V4L/DVB (8257): uvcvideo: Fix possible AB-BA deadlock with videodev_lock and open_mutex
The uvcvideo driver's uvc_v4l2_open() method is called from videodev's
video_open() function, which means it is called with the videodev_lock
mutex held. uvc_v4l2_open() then takes uvc_driver.open_mutex to check
dev->state and avoid racing against a device disconnect, which means
that open_mutex must nest inside videodev_lock.
However uvc_disconnect() takes the open_mutex around setting
dev->state and also around putting its device reference. However, if
uvc_disconnect() ends up dropping the last reference, it will call
uvc_delete(), which calls into the videodev code to unregister its
device, and this will end up taking videodev_lock. This opens a
(unlikely in practice) window for an AB-BA deadlock and also causes a
lockdep warning because of the lock misordering.
Fortunately there is no apparent reason to hold open_mutex when doing
kref_put() in uvc_disconnect(): if uvc_v4l2_open() runs before the
state is set to UVC_DEV_DISCONNECTED, then it will take another
reference to the device and kref_put() won't call uvc_delete; if
uvc_v4l2_open() runs after the state is set, it will run before
uvc_delete(), see the state, and return immediately -- uvc_delete()
does uvc_unregister_video() (and hence video_unregister_device(),
which is synchronized with videodev_lock) as its first thing, so there
is no risk of use-after-free in uvc_v4l2_open().
Bug diagnosed based on a lockdep warning reported by Romano Giannetti
<romano@dea.icai.upcomillas.es>.
Andrew Morton [Fri, 4 Jul 2008 09:33:23 +0000 (06:33 -0300)]
V4L/DVB (8256): uvc/uvc_v4l2.c: suppress uninitialized var warning
drivers/media/video/uvc/uvc_v4l2.c: In function `uvc_v4l2_mmap':
drivers/media/video/uvc/uvc_v4l2.c:1035: warning: 'buffer' might be used uninitialized in this function
/home/v4l/master/v4l/gspca.c: In function 'gspca_frame_add':
/home/v4l/master/v4l/gspca.c:222: warning: format '%d' expects type 'int', but argument 2 has type 'long int'
/home/v4l/master/v4l/gspca.c: In function 'dev_read':
/home/v4l/master/v4l/gspca.c:1568: warning: format '%d' expects type 'int', but argument 2 has type 'size_t'
/home/v4l/master/v4l/gspca.c:1618: warning: format '%d' expects type 'int', but argument 3 has type 'size_t'
Jean Delvare [Thu, 10 Jul 2008 14:23:37 +0000 (11:23 -0300)]
V4L/DVB (8246): tvaudio: Stop I2C driver ID abuse
The tvaudio driver is using "official" I2C device IDs for internal
purpose. There must be some historical reason behind this but anyway,
it shouldn't do that. As the stored values are never used, the easiest
way to fix the problem is simply to remove them altogether.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
Jean Delvare [Thu, 10 Jul 2008 14:04:10 +0000 (11:04 -0300)]
V4L/DVB (8245): ovcamchip: Delete stray I2C bus ID
I2C_HW_SMBUS_OVFX2 is referenced in ovcamchip_core.c, but no bus uses
this driver ID, so we can remove the reference. As far as I can see,
the Cypress FX2 webcam is handled by a different driver (dvb-usb).
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
V4L/DVB (8234): uvcvideo: Make input device support optional
UVC devices can report button events. The uvcvideo driver depends on
CONFIG_INPUT to report events to the input layer. This patch removes the hard
dependency by introducing a new CONFIG_USB_VIDEO_CLASS_INPUT_EVDEV option.
V4L/DVB (8231): gspca: Do not declare the webcams declared by other drivers.
etoms: Do not declare the webcams declared by the driver et61x251.
sonixb, sonixj:
Do not declare the webcams declared by the driver sn9c102.
zc3xx: Do not declare the webcams declared by the driver zc0301.
V4L/DVB (8209): uvcvideo: Don't free URB buffers on suspend.
All submitted URBs must be killed at suspend time, but URB buffers don't have
to be freed. Avoiding a free on suspend/reallocate on resume lowers the presure
on system memory.
V4L/DVB (8208): uvcvideo: Use GFP_NOIO when allocating memory during resume
The swap device might still be asleep, so memory allocated in the resume
handler must use GFP_NOIO. Thanks to Oliver Neukum for catching and reporting
this bug.
Hans de Goede [Thu, 3 Jul 2008 14:20:58 +0000 (11:20 -0300)]
V4L/DVB (8194): gspca: Fix the format of the low resolution mode of spca561.
The low (half) res modes of the spca561 are not spca561 compressed, but are
raw bayer, this patches fixes this and adds a PIX_FMT define for the GBRG
bayer format used by the spca561 in low res mode.
Signed-off-by: Hans de Goede <j.w.r.degoede@hhs.nl> Signed-off-by: Jean-Francois Moine <moinejf@free.fr> Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
Andy Walls [Mon, 7 Jul 2008 13:36:16 +0000 (10:36 -0300)]
V4L/DVB (8189): cx18: Use correct GPIO pin for resetting Xceive 3028 tuner on Yuan MPC718
Change the Yuan MPC718 cards entry to use the correct GPIO pin for resetting
the Xceive 3028 tuner. Thanks to Brian Hope <brian@hopefamily.info> for
taking the time and figuring out which pin to use.
Signed-off-by: Andy Walls <awalls@radix.net> Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
V4L/DVB (8181): gspca: read() did not work (loop in kernel, timeout...)
main: Remove some vidioc_xx traces.
main: read() did not work (user irq instead of mmap irq).
main: Lack of v4l1 compat.
main: Process loop inside kernel when no frame arriving.
main: Double qbuf in read() when too many buffered frames.
