mmap smp problem ?

[Peter Pregler]

On 23-Nov-99 Gleicon S. Moraes wrote:
>>
>> There is some warning in the README that mmap on SMP is a killer. Normal
>> read operation, e.g. the gqcam-application, should work fine. Can you
>> check that?
> 
> is there a link pointing to a simple implementation of the read operation
> using cpia
> driver ? gqcam gave an ideia, but i want to make a layer to put in some v4l
> programs. The
> point is: i use bttv, and most of these programs use mmap to get the picture.
> Can i just set the image size and depth , and after thist, read the entire
> picture ?
> I just want to make those programs run with cpia driver.

That is exactly what the 'synchronous' mmap-implementation I was talking
about does; but at driver level and not at applocation level. It maps the
mmap-calls to the same routines used for read calls (ioctl-CMCAPTURE will
actually read the frame, ioctl-CSYNC will just mark the buffer free). So old
v4l-applications will work (e.g. xawtv, webcam, w3cam do work on my test box).

For new applications use the read-interface as implemented in the gqcam. Memory
mapping does not make that much sense (compared to video capture boards as
the bttv) for a parallel port webcam unless we get interrupt driven operation
to work.z

Just stay tuned for a few more days. There are some small problems with
exposure control and stability during crash tests (e.g. unplugging the camera
during read should not crash the box). I will get rid of those over the weekend
and release the stuff afterwards.

-Peter

-------------------------------
Even the thought that a ship was waiting to take him back to Earth
did not wipe out the sense of loss he felt at that moment.
-------------------------------
Email: Peter.Pregler@risc.uni-linz.ac.at
Peter Pregler / RISC, University of Linz, Austria
-----------------------------------------------------------------------------
To unsubscribe from this mailinglist, send the line "unsubscribe vision-webcam" in the
body of a message to "majordomo@errors.no".