linux-fbdev-users Mailing List for Linux Frame Buffer Device Development

Dear Friends,
I am not able to initialize atyfb for my ATI Rage XL PCI on p4 with 
kernel 2.4.20. After kernel booting starts screen in becoming blank with 
random dots(mostly green and white) specifying redhat init() OK's.
where as vga16fb is working properly. I tried loading as a module still 
same problem persist.
Does any body initialized atyfb framebuffer for ATI Rage XL.
Please send your .config and autoconf.h if you are successful.
I request your help regarding this.
Regards,
Suresh

On recent kernel (2.6.4 and later) my Radeon 7000 doesn't display
anything. My Eizo monitor warns me of really incompatible frequencies
(fV 217.2 Hz, fH 271.5 kHz). fbset returns normal values, by the way.
I've found out that this chunk in patch-2.6.4, applying to
drivers/video/aty/radeon_base.c, causes my problems:
@@ -1329,6 +1320,16 @@
 * not sure which model starts having FP2_GEN_CNTL, I assume anything more
 * recent than an r(v)100...
 */
+#if 0
+ /* XXX I had reports of flicker happening with the cinema display
+ * on TMDS1 that seem to be fixed if I also forbit odd dividers in
+ * this case. This could just be a bandwidth calculation issue, I
+ * haven't implemented the bandwidth code yet, but in the meantime,
+ * forcing uses_dvo to 1 fixes it and shouln't have bad side effects,
+ * I haven't seen a case were were absolutely needed an odd PLL
+ * divider. I'll find a better fix once I have more infos on the
+ * real cause of the problem.
+ */
 while (rinfo->has_CRTC2) {
 u32 fp2_gen_cntl = INREG(FP2_GEN_CNTL);
 u32 disp_output_cntl;
@@ -1362,6 +1363,9 @@
 uses_dvo = 1;
 break;
 }
+#else
+ uses_dvo = 1;
+#endif
 if (freq > rinfo->pll.ppll_max)
 freq = rinfo->pll.ppll_max;
 if (freq*12 < rinfo->pll.ppll_min)
If I change that #if 0 to #if 1, I get a normal image. This is with the
following video-card:
01:00.0 VGA compatible controller: ATI Technologies Inc Radeon R100 QD [Radeon 7200] (prog-if 00 [VGA])
 Subsystem: ATI Technologies Inc Radeon 7000/Radeon
 Flags: bus master, stepping, 66Mhz, medium devsel, latency 64, IRQ 16
 Memory at d8000000 (32-bit, prefetchable) [size=128M]
 I/O ports at a000 [size=256]
 Memory at e5000000 (32-bit, non-prefetchable) [size=512K]
 Expansion ROM at <unassigned> [disabled] [size=128K]
 Capabilities: [58] AGP version 2.0
 Capabilities: [50] Power Management version 2
I've seen other messages about problems with radeonfb. Can other people
confirm this problem, and confirm that removing this chunk fixes the
problem?
Kind regards,
Jurriaan
-- 
Don't like these taglines? Steal your own...
Debian (Unstable) GNU/Linux 2.6.5-rc2-mm3 3940 bogomips 0.18 0.28

On 2004年3月20日, Suresh Chandra Mannava wrote:
> We are facing the problem of initializing the framebuffer for the below
> system.
> We are porting Xfree86 on to a new 32bit RISC processor. We have test
> FPGA system running at 4 Mega Hertz (Bogomips 0.16) on kernel 2.4.7.
> proposed system(ASIC) runs on ~300Mhz.
>
> My main concern is does Xfree86 can be tested on 2.6Mhz FPGA?
> What is the minimum clock speed required to run X?
> Our test system supports 2GB of RAM and it consists of PCI ATI Rage XL
> Graphics card.
> We think that speed might be a problem.We need to findout first whether
> we will be able to run a framebuffer driver at 4 MHZ.Since framebuffer
> does pixel by pixel access,we think it will take a lot of time and
> anything can happen in between.We are now not in a position to predict
> what is happening.Please give your points on whether the framebuffer
> will work at such a low speed.
>
> I request your suggestions on feasibility of initializing framebuffer
> and testing X on the above system.
Since X was developed in mid-1980s (using unaccelerated graphics hardware) it
can definitely run on such slow CPU. Ater all your graphics card is quite fast.
And fbdev was developed on m68k machines, running on 16-50 MHz CPUs with
unaccelerated graphics hardware as well.
Of course you don't want to try GNOME and KDE, but simple things like twm and
xterm should be usable.
Gr{oetje,eeting}s,
						Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- ge...@li...
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							 -- Linus Torvalds

