[dahdi-commits] sruffell: linux/trunk r9353 - /linux/trunk/drivers/dahdi/dahdi-base.c
SVN commits to the DAHDI project
dahdi-commits at lists.digium.com
Mon Sep 20 15:32:34 CDT 2010
Author: sruffell
Date: Mon Sep 20 15:32:29 2010
New Revision: 9353
URL: http://svnview.digium.com/svn/dahdi?view=rev&rev=9353
Log:
dahdi: Be more tolerant of surprise removal of channels.
Enable DAHDI to detect if an operation on a file handle refers to a
channel that may have been unregistered. This can occur, for example,
when a board driver is hot-swapped out in a live system.
This patch ensures that file->private_data is always properly set for
any open channel, and it's set back to NULL when a channel is
unregistered. This way file->private_data can be used to check whether
it's valid to perform an operation on the channel. (NOTE: There is
still a race condition here if the driver was unbound on one processor
during the window of time between when file->private_data was checked
and the system call finishes).
Also, since DAHDI should only return -ENODEV on read or write when there
was a surprise device removal on a running system this sleep can prevent
the system from becoming unresponsive if the userspace application does
not check for the -ENODEV error and constantly tries to call read with
elevated privileges.
(issue #17669)
Reported by: tzafrir
Tested by: sruffell
Review: https://reviewboard.asterisk.org/r/905/
Signed-off-by: Shaun Ruffell <sruffell at digium.com>
Modified:
linux/trunk/drivers/dahdi/dahdi-base.c
Modified: linux/trunk/drivers/dahdi/dahdi-base.c
URL: http://svnview.digium.com/svn/dahdi/linux/trunk/drivers/dahdi/dahdi-base.c?view=diff&rev=9353&r1=9352&r2=9353
==============================================================================
--- linux/trunk/drivers/dahdi/dahdi-base.c (original)
+++ linux/trunk/drivers/dahdi/dahdi-base.c Mon Sep 20 15:32:29 2010
@@ -47,6 +47,7 @@
#include <linux/moduleparam.h>
#include <linux/sched.h>
#include <linux/list.h>
+#include <linux/delay.h>
#ifdef HAVE_UNLOCKED_IOCTL
#include <linux/smp_lock.h>
@@ -1971,6 +1972,15 @@
might_sleep();
+ /* In the case of surprise removal of hardware, make sure any open
+ * file handles to this channel are disassociated with the actual
+ * dahdi_chan. */
+ if (chan->file) {
+ chan->file->private_data = NULL;
+ if (chan->span)
+ module_put(chan->span->ops->owner);
+ }
+
release_echocan(chan->ec_factory);
#ifdef CONFIG_DAHDI_NET
@@ -2024,9 +2034,10 @@
write_unlock_irqrestore(&chan_lock, flags);
}
-static ssize_t dahdi_chan_read(struct file *file, char __user *usrbuf, size_t count, int unit)
-{
- struct dahdi_chan *chan = chans[unit];
+static ssize_t dahdi_chan_read(struct file *file, char __user *usrbuf,
+ size_t count)
+{
+ struct dahdi_chan *chan = file->private_data;
int amnt;
int res, rv;
int oldbuf,x;
@@ -2035,10 +2046,19 @@
/* Make sure count never exceeds 65k, and make sure it's unsigned */
count &= 0xffff;
- if (!chan)
- return -EINVAL;
-
- if (count < 1)
+ if (unlikely(!chan)) {
+ /* We would typically be here because of surprise hardware
+ * removal or driver unbinding while a user space application
+ * has a channel open. Most telephony applications are run at
+ * elevated priorities so this sleep can prevent the high
+ * priority threads from consuming the CPU if they're not
+ * expecting surprise device removal.
+ */
+ msleep(5);
+ return -ENODEV;
+ }
+
+ if (unlikely(count < 1))
return -EINVAL;
for (;;) {
@@ -2150,21 +2170,30 @@
return range1 + range2;
}
-static ssize_t dahdi_chan_write(struct file *file, const char __user *usrbuf, size_t count, int unit)
+static ssize_t dahdi_chan_write(struct file *file, const char __user *usrbuf,
+ size_t count)
{
unsigned long flags;
- struct dahdi_chan *chan = chans[unit];
+ struct dahdi_chan *chan = file->private_data;
int res, amnt, oldbuf, rv, x;
/* Make sure count never exceeds 65k, and make sure it's unsigned */
count &= 0xffff;
- if (!chan)
+ if (unlikely(!chan)) {
+ /* We would typically be here because of surprise hardware
+ * removal or driver unbinding while a user space application
+ * has a channel open. Most telephony applications are run at
+ * elevated priorities so this sleep can prevent the high
+ * priority threads from consuming the CPU if they're not
+ * expecting surprise device removal.
+ */
+ msleep(5);
+ return -ENODEV;
+ }
+
+ if (unlikely(count < 1))
return -EINVAL;
-
- if (count < 1) {
- return -EINVAL;
- }
for (;;) {
spin_lock_irqsave(&chan->lock, flags);
@@ -2748,6 +2777,7 @@
}
if (!res) {
chan->file = file;
+ file->private_data = chan;
spin_unlock_irqrestore(&chan->lock, flags);
} else {
spin_unlock_irqrestore(&chan->lock, flags);
@@ -2912,7 +2942,7 @@
chan = file->private_data;
if (!chan)
return -EINVAL;
- return dahdi_chan_read(file, usrbuf, count, chan->channo);
+ return dahdi_chan_read(file, usrbuf, count);
}
if (unit == 255) {
@@ -2921,12 +2951,12 @@
module_printk(KERN_NOTICE, "No pseudo channel structure to read?\n");
return -EINVAL;
}
- return dahdi_chan_read(file, usrbuf, count, chan->channo);
+ return dahdi_chan_read(file, usrbuf, count);
}
if (count < 0)
return -EINVAL;
- return dahdi_chan_read(file, usrbuf, count, unit);
+ return dahdi_chan_read(file, usrbuf, count);
}
static ssize_t dahdi_write(struct file *file, const char __user *usrbuf, size_t count, loff_t *ppos)
@@ -2944,7 +2974,7 @@
chan = file->private_data;
if (!chan)
return -EINVAL;
- return dahdi_chan_write(file, usrbuf, count, chan->channo);
+ return dahdi_chan_write(file, usrbuf, count);
}
if (unit == 255) {
chan = file->private_data;
@@ -2952,9 +2982,9 @@
module_printk(KERN_NOTICE, "No pseudo channel structure to read?\n");
return -EINVAL;
}
- return dahdi_chan_write(file, usrbuf, count, chan->channo);
- }
- return dahdi_chan_write(file, usrbuf, count, unit);
+ return dahdi_chan_write(file, usrbuf, count);
+ }
+ return dahdi_chan_write(file, usrbuf, count);
}
@@ -5891,6 +5921,12 @@
ret = dahdi_chanandpseudo_ioctl(file, cmd, data, chan->channo);
goto unlock_exit;
}
+
+ if (!file->private_data) {
+ ret = -ENXIO;
+ goto unlock_exit;
+ }
+
ret = dahdi_chan_ioctl(file, cmd, data, unit);
unlock_exit:
More information about the dahdi-commits
mailing list