summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorHans Verkuil <hans.verkuil@cisco.com>2013-04-14 10:53:07 -0300
committerMauro Carvalho Chehab <mchehab@redhat.com>2013-04-15 08:34:37 -0300
commita877e2771a7f74b724cc09d96decd2ff3a37f2cd (patch)
tree2d3b727576dcf4cd06f3df751bedb4bad1bd190a
parenta9af5392520c9c30e355562b0bd641d49130284a (diff)
[media] cx25821: do not expose broken video output streams
The cx25821 driver has support for one audio output channel and two video output channels. This is implemented in a very ugly and very evil way through a custom ioctl that passes the filename of a file containing the video data, which is then read by the driver itself using vfs. There are a number of problems with this: 1) it's very ugly and very evil (I can't say that often enough). 2) V4L2 supports video output, so why not use that? 3) it's very buggy, closing the filehandle through which you passed the ioctl will oops the kernel. 4) it's a nasty security leak since this allows you to load any file in the system as a video or audio source, so in theory you can output /etc/passwd to audio or video out and record & decode it on another device. Because of all these issues we no longer register those output video nodes. Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
-rw-r--r--drivers/media/pci/cx25821/cx25821-video.c2
1 files changed, 1 insertions, 1 deletions
diff --git a/drivers/media/pci/cx25821/cx25821-video.c b/drivers/media/pci/cx25821/cx25821-video.c
index 1465591d000c..6b18320ebef1 100644
--- a/drivers/media/pci/cx25821/cx25821-video.c
+++ b/drivers/media/pci/cx25821/cx25821-video.c
@@ -461,7 +461,7 @@ int cx25821_video_register(struct cx25821_dev *dev)
spin_lock_init(&dev->slock);
- for (i = 0; i < MAX_VID_CHANNEL_NUM - 1; ++i) {
+ for (i = 0; i < VID_CHANNEL_NUM; ++i) {
cx25821_init_controls(dev, i);
cx25821_risc_stopper(dev->pci, &dev->channels[i].vidq.stopper,