nested(sk);
+ unix_state_lock_nested(sk);
if (sk->sk_state != st) {
- unix_state_wunlock(sk);
- unix_state_runlock(other);
+ unix_state_unlock(sk);
+ unix_state_unlock(other);
sock_put(other);
goto restart;
}
err = security_unix_stream_connect(sock, other->sk_socket, newsk);
if (err) {
- unix_state_wunlock(sk);
+ unix_state_unlock(sk);
goto out_unlock;
}
@@ -1096,23 +1130,20 @@ restart:
smp_mb__after_atomic_inc(); /* sock_hold() does an atomic_inc() */
unix_peer(sk) = newsk;
- unix_state_wunlock(sk);
+ unix_state_unlock(sk);
/* take ten and and send info to listening sock */
spin_lock(&other->sk_receive_queue.lock);
__skb_queue_tail(&other->sk_receive_queue, skb);
- /* Undo artificially decreased inflight after embrion
- * is installed to listening socket. */
- atomic_inc(&newu->inflight);
spin_unlock(&other->sk_receive_queue.lock);
- unix_state_runlock(other);
+ unix_state_unlock(other);
other->sk_data_ready(other, 0);
sock_put(other);
return 0;
out_unlock:
if (other)
- unix_state_runlock(other);
+ unix_state_unlock(other);
out:
if (skb)
@@ -1178,10 +1209,10 @@ static int unix_accept(struct socket *sock, struct socket *newsock, int flags)
wake_up_interruptible(&unix_sk(sk)->peer_wait);
/* attach accepted sock to socket */
- unix_state_wlock(tsk);
+ unix_state_lock(tsk);
newsock->state = SS_CONNECTED;
sock_graft(tsk, newsock);
- unix_state_wunlock(tsk);
+ unix_state_unlock(tsk);
return 0;
out:
@@ -1208,7 +1239,7 @@ static int unix_getname(struct socket *sock, struct sockaddr *uaddr, int *uaddr_
}
u = unix_sk(sk);
- unix_state_rlock(sk);
+ unix_state_lock(sk);
if (!u->addr) {
sunaddr->sun_family = AF_UNIX;
sunaddr->sun_path[0] = 0;
@@ -1219,7 +1250,7 @@ static int unix_getname(struct socket *sock, struct sockaddr *uaddr, int *uaddr_
*uaddr_len = addr->len;
memcpy(sunaddr, addr->name, *uaddr_len);
}
- unix_state_runlock(sk);
+ unix_state_unlock(sk);
sock_put(sk);
out:
return err;
@@ -1337,7 +1368,7 @@ restart:
goto out_free;
}
- unix_state_rlock(other);
+ unix_state_lock(other);
err = -EPERM;
if (!unix_may_send(sk, other))
goto out_unlock;
@@ -1347,20 +1378,20 @@ restart:
* Check with 1003.1g - what should
* datagram error
*/
- unix_state_runlock(other);
+ unix_state_unlock(other);
sock_put(other);
err = 0;
- unix_state_wlock(sk);
+ unix_state_lock(sk);
if (unix_peer(sk) == other) {
unix_peer(sk)=NULL;
- unix_state_wunlock(sk);
+ unix_state_unlock(sk);
unix_dgram_disconnected(sk, other);
sock_put(other);
err = -ECONNREFUSED;
} else {
- unix_state_wunlock(sk);
+ unix_state_unlock(sk);
}
other = NULL;
@@ -1397,14 +1428,14 @@ restart:
}
skb_queue_tail(&other->sk_receive_queue, skb);
- unix_state_runlock(other);
+ unix_state_unlock(other);
other->sk_data_ready(other, len);
sock_put(other);
scm_destroy(siocb->scm);
return len;
out_unlock:
- unix_state_runlock(other);
+ unix_state_unlock(other);
out_free:
kfree_skb(skb);
out:
@@ -1494,14 +1525,14 @@ static int unix_stream_sendmsg(struct kiocb *kiocb, struct socket *sock,
goto out_err;
}
- unix_state_rlock(other);
+ unix_state_lock(other);
if (sock_flag(other, SOCK_DEAD) ||
(other->sk_shutdown & RCV_SHUTDOWN))
goto pipe_err_free;
skb_queue_tail(&other->sk_receive_queue, skb);
- unix_state_runlock(other);
+ unix_state_unlock(other);
other->sk_data_ready(other, size);
sent+=size;
}