Hans Verkuil [Fri, 27 Jun 2008 18:33:02 +0000 (15:33 -0300)]
V4L/DVB (8162): cx18: fix PAL/SECAM support
Reverted the 'Fix unintended auto configurations in cx18-av-core' patch,
instead disable the auto config completely.
Fix a bug in cx18_av_vbi_setup() where the standard tests were done
in the wrong order.
Tested with NTSC-M, PAL-BG, PAL-I, PAL-DK, PAL-M, PAL-Nc, SECAM-DK,
SECAM-L and SECAM-BG. The last one does not work at the moment due to
a tda9887.c bug.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
main: V4L2_PIX_FMT_SPCA501 is compressed (thanks to Hans de Goede)
main: return 0 when no change on vidioc_s_fmt_cap (thanks to Hans de Goede)
pac207: cleanup
V4L/DVB (8154): Fix protection problems in the main driver.
- Protect format change when streaming active.
- Protect USB exchanges on close.
- Set a timeout in frame wait.
- Have only one capture file and free the resources when closing this file.
- Simplify the URB buffer.
- Don't reset the control values at open time in pac207.
- Fix compilation warnings of stk014.
Al Viro [Sun, 22 Jun 2008 17:20:49 +0000 (14:20 -0300)]
V4L/DVB (8126): net endianness fix
According to RFC 4326 (4.1) D-bit is MSB in net-endian 16bit.
dvb_net.c did
/* Set D-bit for CRC32 verification,
* if it was set originally. */
ulen |= 0x0080;
which works of little-endian (htons(1<<15) is 0x0080 there), but breaks
on big-endian.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
V4L/DVB (8123): Add support for em2860 based PointNix Intra-Oral Camera
em28xx-cards.c
em28xx-input.c
em28xx-video.c
em28xx.h
- Add support for the PointNix Intra-Oral Camera, which required addition of
a construct for reading the "snapshot" button (provided on the em2860 and
em2880 chips, but this is the first case where I have seen it actually used
in a product). The button is wired to pin 56 on the em2880.
http://www.pointnix.com/ENG/dental/product_02.asp
Thanks to Roberto Mantovani <rmantovani@libero.it> for testing the changes
Hans Verkuil [Wed, 25 Jun 2008 09:54:05 +0000 (06:54 -0300)]
V4L/DVB (8116): videodev: allow PRIVATE_BASE controls when called through VIDIOC_G/S_CTRL.
V4L2_CID_PRIVATE_BASE controls are not allowed when called from
VIDIOC_S/G_EXT_CTRL as extended controls use a better mechanism
for private controls. But still allow it when called from the
VIDIOC_G/S_CTRL to extended control conversion in video_ioctl2()
for backwards compatibility.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
cx18: Improved Raptor card audio input routing defintions, so that muxer values
matched cx18_gpio() values for tuner, line in 1, and radio and added LED
indication of selected audio input. Audio line in 2 doesn't work as it uses
the not yet supported 2nd I2S port. Tuner/FM Radio AF is mono until SIF
support is fixed.
Signed-off-by: Andy Walls <awalls@radix.net> Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
Hans Verkuil [Wed, 25 Jun 2008 09:29:01 +0000 (06:29 -0300)]
V4L/DVB (8112): videodev: improve extended control support in video_ioctl2()
- add sanity checks for the extended controls argument.
- if the driver only supports extended controls, then convert
old-style controls to an extended control callback.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
- use %x rather than %d for control IDs and bitmask fields
- make two arrays const
- show the whole control array for the ext_ctrl ioctls
- print pix_fmt for V4L2_BUF_TYPE_VIDEO_OUTPUT
- show full type name rather than an integer
- fix CROPCAP debugging
- fix G/S_TUNER debugging
- show error code in case of an error
- other small cleanups
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
Michael Krufky [Sun, 22 Jun 2008 01:13:42 +0000 (22:13 -0300)]
V4L/DVB (8098): xc5000: add module option to load firmware during driver attach
The xc5000 firmware download process can be slow on certain devices.
Add a module option, "init_fw" to perform the firmware download during
xc5000_attach(), which would get the firmware download over with well
before any tune request is made.
When this option is enabled, it will reduce the time it takes to get a
signal lock during the first tune after driver load, but will add a few
seconds to the driver initialization.
Signed-off-by: Michael Krufky <mkrufky@linuxtv.org> Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
Ian Armstrong [Sat, 21 Jun 2008 14:09:46 +0000 (11:09 -0300)]
V4L/DVB (8090): ivtv: yuv decoder lock fix (2)
The IVTV_IOC_DMA_FRAME ioctl fails to correctly 'start' the decoder. Although
yuv output will be correct, some functions which should be disabled while the
decoder is in use remain enabled. This can result in hardware registers being
corrupted, causing problems with the mpeg decoder. This patch ensures the
decoder 'start' sequence is called, disabling these functions until the
device is closed.
Signed-off-by: Ian Armstrong <ian@iarmst.demon.co.uk> Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
Ian Armstrong [Sat, 21 Jun 2008 12:25:23 +0000 (09:25 -0300)]
V4L/DVB (8088): ivtv: yuv decoder lock fix
write() operations to the yuv device may not always 'start' the decoder.
Although yuv output will be correct, some functions which should be disabled
while the decoder is in use remain enabled. This can result in hardware
registers being corrupted, causing problems with the mpeg decoder. This patch
simply moves the existing decoder 'start' sequence to earlier in the write()
handler.
Signed-off-by: Ian Armstrong <ian@iarmst.demon.co.uk> Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
projects / linux-2.6 / log