[cpia] bug in cpia_usb_streamRead

[Jon A. Cruz]

I'm usuing an application that sporadically gets an error doing an ioctl
call with VIDIOCMCAPTURE when streaming in frames.

It looks like there's a bug in cpia_usb_streamRead. It's assuming that
once it goes to sleep, it will only be awakend when a frame is ready,
but this is not the case. I have a patch that seems to fix this by
switching

if (ucpia->curbuff->status != FRAME_READY)
to
while(ucpia->curbuff->status != FRAME_READY)
(plus a little checking of other things.)

Since cpia.c is calling this with noblock set to 0, but it's not really
blocking (due to having that if instead of a while), this seems like the
proper fix put put in.

So... what's next to get things fixed? And to get the kernel updated
also? The project seems quite dormant of late.



_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com