Dear Friends, 
We are facing the problem of initializing the framebuffer for the below 
system.
We are porting Xfree86 on to a new 32bit RISC processor. We have test 
FPGA system running at 4 Mega Hertz (Bogomips 0.16) on kernel 2.4.7. 
proposed system(ASIC) runs on ~300Mhz.
My main concern is does Xfree86 can be tested on 2.6Mhz FPGA?
What is the minimum clock speed required to run X?
Our test system supports 2GB of RAM and it consists of PCI ATI Rage XL 
Graphics card.
 We started testing the generic vga16fb.c framebuffer.The basic 
intialization seems to happen perfectly.The problem comes in 
take_over_console,where framebuffer takes over the dummy console.In the 
take over console,at first step he is calling fbcon_startup which is 
getting screen base ,X and Y coordinates and clearing that location.This 
happens fine.This function is called for after some timer interval and 
hence we can say that it will be called repeatedly.The problem is that 
we cann't put any printk in between as it adversely affects the 
flow(This we found after testing vga16fb.c rigorously on intel as well 
as on our FPGA).We traversed the source code and also through object 
dumps and found that nothing is wrong.The system seems to be hung after 
some time(It is able to clear the screen).We came to this conclusion 
after giving a run and waiting for 6 hours.
We think that speed might be a problem.We need to findout first whether 
we will be able to run a framebuffer driver at 4 MHZ.Since framebuffer 
does pixel by pixel access,we think it will take a lot of time and 
anything can happen in between.We are now not in a position to predict 
what is happening.Please give your points on whether the framebuffer 
will work at such a low speed.
I request your suggestions on feasibility of initializing framebuffer 
and testing X on the above system.
Regards,
Suresh Chandra Mannava.
-- 
Suresh Chandra Mannava		
Software Engineer, Cornet Technology India, Chennai.
CSE, Research Student, Vellore Institute of Technology - India.
Email: mannava(at)vit.ac.in, Mobile: +919884278813

> After booting 2.6.5-rc1-mm1 with this command-line:
> 
> kernel /vmlinuz-265rc1mm1 root=/dev/hda3 video=radeonfb:1600x1200-32@75 acpi=force
> 
> I get a warning on my monitor:
> 
> fH 271.5 kHz
> fV 217.2 Hz
I haven't seen that before :-/ The kernel isn't generating that. I just 
looked threw the code.
> That is obviously not what I meant by 1600x1200-32@75.
> 
> fbset -V tells me:
> 
> Linux Frame Buffer Device Configuration Version 2.1 (23/06/1999)
> (C) Copyright 1995-1999 by Geert Uytterhoeven
> 
> 
> mode "1600x1200-75"
> # D: 202.511 MHz, H: 93.755 kHz, V: 75.004 Hz
> geometry 1600 1200 1600 1200 32
> timings 4938 304 64 46 1 192 3
> hsync high
> vsync high
> rgba 8/16,8/8,8/0,0/0
> endmode
This looks right.
> Obviously, it's bad that I define what video-mode I want and radeonfb
> starts changing that to bogus values. The monitor is an Eizo F931,
> connected via a Rose Electronics Vista KVM.
What is you want exactly?

