EPIA system with kernel 2.4.25 + epia2 patch DirectFB-0.9.22 ffmpeg-0.4.8 DFB++-0.9.22 vdr-1.3.30 softdevice-CVS ( 20050823 ) alsa-1.0.9
One installation is based on a stripped down RH9 with gcc 3.2.2 20030222 (RedHat). The other is a Core Linux install with gcc 3.2.
I have used the exact same software packages for both installations.
viafb is loaded without any parameters ( modprobe viafb )
On the RedHat installation I can run vdr -P"softdevice -vo dfb:" and I get a picture on my TFT. Then I reboot with the Core Linux install and I only get a black screen after starting vdr. I have the same directfbrc, the same fb.modes but I just can't get it to work on the Core Linux system. The output log is attached.
When vdr is running ( with a black screen ) I get the following:
# fbset -i
mode "800x600-96" # D: 36.000 MHz, H: 33.834 kHz, V: 96.394 Hz geometry 800 600 800 1200 32 timings 27778 56 80 79 11 128 12 hsync high vsync high laced true rgba 8/16,8/8,8/0,8/24 endmode
Frame buffer device information: Name : Via CLE266 Address : 0xd8000000 Size : 33554432 Type : PACKED PIXELS Visual : TRUECOLOR XPanStep : 0 YPanStep : 1 YWrapStep : 0 LineLength : 3200 MMIO Address: 0xdc000000 MMIO Size : 16777216 Accelerator : Unknown (77)
The only difference as I see it is the compiler ( 3.2.2-redhat versus 3.2 ).
Any clues?
On Mittwoch, 24. August 2005 17:57, Per Mellander wrote:
EPIA system with kernel 2.4.25 + epia2 patch DirectFB-0.9.22
Is that from a tarball or cvs? DirectFB/gfxdrivers/unichrome/uc_overlay.h should be from 2005-02-27 or later, so that there is my tiny patch included (see below - SetLevel()).
My epia system is gentoo based with kernel 2.6.7 and epia patch. Last time I tested this box dfb worked ok.
My modprobe lines are: modprobe cinergyT2 modprobe snd-via82xx modprobe viafb mode=1024x768 bpp=32 refresh=60 TVon=0
and my /etc/directfbrc on this box is: #mode = 1024x768 mode = 1280x1024 #pixelformat=AiRGB pixelformat=ARGB disable-module=cle266
which results in:
root # fbset -i
mode "1024x768-76" # D: 78.653 MHz, H: 59.949 kHz, V: 75.694 Hz geometry 1024 768 1024 1536 32 timings 12714 128 32 16 4 128 4 rgba 8/16,8/8,8/0,8/24 endmode
Frame buffer device information: Name : CLE266 Address : 0xd8000000 Size : 66842624 Type : PACKED PIXELS Visual : TRUECOLOR XPanStep : 0 YPanStep : 1 YWrapStep : 0 LineLength : 4096 MMIO Address: 0xdc000000 MMIO Size : 16777216 Accelerator : Unknown (77)
The only line that looks strange is: [dfb] SetParams: action=IDirectFBDisplayLayer::SetLevel(int), result=Not supported! Failed: SetLevel()
On Thu, 25 Aug 2005 07:54:31 +0200, Stefan Lucke wrote
It's from a tarball, -rwxr-xr-x 1 2009 2009 3556 Nov 23 2004 uc_overlay.h
I will give CVS a try to get it all up to date. Anyway, I finally got it working! I'm now using:
modprobe viafb mode=800x600 bpp=32 refresh=60 TVon=0
directfbrc: mode = 800x600 pixelformat = ARGB disable-module=cle266
I'm not sure if it's the TVon=0 that does it becuase I was able to run mplayer with -vo directfb when vdr+softdevice did not work. What is this disable-module good for? If I had configured with --with-gfxdrivers=unichrome only, will it work then. What I'm actually asking is - What's the difference between unichrome and cle266? ( In DirectFB's perspective )
Thanks,
/Mel
I have patched the file and recompiled. SetLevel() error is gone but same behaviour.
I think I really found the issue by now. I can run mplayer and dfbshow via ssh but not VDR. VDR does only work when I start it from the console.
Is there a way to trick VDR, or ie. softdevice+directfb, to be run via ssh login?
/Mel
On Thu, 25 Aug 2005 11:47:17 +0300, Per Mellander wrote
+----------------------------+ Per Mellander Sjöbonsvägen 20 434 96 Kungsbacka 0300-21345 0708-141613 http://www.mellander.org/per ICQ:288773560 +----------------------------+
Per Mellander a écrit :
The difference should be "running as root" vs. "running as any user". Try the vt switching directives in directfbrc (see man or ML archives).
...
unichrome is a newer DirectFB driver than cle266, adding a lot of new features. This is kind of v0.4 vs. v0.3 disable-module allows to use the unichrome driver by avoiding testing the cle266 one, which should match the AcceleratorID first.