Tux, the Linux mascotte


- News
- Download Latest Release
- Supported cards
- Documentation & FAQ
- Development & Programming
- SourceForge Project Page
- SourceForge CVS Repository


- Ronald Bultje
- Laurent Pinchart
- Mailinglist


- MJPEG-tools
- Linux Video Studio
- XawTV

Manufacturer Links

- Linux Media Labs
- Iomega
- Pinnacle Systems
- Philips Semiconductors
- Analog Devices
- Brooktree (Conexant)
- Micronas
- Zoran

SourceForge Logo

Main Page

Welcome to the website of the Linux Unified Zoran Driver. The Unified Zoran Driver is a generic driver for the zoran-360x7 PCI chipset, that is mostly used in MJPEG hardware capture devices, such as the Iomega Buz, Pinnacle/Miro DC10+ and the Linux Media Labs LML33. This website provides a driver for these cards, and it aims to provide some of the programming information available for these cards.


Ronald Bultje writes (13/10/2004): Just for the archives: the driver on this webpage should not be used unless you're running an older 2.4.x kernel. Your current 2.6.x kernel contains the most up-to-date Zoran driver, and I intend to keep the current CVS here in sync with the most recent 2.6.x kernels. In your kernel, the module is called zr36067.ko, so just compile a vanilla kernel, run "/sbin/modprobe zr36067" and enjoy. I would also like to mention that the CVS tree that was once called ZORAN_VIDEODEV_2 has been moved to the HEAD CVS. The old (0.7.x) tree is branched off, see the webCVS for details.

Ronald Bultje writes (26/03/2003): A new release of your all-time favourite driver (well, the WLAN drivers are cooler, but still ;-) ) is available! This time, we (me) have decided unanymously to call this release 0.9.4!

Changes since our last episode:

  • Add YUY2 overlay (fullscreen/hardware-scaled)
  • use __stringify() for array insmod options
  • fix stupid flaw in open() logic (could result in oops)
  • fix 2.5.x build
  • remove lots of 'dead' includes from zoran_driver.c
  • remove DataErr from the zr36060 IRQMask, this autoresets the chip when it loses sync, which means we don't get a timeout. Much better than all the manual restart handling (which only worked partly).
  • introduce workaround for kfree-in-munmap-handler oops
  • implement support for bitmaps rather than clips for overlay
  • use try_module_get()/module_put() instead of MOD_INC/DEC_USE_COUNT in 2.5.x... Also implement i2c_driver->owner for all i2c clients

This means, in short, that it shouldn't timeout anymore, it shouldn't oops anymore, and it should use 'true' fullscreen now (hardware-scaled) on Xv-capable videocards. Also, it should now work under 2.5.x (compilation is tested). I've had one report that the field order is sometimes spontaneously switched after a signal glitch. If anyone encounters it, I'd be interested in knowning what conditions trigger it (of course, it shouldn't do this).

Go to the Download-page to fetch this latest release!

Last modified: October 13, 2004 - 06:43:36
© Ronald Bultje, 2003 - under the terms of the GNU Free Documentation License