Skip to contentSkip to conversation

2.6-test11 framebuffer Matrox

DISCUSSION Steffen Schwientek 7 participants 12 messages Dec 19, 2003 – Dec 21, 2003

lore
Share & tools
lore
#0 lore raw

My Matrox-framebuffer is not working properly. Build direct into the kernel, the monitor will be black with some stripes at startup, just the reset button works.
Build as a modules, the same happens if I load the module.

The make xconfig script also advice me to compile some 8,16,24 and 32 bpp packed pixel too, but I cant find them in the 2.6 kernel configuration

Steffen

#1 lore raw

Hello Steffen,

Quoted 4 lines

My Matrox-framebuffer is not working properly. Build direct into the kernel, the monitor will be black with some stripes at startup, just the reset button works.
Build as a modules, the same happens if I load the module.

All I can say is "me, too". 8-(

Quoted 1 line

Steffen

Can somebody definately say if "matroxfb" is working for 2.6? I haven't tested it for a while, but I was surprised to find it non-working in 2.6...

CU
Michael.

#2 lore raw
Quoted 8 lines · Michael Hunold

Hello Steffen,

Quoted 4 lines

My Matrox-framebuffer is not working properly. Build direct into the kernel, the monitor will be black with some stripes at startup, just the reset button works.
Build as a modules, the same happens if I load the module.

Quoted 6 lines

All I can say is "me, too". 8-(

Quoted 1 line

Steffen

Can somebody definately say if "matroxfb" is working for 2.6? I haven't tested it for a while, but I was surprised to find it non-working in 2.6...

The matrox framebuffer works with 2.6.0-test11 and a matrox G550. This is statically compiled, with smp and preempt.

It dies if I use fbcon + the ruby patch (for multiple keyboard support) though. So something is wrong - this patch changes the console but not the fb driver so it isn't supposed to crash. I guess there's some problem with the driver that gets triggered only in some circumstances.

Helge Hafting

#3 lore raw
In reply to Michael Hunold
Quoted 2 lines · Michael Hunold

Can somebody definately say if "matroxfb" is working for 2.6? I haven't tested it for a while, but I was surprised to find it non-working in 2.6...

Not working for me. G550 driving Samsung SyncMaster 171s (LCD).
It locked system several times in the past when switching between XFree and fb console.
I was told to use exact resolution and color depth wi fbcon and X.
In my case, it's 1280x1024-16bpp. It is working great in X, but fbcon do not work - all I get is a black screen and lock when I try to blinfly run XFree.

Signature · 3 lines
Tomasz Torcz            There exists no separation between gods and men:
[email protected]   one blends softly casual into the other. 
|> Playing: Marlow - exposed_(frankman_remix) ...
#4 lore raw
Quoted 5 lines · Tomasz Torcz
Quoted 1 line · Michael Hunold

Can somebody definately say if "matroxfb" is working for 2.6? I haven't tested it for a while, but I was surprised to find it non-working in 2.6...

Quoted 8 lines

Not working for me. G550 driving Samsung SyncMaster 171s (LCD).
It locked system several times in the past when switching between XFree and fb console.
I was told to use exact resolution and color depth wi fbcon and X.
In my case, it's 1280x1024-16bpp. It is working great in X, but fbcon do not work - all I get is a black screen and lock when I try to blinfly run XFree.

last time i tried it (-test7 or so) it worked. your normal fbcon and what not drivers do not work. They haven't been updated to handle the new fb code. You'll need to do all your configuring via kernel or module loading and cross your fingers and hope xfree86 can load based off it.
Also, i never got a non-blank console when using matroxfb... once it
loaded i always had to blindly move around and start X and what not.

But, what's really the point in using X with matroxfb? You lose half your memory off the bat that X cannot access and you get no added performance or anything. It really does not seem worth it. Use matroxfb if you need to do fb stuff in a console without X. Otherwise, stick to just plain console and X and forget about matroxfb.

#5 lore raw
Quoted 2 lines · Ed Sweetman

But, what's really the point in using X with matroxfb?

In my case, the point is to run two instances of X, one on each of the two framebuffers. I can then have two simultaneous users, each using one of the two framebuffers. Two pc's in one!

Quoted 3 lines

You lose half
your memory off the bat that X cannot access and you get no added performance or anything.

Another X accesses the other memory :-)

Quoted 3 lines

It really does not seem worth it. Use
matroxfb if you need to do fb stuff in a console without X. Otherwise, stick to just plain console and X and forget about matroxfb.

Sure - for single-user setups.

Helge Hafting

#6 lore raw
In reply to Ed Sweetman
Quoted 3 lines · Ed Sweetman

But, what's really the point in using X with matroxfb? You lose half your memory off the bat that X cannot access and you get no added performance or anything. It really does not seem worth it. Use

32 MiB of videoram is more than enough.
And the point is: comfortable work in console (bigger resolution -> more information on screen), sometimes even with small fbtv window in corner. And ability to run browsers like mozilla in X, when it's needed. And ability to watch movies with mplayer in both.

I don't understand why matroxfb has regressed in 2.6 compared to 2.4.

Signature · 3 lines
Tomasz Torcz               RIP is irrevelant. Spoofing is futile.
[email protected]     Your routes will be aggreggated. -- Alex Yuriev
|> Playing: Electric Rudeboyz - Beton ...
#7 lore raw
In reply to Steffen Schwientek

Hello Steffen,

Quoted 4 lines

My Matrox-framebuffer is not working properly. Build direct into the kernel, the monitor will be black with some stripes at startup, just the reset button works.
Build as a modules, the same happens if I load the module.

Hmm, I just tested with the 2.6.0 release and my both cards are working properly now. The only thing is a huge white box around the penguin logo.