@@ -1512,7 +1543,7 @@ static int unix_stream_sendmsg(struct kiocb *kiocb, struct socket *sock,
return sent;
pipe_err_free:
- unix_state_runlock(other);
+ unix_state_unlock(other);
kfree_skb(skb);
pipe_err:
if (sent==0 && !(msg->msg_flags&MSG_NOSIGNAL))
@@ -1641,7 +1672,7 @@ static long unix_stream_data_wait(struct sock * sk, long timeo)
{
DEFINE_WAIT(wait);
- unix_state_rlock(sk);
+ unix_state_lock(sk);
for (;;) {
prepare_to_wait(sk->sk_sleep, &wait, TASK_INTERRUPTIBLE);
@@ -1654,14 +1685,14 @@ static long unix_stream_data_wait(struct sock * sk, long timeo)
break;
set_bit(SOCK_ASYNC_WAITDATA, &sk->sk_socket->flags);
- unix_state_runlock(sk);
+ unix_state_unlock(sk);
timeo = schedule_timeout(timeo);
- unix_state_rlock(sk);
+ unix_state_lock(sk);
clear_bit(SOCK_ASYNC_WAITDATA, &sk->sk_socket->flags);
}
finish_wait(sk->sk_sleep, &wait);
- unix_state_runlock(sk);
+ unix_state_unlock(sk);
return timeo;
}
@@ -1711,20 +1742,23 @@ static int unix_stream_recvmsg(struct kiocb *iocb, struct socket *sock,
int chunk;
struct sk_buff *skb;
+ unix_state_lock(sk);
skb = skb_dequeue(&sk->sk_receive_queue);
if (skb==NULL)
{
if (copied >= target)
- break;
+ goto unlock;
/*
* POSIX 1003.1g mandates this order.
*/
if ((err = sock_error(sk)) != 0)
- break;
+ goto unlock;
if (sk->sk_shutdown & RCV_SHUTDOWN)
- break;
+ goto unlock;
+
+ unix_state_unlock(sk);
err = -EAGAIN;
if (!timeo)
break;
@@ -1738,7 +1772,11 @@ static int unix_stream_recvmsg(struct kiocb *iocb, struct socket *sock,
}
mutex_lock(&u->readlock);
continue;
+ unlock:
+ unix_state_unlock(sk);
+ break;
}
+ unix_state_unlock(sk);
if (check_creds) {
/* Never glue messages from different writers */
@@ -1816,12 +1854,12 @@ static int unix_shutdown(struct socket *sock, int mode)
mode = (mode+1)&(RCV_SHUTDOWN|SEND_SHUTDOWN);
if (mode) {
- unix_state_wlock(sk);
+ unix_state_lock(sk);
sk->sk_shutdown |= mode;
other=unix_peer(sk);
if (other)
sock_hold(other);
- unix_state_wunlock(sk);
+ unix_state_unlock(sk);
sk->sk_state_change(sk);
if (other &&
@@ -1833,9 +1871,9 @@ static int unix_shutdown(struct socket *sock, int mode)
peer_mode |= SEND_SHUTDOWN;
if (mode&SEND_SHUTDOWN)
peer_mode |= RCV_SHUTDOWN;
- unix_state_wlock(other);
+ unix_state_lock(other);
other->sk_shutdown |= peer_mode;
- unix_state_wunlock(other);
+ unix_state_unlock(other);
other->sk_state_change(other);
read_lock(&other->sk_callback_lock);
if (peer_mode == SHUTDOWN_MASK)
@@ -1973,7 +2011,7 @@ static int unix_seq_show(struct seq_file *seq, void *v)
else {
struct sock *s = v;
struct unix_sock *u = unix_sk(s);
- unix_state_rlock(s);
+ unix_state_lock(s);
seq_printf(seq, "%p: %08X %08X %08X %04X %02X %5lu",
s,
@@ -2001,14 +2039,14 @@ static int unix_seq_show(struct seq_file *seq, void *v)
for ( ; i < len; i++)
seq_putc(seq, u->addr->name->sun_path[i]);
}
- unix_state_runlock(s);
+ unix_state_unlock(s);
seq_putc(seq, '\n');
}
return 0;
}
-static struct seq_operations unix_seq_ops = {
+static const struct seq_operations unix_seq_ops = {
.start = unix_seq_start,
.next = unix_seq_next,
.stop = unix_seq_stop,
diff --git a/net/unix/garbage.c b/net/unix/garbage.c
index f20b7ea7c555..406b6433e467 100644
--- a/net/unix/garbage.c
+++ b/net/unix/garbage.c
@@ -62,6 +62,10 @@
* AV 1 Mar 1999
* Damn. Added missing check for ->dead in listen queues scanning.
