This subject doesn't ever seem to go away ;) From my point of view, there is no current solution for the CN400 & CRT. I've bodged my own together - see further down.
The following drivers are the ones i've tested, along with my findings with them (don't forget I'm only testing on the CN400 with CRT - you're milage may vary). I'm using Gentoo with the 2.6.15-r1 kernel. DirectFB CVS Version (linux-viafb) ---------------------------------- White Screen on boot (reported fixed so should be ok now!) fbset doesnt work (even -i output is incorrect) refresh rate appears to be hardcoded fb.modes is ignored bug in kernel line parameters which crashes when you add mode= Via Official Version (linux-fbdev-kernel-src_20050726) ------------------------------------------------------ accel=1 (hardware acceleration) crashes (garbage filled screen) ypaning not available without acceleration WAITFORVSYNC appears to be missing? directfb works, but no double buffering (so useless really) I'm sure there are plenty of other differences, (for instance everyone says the tv timings on the official version are rubbish). The only way i've managed to get a working directfb framebuffer with my CN400 board is bodging the CN400 crt setup code from the offical dirver into the DirectFB cvs version. It only took a couple of days, and is pretty awful - but it works fine, fbset (refresh and all) and correct timings (i can now go to any supported res/refresh at any time, and jump between terminals!). I'm not sure how useful it is, but like i said before if anyone wants it just give me a yell (someone did take a copy - but no idea how they got on with it!). I won't post it on the list as it contains large official via files and it's most probably not useful to many people. Remember, this is only for CN400 + CRT ONLY + KERNEL 2.6 - NO TV!. As a side note, I have some weird findings with DirectFB and different drivers. No matter what combination of source / destination RGB 24 and ARGB 32 I cannot always get accelerated blits of different types (colourkey/stretch/alpha). And the new frankenstien driver I've created has significantly improved my line drawing speeds...whats that all about??? Liab Nicolas Huillard wrote: > (I cc. the DirectFB ML, as it could interest people there. We're talking > about a newer viafb driver found at > http://davesdomain.org.uk/viafb/index.php) > > David Boucher a écrit : > >>I can certainly see the point of collaborating on this, but I am >>uncertain as to what would be the best way to proceed. My driver is >>quite immature at the moment and lacks many of the features found in the >>DirectFB CVS driver. Also, if the DirectFB version contains any code >>from the official driver, would merging the two version harm the chances >>of getting the driver into the kernel some day (I'm not sure)? I would >>appreciate any comments that you have about these points. > > > I'm not sure there is problems with licence from VIA, WRT the kernel > frame-buffer driver. There's only a problem in the way they handle the > open-source *process*. > > The newest DirectFB driver has the following advantages : > * adapted to kernel 2.6 (up to 2.6.14 or so, as CVS comments say) > * fully functionnal under DirectFB (layers, blits, etc.) > * fine-tuned no-scale TV-out tables (very important) > and disadvantages : > * does not fully implement newer things for newer chipsets, except with > quick patch hacks (VIA is more up-to-date in this regard : CN400) > * timing tables for TV-out work for only a subset of TV encoders, and > their format is incompatible with VIA's newset ones (which are crap, as > some say) and openchrome's ones. > > I think the amount of work to update the new driver in DirectFB's CVS is > not that much. Maintenance will imply copying code from VIA's driver, > from time to time. Improving TV tables can be easier if they had the > same format as their counterpart in the openchrome X driver (not that > difficult, either). > > >>That said, I would be happy to make a contribution if I can, as the main >>use of the driver appears to be to run DirectFB programs (from the >>feedback I have been getting). > > > That's what I also understand. Simple bootsplash could be implemented on > vesafb. That only leave DirectFB, MPlayer (and maybe Xine) users for the > audience of this kind of software. > _______________________________________________ directfb-users mailing list [email protected] http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-users
