This patch afixes several driver bugs
* Make the windows disabled by default at module load and unload
* Enable the attribute window
* Change the order of the windows, now osd0 is fb0, so the video mode change
  must be done on vid0 which is fb1
* Remove more unsued code
* Make the enable_digital output to not change the window configurations
* Add a new ioctl to enable color bar test mode
* Fix some of the rules of the check_var function, now the virtual size of the
  window is handled automatic based on the 32 bytes align constraints

This adds more memory to the attribute window, to make it big enough for
a 720x480 screen, this is enough for now.

Remove the logic of auto detect on valid input signal.
Let user decide choose and decide which one is valid.
Signed-off-by: 邱宗炽 <tqiu@neuros.com.cn>

To avoid waste time on needless hwif probe, add logic for wait_hwif_ready
function: if all drives in a HW interface was marked as "noprobe", then
skip the HW interface probe.

Signed-off-by: 邱宗炽 <tqiu@neuros.com.cn>

There are two port of composite input, can select which one input.
Signed-off-by: 邱宗炽 <tqiu@neuros.com.cn>

Remove all OR'ed return values, this mades the driver to work on vlc. Now the
driver is more standard.

device driver must register class device, or the mdev(udev) can not
get enough message to create the device node

Merge branch 'neuros' of ssh://git@git.neuros.com.cn/git/git-pub/osd20/linux-davinci-2.6 into neuros

the previous way will clear all pins signal when
set any thing to the clear register

Lots of new stuff has been addded, here is a brief summary:
* New ioctl to set transparency color
* New ioctl to enable transparency on osd0 and osd1
* The memory allocation now is provided by a struct on the machine description
  Basically, it tells how big the buffers for every plane should be
* Clean up the code, remove unusued code
* Add support for mode switching on vid0 plane, SD / HD is supported, the
  latter is not tested
* Add support for mode list, you can list the modes with:
  cat /sys/class/graphics/fb0/modes
* Add support for mode set, you can se the mode with:
  echo MODE > /sys/class/grpahics/mode
  where MODE is one of the returned my the previous command
* Output switch works, by writing the enum value found on davincifb.h
  to /sys/class/video_output/venc

Merge branch 'neuros' of ssh://git@git.neuros.com.cn/git/git-pub/osd20/linux-davinci-2.6 into neuros

fix some ioctls to support multi-backend device
Signed-off-by: 邱宗炽 <tqiu@neuros.com.cn>

remove the ioctl of active device.
Auto detect the sinal input when open the device. tvp5150 work.
tvp7000 need be added after it's driver be implemented.
Signed-off-by: 邱宗炽 <tqiu@neuros.com.cn>

the bug is: read wrong register to get current video standard.
Signed-off-by: 邱宗炽 <tqiu@neuros.com.cn>

Signed-off-by: 邱宗炽 <tqiu@neuros.com.cn>

Merge branch 'neuros' of ssh://git@git.neuros.com.cn/git/git-pub/osd20/linux-davinci-2.6 into neuros

initial version, only with a few command, have not test

With this patch, the name of the kernel is always 2.6.23-davinci1
instead of changing with local changes with names such as
2.6.23-davinci1-g723849a8-dirty. At the moment we don't need this
feature and it just create confusion and leave cruft in
/lib/modules, so it's better to keep it disabled.

Kernel is hanging for a long time on probing the IDE, even
when there's no hard disk connected. This patch fix a bug,
Now we can pass parameters to kernel to skip drive probing
if none exist.
Summary of ide driver parameters for kernel command line:
"hdx=noprobe"  : drive may be present, but don't probe for it

Merge branch 'neuros' of ssh://git@git.neuros.com.cn/git/git-pub/osd20/linux-davinci-2.6 into neuros

make driver more readable
add hdcp exchange function

Merge branch 'neuros' of ssh://git@git.neuros.com.cn/git/git-pub/osd20/linux-davinci-2.6 into neuros

initialize gpio 47 as output and low,
detect the first wave logic change

New DM driver is added. This driver will replace the current davincifb soon
A hard refactor has been done on the davincifb to create this driver and
several bugs fixed. Right now is TEST ONLY.

When a kernel for a davinci board is built, increase the order of the
allocator to 13, i.e 2^12 max contiguous chunk size (16MB). Looks like it is
harmless. Also add the davincifb.h header file when the kernel headers are
installed.

This patch will be included on mainstream version soon.
alloc_pages_exact() is similar to alloc_pages(), except that it allocates
the minimum number of pages to fulfill the request.  This is useful if you
want to allocate a very large buffer that is slightly larger than an
even power-of-two number of pages.  In that case, alloc_pages() will waste
a lot of memory

adjust PWM timer GPIO configure and parameter
 now ir receive and blaster data consistent.

This is a hack to avoid waste time on IDE probe.

Signed-off-by: 邱宗炽 <tqiu@neuros.com.cn>

The hang was caused because the driver attempted to call dev_dbg with the
rsz_dev struct being NULL. This patch fixes the bug number 4690. Some dev_dbg
functions were missing the format string. Removed unused functions. Initialize
correctly the rsz_dev. The real error now is "Resizer write buffer overflow".

Merge branch 'neuros' of ssh://git@git.neuros.com.cn/git/git-pub/osd20/linux-davinci-2.6 into neuros