*
+ * Miklos Szeredi 25 Jun 2007
+ * Reimplement with a cycle collecting algorithm. This should
+ * solve several problems with the previous code, like being racy
+ * wrt receive and holding up unrelated socket operations.
*/
#include <linux/kernel.h>
@@ -84,10 +88,9 @@
/* Internal data structures and random procedures: */
-#define GC_HEAD ((struct sock *)(-1))
-#define GC_ORPHAN ((struct sock *)(-3))
-
-static struct sock *gc_current = GC_HEAD; /* stack of objects to mark */
+static LIST_HEAD(gc_inflight_list);
+static LIST_HEAD(gc_candidates);
+static DEFINE_SPINLOCK(unix_gc_lock);
atomic_t unix_tot_inflight = ATOMIC_INIT(0);
@@ -122,8 +125,16 @@ void unix_inflight(struct file *fp)
{
struct sock *s = unix_get_socket(fp);
if(s) {
- atomic_inc(&unix_sk(s)->inflight);
+ struct unix_sock *u = unix_sk(s);
+ spin_lock(&unix_gc_lock);
+ if (atomic_inc_return(&u->inflight) == 1) {
+ BUG_ON(!list_empty(&u->link));
+ list_add_tail(&u->link, &gc_inflight_list);
+ } else {
+ BUG_ON(list_empty(&u->link));
+ }
atomic_inc(&unix_tot_inflight);
+ spin_unlock(&unix_gc_lock);
}
}
@@ -131,182 +142,218 @@ void unix_notinflight(struct file *fp)
{
struct sock *s = unix_get_socket(fp);
if(s) {
- atomic_dec(&unix_sk(s)->inflight);
+ struct unix_sock *u = unix_sk(s);
+ spin_lock(&unix_gc_lock);
+ BUG_ON(list_empty(&u->link));
+ if (atomic_dec_and_test(&u->inflight))
+ list_del_init(&u->link);
atomic_dec(&unix_tot_inflight);
+ spin_unlock(&unix_gc_lock);
}
}
+static inline struct sk_buff *sock_queue_head(struct sock *sk)
+{
+ return (struct sk_buff *) &sk->sk_receive_queue;
+}
-/*
- * Garbage Collector Support Functions
- */
+#define receive_queue_for_each_skb(sk, next, skb) \
+ for (skb = sock_queue_head(sk)->next, next = skb->next; \
+ skb != sock_queue_head(sk); skb = next, next = skb->next)
-static inline struct sock *pop_stack(void)
+static void scan_inflight(struct sock *x, void (*func)(struct sock *),
+ struct sk_buff_head *hitlist)
{
- struct sock *p = gc_current;
- gc_current = unix_sk(p)->gc_tree;
- return p;
+ struct sk_buff *skb;
+ struct sk_buff *next;
+
+ spin_lock(&x->sk_receive_queue.lock);
+ receive_queue_for_each_skb(x, next, skb) {
+ /*
+ * Do we have file descriptors ?
+ */
+ if (UNIXCB(skb).fp) {
+ bool hit = false;
+ /*
+ * Process the descriptors of this socket
+ */
+ int nfd = UNIXCB(skb).fp->count;
+ struct file **fp = UNIXCB(skb).fp->fp;
+ while (nfd--) {
+ /*
+ * Get the socket the fd matches
+ * if it indeed does so
+ */
+ struct sock *sk = unix_get_socket(*fp++);
+ if(sk) {
+ hit = true;
+ func(sk);
+ }
+ }
+ if (hit && hitlist != NULL) {
+ __skb_unlink(skb, &x->sk_receive_queue);
+ __skb_queue_tail(hitlist, skb);
+ }
+ }
+ }
+ spin_unlock(&x->sk_receive_queue.lock);
}
-static inline int empty_stack(void)
+static void scan_children(struct sock *x, void (*func)(struct sock *),
+ struct sk_buff_head *hitlist)
{
- return gc_current == GC_HEAD;
+ if (x->sk_state != TCP_LISTEN)
+ scan_inflight(x, func, hitlist);
+ else {
+ struct sk_buff *skb;
+ struct sk_buff *next;
+ struct unix_sock *u;
+ LIST_HEAD(embryos);
+
+ /*
+ * For a listening socket collect the queued embryos
+ * and perform a scan on them as well.