[this was meant to go to linux-kernel and fbdev-users at the same time,
but I made a typo. sorry]
After booting 2.6.5-rc1-mm1 with this command-line:
kernel /vmlinuz-265rc1mm1 root=/dev/hda3 video=radeonfb:1600x1200-32@75 acpi=force
I get a warning on my monitor:
fH 271.5 kHz
fV 217.2 Hz
That is obviously not what I meant by 1600x1200-32@75.
fbset -V tells me:
Linux Frame Buffer Device Configuration Version 2.1 (23/06/1999)
(C) Copyright 1995-1999 by Geert Uytterhoeven
mode "1600x1200-75"
 # D: 202.511 MHz, H: 93.755 kHz, V: 75.004 Hz
 geometry 1600 1200 1600 1200 32
 timings 4938 304 64 46 1 192 3
 hsync high
 vsync high
 rgba 8/16,8/8,8/0,0/0
endmode
dmesg:
Console: switching to colour frame buffer device 133x54
radeonfb: Found Intel x86 BIOS ROM Image
radeonfb: Retreived PLL infos from BIOS
radeonfb: Reference=27.00 MHz (RefDiv=60) Memory=166.00 Mhz, System=166.00 MHz
radeonfb: No connector info table detected
radeonfb: Monitor 1 type CRT found
radeonfb: ATI Radeon QD DDR SGRAM 64 MB
lspci:
01:00.0 VGA compatible controller: ATI Technologies Inc Radeon R100 QD [Radeon 7200] (prog-if 00 [VGA])
 Subsystem: ATI Technologies Inc Radeon 7000/Radeon
 Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping+ SERR- FastB2B-
 Status: Cap+ 66Mhz+ UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
 Latency: 64 (2000ns min), Cache Line Size: 0x08 (32 bytes)
 Interrupt: pin A routed to IRQ 16
 Region 0: Memory at d8000000 (32-bit, prefetchable) [size=128M]
 Region 1: I/O ports at a000 [size=256]
 Region 2: Memory at e5000000 (32-bit, non-prefetchable) [size=512K]
 Expansion ROM at <unassigned> [disabled] [size=128K]
 Capabilities: [58] AGP version 2.0
 Status: RQ=48 Iso- ArqSz=0 Cal=0 SBA+ ITACoh- GART64- HTrans- 64bit- FW- AGP3- Rate=x1,x2
 Command: RQ=1 ArqSz=0 Cal=0 SBA+ AGP- GART64- 64bit- FW- Rate=<none>
 Capabilities: [50] Power Management version 2
 Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
 Status: D0 PME-Enable- DSel=0 DScale=0 PME-
.config:
CONFIG_FB=y
CONFIG_FB_RADEON=y
CONFIG_FB_RADEON_I2C=y
Obviously, it's bad that I define what video-mode I want and radeonfb
starts changing that to bogus values. The monitor is an Eizo F931,
connected via a Rose Electronics Vista KVM.
Any hints on solving this?
Thanks,
Jurriaan
-- 
Thesaurus (n): ancient reptile with excellent vocabulary.
Debian (Unstable) GNU/Linux 2.6.4-rc1-mm2 3940 bogomips load av: 0.00 0.00 0.00

> > I have the specs and the hardware. 
> 
> Any news on that ? Did you manage to rescue the specs from your damaged
> harddisk ? 
Not yet. I managed last night to finish the install on my system. 

On Mon, Mar 08, 2004 at 07:14:02PM +0000, James Simmons wrote:
> 
> > On Fri, Feb 27, 2004 at 05:36:22PM +0000, James Simmons wrote:
> > > 
> > > Its a endian issue. Been there :-( The voodoo card needs to be set to big 
> > > endian access.
> > 
> > Any hint on how to do that ? I have neither voodoo specs, nor access to
> > the hardware, still i would very much like to have it working.
> 
> I have the specs and the hardware. 
Any news on that ? Did you manage to rescue the specs from your damaged
harddisk ? 
Friendly,
Sven Luther

That is correct... Had it worked immediately, I'd not have bothered to
post ;-)
With the adjusted pixel clock, I am able to select between 640x480x16bpp
and 800x600x16bpp (perhaps others) using the vesa modes, without
trouble. The weirdest thing was that fbset would change the patterns on
the TV, so I thought it was working. It looks like everything was
changing except the pixel clock...
=09
-----Original Message-----
From: Geert Uytterhoeven [mailto:ge...@li...]=20
Sent: Tuesday, March 09, 2004 5:17 AM
To: Hihn, Jason
Cc: lin...@li...
Subject: Re: [Linux-fbdev-users] Frambeffer to TV troubles
On Mon, 8 Mar 2004, Hihn, Jason wrote:
> I've got a ATI Graphics Expression+ PC2TV (ATI MACH64/Rage II GT) and
I
> finally got TV out to work. The perplexing thing was it's boot up and
> you could see output on the TV (using composite out), but when asked
for
> a video mode it's lose it after that. I played around with fbset,
> getting the monitor to do all kinds of funky things, but never a
> picture. (I was playing with resolutions and timings) Oh, and I was
> using the Vesa driver
>
>
>
> Then I went into include/linux/fb.h and made the default videomode's
> timings as I wanted: 20330, 128, 32, 32, 8, 128, 5 (NTSC) which
worked!
> Now I use mode 0x311 and I can get a picture. It seems that the
pixclock
> is hardcoded to some non-NTSC compatible values.
So if you pass these values using fbset, it doesn't work?
Strange...
Gr{oetje,eeting}s,
						Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 --
ge...@li...
In personal conversations with technical people, I call myself a hacker.
But
when I'm talking to journalists I just say "programmer" or something
like that.
							 -- Linus
Torvalds

