To ov511 transplant V4l2 Drive, challenge.

Now the project needs, through the ov511 to do this thing real time to keep up, so to the most I do not want to go that part, originally thought to change v4l2 on it, who know unexpectedly drive not on,

Another opportunity to improve, this step away if still can't keep up, can only explain my capture method is not, I can't think of other ways to improve speed.

But say is transplant v4l2 Drive, not easy ah, I do not know under drivers/video/to modify which file, can only bite the bullet down to do.

Come on.

1. According to make Menuconfig find the appropriate drive location help look at the location of Kconfig

2. Find ov511.c ov511.h files under drivers/media/video/, replace them directly with files that provide v4l2 support, 2.6.30.4 files in the kernel

3.make

Wrong, good, the first mistake I forget, said is lack of v4l2-ioctl.h I will go to 2.6.30.4 copy over.

Then again a large series of warnings (ignore), and came to struct Video_device no parent This member, I look for

Where is this structure, tangled,

I compare Ov511.c's head file to find, still did not find, Baidu Bar, found the Include/media/v4l2-dev.h

Here Ctrl+f no parent This member, there is struct video_device this structure, simply the same, copy some come over,

4 Continue Make

Good point the end of the near misses, now the test, but the camera is not. Wait a minute.

I am now very silent on the above practice, do not know how many misguided heroic teenager, I now know.

In fact what Ah, that I modified a lot of documents are not successful, and finally I began to suspect that the driver is the way I transplanted it.

Helpless under I sleep feel to start again, see Kconfig that vivi.c in the end in which directory, Ah, is really the eyes of heaven ah.

Vivi.c was found by me, in that virtual video drivers, so it would be good to do.

Make Menuconfig

Pick up

Make

Wow, there's vivi.o haha.

Test, Oh, my Lady Gaga, succeeded.

The following question is the application problem, haha

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

V4L2 is not like the above said to blindly make files, the network said VIVI.C did not tell us how to configure V4L2, I went to kconfig under Search Vivi

Also really give me found, looked under Transite name, originally is virtual video Driver, and then see Help V4L2, continue to search to see if there is

There are other, no, OK, basically finished. Wrap Added into the kernel, the compilation was successful.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

Program written, in the PC test success, you can save the file, but to arm under the die, do not know what the reason ...

I tried a lot of things.

With struct stat the size of the test file, it is so large, but fwrite saved binary file clearly returns the value is 0?????

This question, what do you suggest?

Drive the question I don't doubt that nothing has changed anything ... (may be the driver problem)

But except to keep everything normal. The save is not driven by the driver, it is not the data from the buffer queue.

XOR or fwrite the data to be saved is too large for the transmission to fail (by the way, I am using an NFS mounted filesystem).

In doubt

##################################################################################

Problem solved, is my Linux under the disk space capacity is not enough, NFS permissions or no native operation of the large.

Windows Firewall is not off, I gave it off.

VI text in the Linux editor, and vi text on ARM edit

Linux can save files, arm can not,

CP xx pp under Linux can arm, but also with tips, write error ... I see the disk capacity of the problem (a few days ago also prompted, hehe)

##################################################################################