+ */
+ spin_lock(&x->sk_receive_queue.lock);
+ receive_queue_for_each_skb(x, next, skb) {
+ u = unix_sk(skb->sk);
+
+ /*
+ * An embryo cannot be in-flight, so it's safe
+ * to use the list link.
+ */
+ BUG_ON(!list_empty(&u->link));
+ list_add_tail(&u->link, &embryos);
+ }
+ spin_unlock(&x->sk_receive_queue.lock);
+
+ while (!list_empty(&embryos)) {
+ u = list_entry(embryos.next, struct unix_sock, link);
+ scan_inflight(&u->sk, func, hitlist);
+ list_del_init(&u->link);
+ }
+ }
}
-static void maybe_unmark_and_push(struct sock *x)
+static void dec_inflight(struct sock *sk)
{
- struct unix_sock *u = unix_sk(x);
+ atomic_dec(&unix_sk(sk)->inflight);
+}
- if (u->gc_tree != GC_ORPHAN)
- return;
- sock_hold(x);
- u->gc_tree = gc_current;
- gc_current = x;
+static void inc_inflight(struct sock *sk)
+{
+ atomic_inc(&unix_sk(sk)->inflight);
}
+static void inc_inflight_move_tail(struct sock *sk)
+{
+ struct unix_sock *u = unix_sk(sk);
+
+ atomic_inc(&u->inflight);
+ /*
+ * If this is still a candidate, move it to the end of the
+ * list, so that it's checked even if it was already passed
+ * over
+ */
+ if (u->gc_candidate)
+ list_move_tail(&u->link, &gc_candidates);
+}
/* The external entry point: unix_gc() */
void unix_gc(void)
{
- static DEFINE_MUTEX(unix_gc_sem);
- int i;
- struct sock *s;
- struct sk_buff_head hitlist;
- struct sk_buff *skb;
+ static bool gc_in_progress = false;
- /*
- * Avoid a recursive GC.
- */
+ struct unix_sock *u;
+ struct unix_sock *next;
+ struct sk_buff_head hitlist;
+ struct list_head cursor;
- if (!mutex_trylock(&unix_gc_sem))
- return;
+ spin_lock(&unix_gc_lock);
- spin_lock(&unix_table_lock);
+ /* Avoid a recursive GC. */
+ if (gc_in_progress)
+ goto out;
- forall_unix_sockets(i, s)
- {
- unix_sk(s)->gc_tree = GC_ORPHAN;
- }
+ gc_in_progress = true;
/*
- * Everything is now marked
- */
-
- /* Invariant to be maintained:
- - everything unmarked is either:
- -- (a) on the stack, or
- -- (b) has all of its children unmarked
- - everything on the stack is always unmarked
- - nothing is ever pushed onto the stack twice, because:
- -- nothing previously unmarked is ever pushed on the stack
+ * First, select candidates for garbage collection. Only
+ * in-flight sockets are considered, and from those only ones
+ * which don't have any external reference.
+ *
+ * Holding unix_gc_lock will protect these candidates from
+ * being detached, and hence from gaining an external
+ * reference. This also means, that since there are no
+ * possible receivers, the receive queues of these sockets are
+ * static during the GC, even though the dequeue is done
+ * before the detach without atomicity guarantees.
*/
+ list_for_each_entry_safe(u, next, &gc_inflight_list, link) {
+ int total_refs;
+ int inflight_refs;
+
+ total_refs = file_count(u->sk.sk_socket->file);
+ inflight_refs = atomic_read(&u->inflight);
+
+ BUG_ON(inflight_refs < 1);
+ BUG_ON(total_refs < inflight_refs);
+ if (total_refs == inflight_refs) {
+ list_move_tail(&u->link, &gc_candidates);
+ u->gc_candidate = 1;
+ }
+ }
/*
- * Push root set
+ * Now remove all internal in-flight reference to children of
+ * the candidates.
*/
-
- forall_unix_sockets(i, s)
- {
- int open_count = 0;
-
- /*
- * If all instances of the descriptor are not
- * in flight we are in use.