On Mon, 8 Mar 2004, Hihn, Jason wrote:
> I've got a ATI Graphics Expression+ PC2TV (ATI MACH64/Rage II GT) and I
> finally got TV out to work. The perplexing thing was it's boot up and
> you could see output on the TV (using composite out), but when asked for
> a video mode it's lose it after that. I played around with fbset,
> getting the monitor to do all kinds of funky things, but never a
> picture. (I was playing with resolutions and timings) Oh, and I was
> using the Vesa driver
>
>
>
> Then I went into include/linux/fb.h and made the default videomode's
> timings as I wanted: 20330, 128, 32, 32, 8, 128, 5 (NTSC) which worked!
> Now I use mode 0x311 and I can get a picture. It seems that the pixclock
> is hardcoded to some non-NTSC compatible values.
So if you pass these values using fbset, it doesn't work?
Strange...
Gr{oetje,eeting}s,
						Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- ge...@li...
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							 -- Linus Torvalds

I've got a ATI Graphics Expression+ PC2TV (ATI MACH64/Rage II GT) and I
finally got TV out to work. The perplexing thing was it's boot up and
you could see output on the TV (using composite out), but when asked for
a video mode it's lose it after that. I played around with fbset,
getting the monitor to do all kinds of funky things, but never a
picture. (I was playing with resolutions and timings) Oh, and I was
using the Vesa driver
=20
Then I went into include/linux/fb.h and made the default videomode's
timings as I wanted: 20330, 128, 32, 32, 8, 128, 5 (NTSC) which worked!
Now I use mode 0x311 and I can get a picture. It seems that the pixclock
is hardcoded to some non-NTSC compatible values.=20
=20
I quickly tried to get the ATI driver working, but I never succeeded.
The Vesa driver works, but to try a new setting, I have to re-compile
and reboot.
=20
But I'd like to have it easier to do, I feel that I hacked it up and
that I'm doing something wrong. Can anyone give me some pointers? I'd
like to see my screen in 768x576 resolution, but there is no vesa mode
for that.
=20
Thanks
=20

> On Fri, Feb 27, 2004 at 05:36:22PM +0000, James Simmons wrote:
> > 
> > Its a endian issue. Been there :-( The voodoo card needs to be set to big 
> > endian access.
> 
> Any hint on how to do that ? I have neither voodoo specs, nor access to
> the hardware, still i would very much like to have it working.
I have the specs and the hardware. 

On Mon, Mar 08, 2004 at 05:44:17PM +0000, James Simmons wrote:
> 
> I have the specs on my hard drive that just crashed. I plan to recover the 
> data this week. It is possible to fix.
Ok, thanks.
Friendly,
Sven Luther

I have the specs on my hard drive that just crashed. I plan to recover the 
data this week. It is possible to fix.
On Mon, 8 Mar 2004, Sven Luther wrote:
> On Fri, Feb 27, 2004 at 05:36:22PM +0000, James Simmons wrote:
> > 
> > Its a endian issue. Been there :-( The voodoo card needs to be set to big 
> > endian access.
> 
> Any hint on how to do that ? I have neither voodoo specs, nor access to
> the hardware, still i would very much like to have it working.
> 
> Friendly,
> 
> Sven Luther
> 

On Fri, Feb 27, 2004 at 05:36:22PM +0000, James Simmons wrote:
> 
> Its a endian issue. Been there :-( The voodoo card needs to be set to big 
> endian access.
Any hint on how to do that ? I have neither voodoo specs, nor access to
the hardware, still i would very much like to have it working.
Friendly,
Sven Luther

On Fri, Feb 27, 2004 at 05:40:36PM +0000, James Simmons wrote:
> 
> > Hello,
> > Aty128fb in 2.6.3 and earlier doesn't work as it should for me.
> > First, when CONFIG_VESA=Y I cannot load module aty128fb.
> 
> You can load the module. You need to use the con2fb utility to map the 
> consoles to that framebuffer device.
> 
> > Second, if VESA support is not compiled in I got black screen when switch
> > to another console. I used to use mode 1024x768-75. (128x76)
> 
> ??? What is your config like? What do you mean by switching to another 
> console.
I'm sorry to bother you, it's OK now.
Framebuffer was set by:
for i in 1 2 3 4 5 6 7 8; do
	/usr/bin/open -c $i -- /usr/bin/fbset -fb /dev/fb/0 -depth 16 1024x768-75
done
Adding -w resolves problems.
> > Third, it is still slow.
> 
> Its unaccelerated. It needs hardware acceleration.
ATI 128 Rage Pro has hardware acceleration. What else is required?
-- 
Witold Filipczyk tohu-bohu
<wi...@po...>