[MPlayer-dev-eng] fbdev:vidix / vidix bug

[Alex Beregszaszi]

On Fri, Jan 18, 2002 at 10:01:41AM +0200, Arpi wrote:
> Hi,
> 
> > Anyway - genfb is bad driver for VIDIX.
> > Even if linux-2.6 will have such extension for example from RUBY
> > tree of linuxconsole.sf.net (FBVID.H) or from V4L2 project then
> > it will not cause to implement such driver for VIDIX.
> > 
> > Please, REMEMBER:
> >  NEVER GENERIC DRIVERS FOR VIDIX ONLY NATIVE ONES (VENDOR_ID DEPENDED).
> 
> Alex, I agree with Nick in this. Vidix is a direct hardware access layer.
> If fbdev provides any hw accel, write a new vo_fbdevhwaccel or modify old
> vo_fbdev. it does not require direct hw access, it is handled by kernel and
> you can use ioctls to /dev/fb0.
> 
> btw, about vidix: i miss some important features:
> - IRQ handling
> - working as non-root
yes, this is needed to get users to change from mga_vid to vidix/mga_vid

> i think we should write (at least for linux, and maybe optional) a generic
> kernel driver to do the mapping to userspace, and to handle irqs.
You're right, this is a good idea

How do you think, how it can be done? Throught a block/char device (ioctls?) or
throught proc entries? I think the char_device&ioctl's is the more portable
way (under unix systems). (we had a portable project using this way to
communicate with userspace)

--
alex

> 
> 
> A'rpi / Astral & ESP-team
> 
> --
> mailto:arpi at thot.banki.hu
> http://esp-team.scene.hu
> _______________________________________________
> MPlayer-dev-eng mailing list
> MPlayer-dev-eng at mplayerhq.hu
> http://mplayerhq.hu/mailman/listinfo/mplayer-dev-eng