提交 5342b77c 编写于 作者: A Alan Cox 提交者: Live-CD User

slip: Clean up create and destroy

The network layer now has a destructor we can hook to clean up the slip
devices array. That needs us to initiate unregister events in the right
places which with the current tty layer we can do, and with network
refcounting is safe to do.
Signed-off-by: NAlan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>
上级 46fb7825
......@@ -616,6 +616,14 @@ static void sl_uninit(struct net_device *dev)
sl_free_bufs(sl);
}
/* Hook the destructor so we can free slip devices at the right point in time */
static void sl_free_netdev(struct net_device *dev)
{
int i = dev->base_addr;
free_netdev(dev);
slip_devs[i] = NULL;
}
static const struct net_device_ops sl_netdev_ops = {
.ndo_init = sl_init,
.ndo_uninit = sl_uninit,
......@@ -634,7 +642,7 @@ static const struct net_device_ops sl_netdev_ops = {
static void sl_setup(struct net_device *dev)
{
dev->netdev_ops = &sl_netdev_ops;
dev->destructor = free_netdev;
dev->destructor = sl_free_netdev;
dev->hard_header_len = 0;
dev->addr_len = 0;
......@@ -712,8 +720,6 @@ static void sl_sync(void)
static struct slip *sl_alloc(dev_t line)
{
int i;
int sel = -1;
int score = -1;
struct net_device *dev = NULL;
struct slip *sl;
......@@ -724,55 +730,7 @@ static struct slip *sl_alloc(dev_t line)
dev = slip_devs[i];
if (dev == NULL)
break;
sl = netdev_priv(dev);
if (sl->leased) {
if (sl->line != line)
continue;
if (sl->tty)
return NULL;
/* Clear ESCAPE & ERROR flags */
sl->flags &= (1 << SLF_INUSE);
return sl;
}
if (sl->tty)
continue;
if (current->pid == sl->pid) {
if (sl->line == line && score < 3) {
sel = i;
score = 3;
continue;
}
if (score < 2) {
sel = i;
score = 2;
}
continue;
}
if (sl->line == line && score < 1) {
sel = i;
score = 1;
continue;
}
if (score < 0) {
sel = i;
score = 0;
}
}
if (sel >= 0) {
i = sel;
dev = slip_devs[i];
if (score > 1) {
sl = netdev_priv(dev);
sl->flags &= (1 << SLF_INUSE);
return sl;
}
}
/* Sorry, too many, all slots in use */
if (i >= slip_maxdev)
return NULL;
......@@ -908,31 +866,14 @@ static int slip_open(struct tty_struct *tty)
return err;
}
/*
FIXME: 1,2 are fixed 3 was never true anyway.
Let me to blame a bit.
1. TTY module calls this funstion on soft interrupt.
2. TTY module calls this function WITH MASKED INTERRUPTS!
3. TTY module does not notify us about line discipline
shutdown,
Seems, now it is clean. The solution is to consider netdevice and
line discipline sides as two independent threads.
By-product (not desired): sl? does not feel hangups and remains open.
It is supposed, that user level program (dip, diald, slattach...)
will catch SIGHUP and make the rest of work.
I see no way to make more with current tty code. --ANK
*/
/*
* Close down a SLIP channel.
* This means flushing out any pending queues, and then returning. This
* call is serialized against other ldisc functions.
*
* We also use this method fo a hangup event
*/
static void slip_close(struct tty_struct *tty)
{
struct slip *sl = tty->disc_data;
......@@ -951,10 +892,16 @@ static void slip_close(struct tty_struct *tty)
del_timer_sync(&sl->keepalive_timer);
del_timer_sync(&sl->outfill_timer);
#endif
/* Count references from TTY module */
/* Flush network side */
unregister_netdev(sl->dev);
/* This will complete via sl_free_netdev */
}
static int slip_hangup(struct tty_struct *tty)
{
slip_close(tty);
return 0;
}
/************************************************************************
* STANDARD SLIP ENCAPSULATION *
************************************************************************/
......@@ -1311,6 +1258,7 @@ static struct tty_ldisc_ops sl_ldisc = {
.name = "slip",
.open = slip_open,
.close = slip_close,
.hangup = slip_hangup,
.ioctl = slip_ioctl,
.receive_buf = slip_receive_buf,
.write_wakeup = slip_write_wakeup,
......@@ -1384,6 +1332,8 @@ static void __exit slip_exit(void)
}
} while (busy && time_before(jiffies, timeout));
/* FIXME: hangup is async so we should wait when doing this second
phase */
for (i = 0; i < slip_maxdev; i++) {
dev = slip_devs[i];
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册