diff mbox

[3.8.y.z,extended,stable] Patch "staging: comedi: fix a race between do_cmd_ioctl() and read/write" has been added to staging queue

Message ID 1376528714-32503-1-git-send-email-kamal@canonical.com
State New
Headers show

Commit Message

Kamal Mostafa Aug. 15, 2013, 1:05 a.m. UTC
This is a note to let you know that I have just added a patch titled

    staging: comedi: fix a race between do_cmd_ioctl() and read/write

to the linux-3.8.y-queue branch of the 3.8.y.z extended stable tree 
which can be found at:

 http://kernel.ubuntu.com/git?p=ubuntu/linux.git;a=shortlog;h=refs/heads/linux-3.8.y-queue

This patch is scheduled to be released in version 3.8.13.7.

If you, or anyone else, feels it should not be added to this tree, please 
reply to this email.

For more information about the 3.8.y.z tree, see
https://wiki.ubuntu.com/Kernel/Dev/ExtendedStable

Thanks.
-Kamal

------

From 35fc0f0718524f53196f91e25de2495942d11bb3 Mon Sep 17 00:00:00 2001
From: Ian Abbott <abbotti@mev.co.uk>
Date: Fri, 5 Jul 2013 16:49:34 +0100
Subject: staging: comedi: fix a race between do_cmd_ioctl() and read/write

commit 4b18f08be01a7b3c7b6df497137b6e3cb28adaa3 upstream.

`do_cmd_ioctl()` is called with the comedi device's mutex locked to
process the `COMEDI_CMD` ioctl to set up comedi's asynchronous command
handling on a comedi subdevice.  `comedi_read()` and `comedi_write()`
are the `read` and `write` handlers for the comedi device, but do not
lock the mutex (for performance reasons, as some things can hold the
mutex for quite a long time).

There is a race condition if `comedi_read()` or `comedi_write()` is
running at the same time and for the same file object and comedi
subdevice as `do_cmd_ioctl()`.  `do_cmd_ioctl()` sets the subdevice's
`busy` pointer to the file object way before it sets the `SRF_RUNNING` flag
in the subdevice's `runflags` member.  `comedi_read() and
`comedi_write()` check the subdevice's `busy` pointer is pointing to the
current file object, then if the `SRF_RUNNING` flag is not set, will call
`do_become_nonbusy()` to shut down the asyncronous command.  Bad things
can happen if the asynchronous command is being shutdown and set up at
the same time.

To prevent the race, don't set the `busy` pointer until
after the `SRF_RUNNING` flag has been set.  Also, make sure the mutex is
held in `comedi_read()` and `comedi_write()` while calling
`do_become_nonbusy()` in order to avoid moving the race condition to a
point within that function.

Change some error handling `goto cleanup` statements in `do_cmd_ioctl()`
to simple `return -ERRFOO` statements as a result of changing when the
`busy` pointer is set.

Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ kamal: backport to 3.8 (context) ]
Signed-off-by: Kamal Mostafa <kamal@canonical.com>
---
 drivers/staging/comedi/comedi_fops.c | 25 +++++++++++++++----------
 1 file changed, 15 insertions(+), 10 deletions(-)

--
1.8.1.2
diff mbox

Patch

diff --git a/drivers/staging/comedi/comedi_fops.c b/drivers/staging/comedi/comedi_fops.c
index 929452e..fbd9231 100644
--- a/drivers/staging/comedi/comedi_fops.c
+++ b/drivers/staging/comedi/comedi_fops.c
@@ -1180,22 +1180,19 @@  static int do_cmd_ioctl(struct comedi_device *dev,
 		DPRINTK("subdevice busy\n");
 		return -EBUSY;
 	}
-	s->busy = file;

 	/* make sure channel/gain list isn't too long */
 	if (cmd.chanlist_len > s->len_chanlist) {
 		DPRINTK("channel/gain list too long %u > %d\n",
 			cmd.chanlist_len, s->len_chanlist);
-		ret = -EINVAL;
-		goto cleanup;
+		return -EINVAL;
 	}

 	/* make sure channel/gain list isn't too short */
 	if (cmd.chanlist_len < 1) {
 		DPRINTK("channel/gain list too short %u < 1\n",
 			cmd.chanlist_len);
-		ret = -EINVAL;
-		goto cleanup;
+		return -EINVAL;
 	}

 	async->cmd = cmd;
@@ -1205,8 +1202,7 @@  static int do_cmd_ioctl(struct comedi_device *dev,
 	    kmalloc(async->cmd.chanlist_len * sizeof(int), GFP_KERNEL);
 	if (!async->cmd.chanlist) {
 		DPRINTK("allocation failed\n");
-		ret = -ENOMEM;
-		goto cleanup;
+		return -ENOMEM;
 	}

 	if (copy_from_user(async->cmd.chanlist, user_chanlist,
@@ -1258,6 +1254,9 @@  static int do_cmd_ioctl(struct comedi_device *dev,

 	comedi_set_subdevice_runflags(s, ~0, SRF_USER | SRF_RUNNING);

+	/* set s->busy _after_ setting SRF_RUNNING flag to avoid race with
+	 * comedi_read() or comedi_write() */
+	s->busy = file;
 	ret = s->do_cmd(dev, s);
 	if (ret == 0)
 		return 0;
@@ -1860,6 +1859,7 @@  static ssize_t comedi_write(struct file *file, const char __user *buf,

 		if (!(comedi_get_subdevice_runflags(s) & SRF_RUNNING)) {
 			if (count == 0) {
+				mutex_lock(&dev->mutex);
 				if (comedi_get_subdevice_runflags(s) &
 					SRF_ERROR) {
 					retval = -EPIPE;
@@ -1867,6 +1867,7 @@  static ssize_t comedi_write(struct file *file, const char __user *buf,
 					retval = 0;
 				}
 				do_become_nonbusy(dev, s);
+				mutex_unlock(&dev->mutex);
 			}
 			break;
 		}
@@ -1981,6 +1982,7 @@  static ssize_t comedi_read(struct file *file, char __user *buf, size_t nbytes,

 		if (n == 0) {
 			if (!(comedi_get_subdevice_runflags(s) & SRF_RUNNING)) {
+				mutex_lock(&dev->mutex);
 				do_become_nonbusy(dev, s);
 				if (comedi_get_subdevice_runflags(s) &
 				    SRF_ERROR) {
@@ -1988,6 +1990,7 @@  static ssize_t comedi_read(struct file *file, char __user *buf, size_t nbytes,
 				} else {
 					retval = 0;
 				}
+				mutex_unlock(&dev->mutex);
 				break;
 			}
 			if (file->f_flags & O_NONBLOCK) {
@@ -2025,9 +2028,11 @@  static ssize_t comedi_read(struct file *file, char __user *buf, size_t nbytes,
 		buf += n;
 		break;		/* makes device work like a pipe */
 	}
-	if (!(comedi_get_subdevice_runflags(s) & (SRF_ERROR | SRF_RUNNING)) &&
-	    async->buf_read_count - async->buf_write_count == 0) {
-		do_become_nonbusy(dev, s);
+	if (!(comedi_get_subdevice_runflags(s) & (SRF_ERROR | SRF_RUNNING))) {
+		mutex_lock(&dev->mutex);
+		if (async->buf_read_count - async->buf_write_count == 0)
+			do_become_nonbusy(dev, s);
+		mutex_unlock(&dev->mutex);
 	}
 	set_current_state(TASK_RUNNING);
 	remove_wait_queue(&async->wait_head, &wait);