- *
- * Special case: when socket s is embrion, it may be
- * hashed but still not in queue of listening socket.
- * In this case (see unix_create1()) we set artificial
- * negative inflight counter to close race window.
- * It is trick of course and dirty one.
- */
- if (s->sk_socket && s->sk_socket->file)
- open_count = file_count(s->sk_socket->file);
- if (open_count > atomic_read(&unix_sk(s)->inflight))
- maybe_unmark_and_push(s);
- }
+ list_for_each_entry(u, &gc_candidates, link)
+ scan_children(&u->sk, dec_inflight, NULL);
/*
- * Mark phase
+ * Restore the references for children of all candidates,
+ * which have remaining references. Do this recursively, so
+ * only those remain, which form cyclic references.
+ *
+ * Use a "cursor" link, to make the list traversal safe, even
+ * though elements might be moved about.
*/
+ list_add(&cursor, &gc_candidates);
+ while (cursor.next != &gc_candidates) {
+ u = list_entry(cursor.next, struct unix_sock, link);
- while (!empty_stack())
- {
- struct sock *x = pop_stack();
- struct sock *sk;
-
- spin_lock(&x->sk_receive_queue.lock);
- skb = skb_peek(&x->sk_receive_queue);
-
- /*
- * Loop through all but first born
- */
+ /* Move cursor to after the current position. */
+ list_move(&cursor, &u->link);
- while (skb && skb != (struct sk_buff *)&x->sk_receive_queue) {
- /*
- * Do we have file descriptors ?
- */
- if(UNIXCB(skb).fp)
- {
- /*
- * Process the descriptors of this socket
- */
- int nfd=UNIXCB(skb).fp->count;
- struct file **fp = UNIXCB(skb).fp->fp;
- while(nfd--)
- {
- /*
- * Get the socket the fd matches if
- * it indeed does so
- */
- if((sk=unix_get_socket(*fp++))!=NULL)
- {
- maybe_unmark_and_push(sk);
- }
- }
- }
- /* We have to scan not-yet-accepted ones too */
- if (x->sk_state == TCP_LISTEN)
- maybe_unmark_and_push(skb->sk);
- skb=skb->next;
+ if (atomic_read(&u->inflight) > 0) {
+ list_move_tail(&u->link, &gc_inflight_list);
+ u->gc_candidate = 0;
+ scan_children(&u->sk, inc_inflight_move_tail, NULL);
}
- spin_unlock(&x->sk_receive_queue.lock);
- sock_put(x);
}
+ list_del(&cursor);
+ /*
+ * Now gc_candidates contains only garbage. Restore original
+ * inflight counters for these as well, and remove the skbuffs
+ * which are creating the cycle(s).
+ */
skb_queue_head_init(&hitlist);
+ list_for_each_entry(u, &gc_candidates, link)
+ scan_children(&u->sk, inc_inflight, &hitlist);
- forall_unix_sockets(i, s)
- {
- struct unix_sock *u = unix_sk(s);
+ spin_unlock(&unix_gc_lock);
- if (u->gc_tree == GC_ORPHAN) {
- struct sk_buff *nextsk;
+ /* Here we are. Hitlist is filled. Die. */
+ __skb_queue_purge(&hitlist);
- spin_lock(&s->sk_receive_queue.lock);
- skb = skb_peek(&s->sk_receive_queue);
- while (skb &&
- skb != (struct sk_buff *)&s->sk_receive_queue) {
- nextsk = skb->next;
- /*
- * Do we have file descriptors ?
- */
- if (UNIXCB(skb).fp) {
- __skb_unlink(skb,
- &s->sk_receive_queue);
- __skb_queue_tail(&hitlist, skb);
- }
- skb = nextsk;
- }
- spin_unlock(&s->sk_receive_queue.lock);
- }
- u->gc_tree = GC_ORPHAN;
- }
- spin_unlock(&unix_table_lock);
+ spin_lock(&unix_gc_lock);
- /*
- * Here we are. Hitlist is filled. Die.
- */
+ /* All candidates should have been detached by now. */
+ BUG_ON(!list_empty(&gc_candidates));
+ gc_in_progress = false;
- __skb_queue_purge(&hitlist);
- mutex_unlock(&unix_gc_sem);
+ out:
+ spin_unlock(&unix_gc_lock);
}
Powered by cgit, Gruvboxed by Blaster4385.