Here is "lspci -v" of the cards I tested:

00:0a.0 VGA compatible controller: Matrox Graphics, Inc. MGA G200 (rev 01) (prog-if 00 [VGA])
Subsystem: Matrox Graphics, Inc. Millennium G200 SD
Flags: bus master, medium devsel, latency 64, IRQ 11
Memory at dd000000 (32-bit, prefetchable) [size=16M]
Memory at dfffc000 (32-bit, non-prefetchable) [size=16K]
Memory at df000000 (32-bit, non-prefetchable) [size=8M]
Expansion ROM at dffe0000 [disabled] [size=64K]
Capabilities: [dc] Power Management version 1

01:00.0 VGA compatible controller: Matrox Graphics, Inc. MGA G400 AGP (rev 03) (prog-if 00 [VGA])
Subsystem: Giga-byte Technology GA-G400
Flags: bus master, medium devsel, latency 64, IRQ 11
Memory at dc000000 (32-bit, prefetchable) [size=32M]
Memory at dfafc000 (32-bit, non-prefetchable) [size=16K]
Memory at df000000 (32-bit, non-prefetchable) [size=8M]
Expansion ROM at dfae0000 [disabled] [size=64K]
Capabilities: [dc] Power Management version 2
Capabilities: [f0] AGP version 2.0

Quoted 1 line

Steffen

CU
Michael.

#8 lore raw
Quoted 2 lines · Michael Hunold

Hmm, I just tested with the 2.6.0 release and my both cards are working properly now. The only thing is a huge white box around the penguin logo.

As I said, this happens to me without Petr's patch, and goes away with it. Can you check out my earlier post in this thread, and analyze your X+fbcon interaction with reagrds to fbdev vs. mga driver, and Petr's patch?

Signature · 3 lines
nick black <[email protected]>
"np:  nondeterministic polynomial-time
the class of dashed hopes and idle dreams." - the complexity zoo
#9 lore raw
In reply to Steffen Schwientek

Hmm, as I should have noticed, matrox framebuffer is totally UNUSABLE with any 2.6.0-test kernels ;-( Several problems occured, like no picture all, unstable picture, corruption on screen all the times, crashes, etc. Both with G400 and G200. Everything works with 2.4.x kernel though. This is the major point I can't upgrade to 2.6.0 ...
Maybe somebody try to port fb layer of 2.4.x to 2.6? I write this because AFAIK in 2.6 there were a major rewrite of framebuffer support 'needed for the embedded industry' (or something similar). As I've noticed, somebody even thinks that framebuffer is only used for 'getting the tux logo'. But I really need it, because of playing video on framebuffer probably with multiple heads, hardware acceleration etc etc etc ...

Quoted 14 lines · Steffen Schwientek

My Matrox-framebuffer is not working properly. Build direct into the kernel, the monitor will be black with some stripes at startup, just the reset button works.
Build as a modules, the same happens if I load the module.

The make xconfig script also advice me to compile some 8,16,24 and 32 bpp packed pixel too, but I cant find them in the 2.6 kernel configuration

Steffen
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Signature · 1 line
- Gábor (larta'H)
#10 lore raw
In reply to Steffen Schwientek
Quoted 4 lines · Steffen Schwientek

My Matrox-framebuffer is not working properly. Build direct into the kernel, the monitor will be black with some stripes at startup, just the reset button works.
Build as a modules, the same happens if I load the module.

I've managed to get my AGP G400 working in the following setups under 2.6.0 since -test7:

with Petr's patch, reverting to 2.4 interfaces:
video=matroxfb:vesa:0x1bb gives me a great fbcon. I can either use X's fbdev driver at the same resolution, or the mga driver at any resolution, with no problems.

without Petr's patch:
video=matroxfb:vesa:0x1bb gives me a great fbcon. I can only use X's fbdev driver at the same resolution; the mga driver at any resolution causes massive console corruption upon switching from X to console and causing any screen changes. Also, I get a large white block underneath the logo on boot.

Signature · 3 lines
nick black <[email protected]>
"np:  nondeterministic polynomial-time
the class of dashed hopes and idle dreams." - the complexity zoo
#11 lore raw
Quoted 14 lines · nick black
Quoted 4 lines · Steffen Schwientek

My Matrox-framebuffer is not working properly. Build direct into the kernel, the monitor will be black with some stripes at startup, just the reset button works.
Build as a modules, the same happens if I load the module.

I've managed to get my AGP G400 working in the following setups under 2.6.0 since -test7:

with Petr's patch, reverting to 2.4 interfaces:
video=matroxfb:vesa:0x1bb gives me a great fbcon. I can either use X's fbdev driver at the same resolution, or the mga driver at any resolution, with no problems.

Quoted 6 lines

without Petr's patch:
video=matroxfb:vesa:0x1bb gives me a great fbcon. I can only use X's fbdev driver at the same resolution; the mga driver at any resolution causes massive console corruption upon switching from X to console and causing any screen changes. Also, I get a large white block underneath the logo on boot.

Nice. :-) and :-(
With that patch, now I'm happy user of 2.6.0 ;-)
[I was that since 2.6.0-test1, but only on one machine of mine, where fb
was not important at all]

My only question is: why was fb layer modofied to a state which is UNUSABLE without a patch which reverts to the 2.4 interface? I'm trying to search archives and various document, and now I've got the feeling that several author thinks fbcon only a toy and its ONLY purpose to get the tux logo on booting. Which is completly false of course ;-(

SORRY (!) if I missunderstood something ...
Or at least please give me some points what is the advantage of the framebuffer in the state it exists in the official 2.6.0 if it is unusable. Or maybe I've missed something. My only headache that I would like know what is the goal with fb in 2.6.x then?

And again: sorry, I would not like to be offensive or something, but I'm really interested in this topic!

- Gábor (larta'H)