On Thursday 25 January 2007 15:09, Nicolas Huillard wrote: > Laz a écrit : > >>viafb 2006-10-08 > >>DirectFB 1.0.0rc2 patched for field parity ioctl > >>DFB++ 2007-01-24 > >>libcle266mpegdec 0.5 > >>softdevice 0.3.1 > >>VDR 1.4.5 > > > > Is viafb patched to actually "listen for" the synching ioctl? > > I'm using viafb from DirectFB CVS 2006-10-08, which seems to be the > lastest one.
Looks like my version of viafb is from cvs on 2006-10-24. > Maybe you're talking about this patch to DirectFB, which makes DFB talk > to viafb's ioctl : > http://www.directfb.org/index.php/viewcvs.cgi/linux-viafb/patches/direc >tfb-viafb-assisted-field-parity.patch?rev=1.1&content-type=text/vnd.view >cvs-markup > > Yes, I'm using it, in an older form, which seems to be the same, tough. > > Software decoding leaves no interlacing artifacts, so it should be OK > on the lower levels (DirectFB / viafb). There is a patch to DirectFB which sends the FBIO_FLIPONVSYNC ioctl and a patch to viafb which "looks out for it". Looks like I still needed to patch viafb from 2006-10-24. Try doing: $ grep FBIO_FLIPONVSYNC linux-viafb/linux/drivers/video/c le266/via_fbobj.c I get: #ifndef FBIO_FLIPONVSYNC #define FBIO_FLIPONVSYNC _IOWR('F', 0x21, struct fb_flip) case FBIO_FLIPONVSYNC: with my (patched) version. Not sure whether this ever made it into linux-viafb cvs or not: I think Mark was going to rewrite it as less of a hack! I have found two versions of this patch, one from August 2005 and one dated 2006-10-24 (not sure whether I made the new one by patching by hand or Mark posted a new version). I can post this on here, if you like. If your viafb is from earlier than mine then it will need the patch (maybe it has now made it into cvs if the DirectFB patch is in there now...see below). > > I'm using vdr-1.5.0 with softdevice cvs from last October (patched > > with the lip-sync patch from yesterday), DirectFB, DFB++, and viafb > > from cvs back from last October (with Mark Adams' patches from a long > > while back), and libcle266mpegdec 0.5. > > Can you elaborate on DirectFB/DFB++ versions ? > It seems we have quite the same setup. You also said earlier that field > parity works for you since a long time, with earlier versions. Looking more closely, I _think_ DirectFB and DFB++ are both cvs from 2006-04-18. I have lots of different cvs trees which I really should clean out. > > I'm running vdr with "-vo dfb:viatv:cle266" and TV-out to connect to > > a CRT TV and, apart from the odd glitches and A-V sync wandering a > > bit, I get nice, smooth scrolling text (BBC News 24 is good for > > testing this!). > > What kernel version do you run ? It is currently running a 2.6.18.1 kernel. > > If I get a chance this weekend, I might have a go at upgrading some > > bits to see if something has changed to break this... > > If you could recall all patches you applied to everything, that could > help. Not that much, these times, I think. > I only patched DirectFB with something named > "dfb_unichrome_flip_ioctl-libcle266mpegdec" from Mark Adams, which > happens to be the one in linux-viafb CVS now (commited by Mark, BTW). My cvs version of linux-viafb does not contain the patch (sounds like I was a couple of weeks early for that!). The only patches I have are named: dfb_unichrome_flip_ioctl.patch dated 2006-08-14 (presumably close to the one now in linux-viafb cvs) and viafb_flip_ioctl.patch dated 2006-10-24. Just had a quick look at directfb.org and it would appear that Mark added the patch (or something looking very similar) a few days back: http://www.directfb.org/index.php/viewcvs.cgi/linux-viafb/linux/drivers/video/cle266/via_fbobj.c Your best bet may be to check out a new version of linux-viafb and see if that sorts it out. As you can see, I haven't really touched much for a while because it has been working quite well. I will have a go at updating DirectFB, DFB++, and linux-viafb to the current CVS versions this weekend and see if anything breaks for me! Cheers, Laz _______________________________________________ Softdevice-devel mailing list [email protected] https://lists.berlios.de/mailman/listinfo/softdevice-devel
