linux-kernel - [PATCH 03/10] staging: comedi: do extra checks for becoming non-busy for "read"

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <1444666890-14353-4-git-send-email-abbotti@mev.co.uk>
Date:	Mon, 12 Oct 2015 17:21:22 +0100
From:	Ian Abbott <abbotti@....co.uk>
To:	driverdev-devel@...uxdriverproject.org
Cc:	Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
	Ian Abbott <abbotti@....co.uk>,
	H Hartley Sweeten <hsweeten@...ionengravers.com>,
	linux-kernel@...r.kernel.org
Subject: [PATCH 03/10] staging: comedi: do extra checks for becoming non-busy for "read"

`comedi_read()` is the handler for the "read" file operation for COMEDI
devices.  It mostly runs without using the main mutex of the COMEDI
device, but uses the `attach_lock` rwsemaphore to protect against the
COMEDI device becoming "detached".  A file object can read data
resulting from a COMEDI asynchonous command if it initiated the command.
The COMEDI subdevice is marked as busy when the command is started.  At
some point, the "read" handler detects that the command has terminated
and all available data has been read and so marks the subdevice as
non-busy.

In order to mark the subdevice as non-busy, the "read" handler needs to
release the `attach_lock` rwsemaphore and `acquire the main `mutex`.
There is a vulnerable point between the two, so it checks that the
device is still attached after acquiring the mutex.  However, it does
not currently check that the conditions for becoming non-busy still
hold.  Add some more checks that the subdevice is still busy with a
command initiated by the same file object, that command is in the correct
direction (in case the subdevice supports both "read" and "write"), that
command has terminated, and has no data available to be read.

Signed-off-by: Ian Abbott <abbotti@....co.uk>
---
 drivers/staging/comedi/comedi_fops.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/drivers/staging/comedi/comedi_fops.c b/drivers/staging/comedi/comedi_fops.c
index 89e8e87..cca3fb1 100644
--- a/drivers/staging/comedi/comedi_fops.c
+++ b/drivers/staging/comedi/comedi_fops.c
@@ -2566,14 +2566,17 @@ static ssize_t comedi_read(struct file *file, char __user *buf, size_t nbytes,
 		 * sufficient (unless there have been 2**32 detaches in the
 		 * meantime!), but check the subdevice pointer as well just in
 		 * case.
+		 *
+		 * Also check the subdevice is still in a suitable state to
+		 * become non-busy in case it changed behind our back.
 		 */
 		new_s = comedi_file_read_subdevice(file);
 		if (dev->attached && old_detach_count == dev->detach_count &&
-		    s == new_s && new_s->async == async) {
-			if (become_nonbusy ||
-			    comedi_buf_read_n_available(s) == 0)
-				do_become_nonbusy(dev, s);
-		}
+		    s == new_s && new_s->async == async && s->busy == file &&
+		    !(async->cmd.flags & CMDF_WRITE) &&
+		    !comedi_is_subdevice_running(s) &&
+		    comedi_buf_read_n_available(s) == 0)
+			do_become_nonbusy(dev, s);
 		mutex_unlock(&dev->mutex);
 	}
 out:
-- 
2.6.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/