Re: 2.6.0-test2 does not boot with matroxfb

Re: 2.6.0-test2 does not boot with matroxfb

[Jocelyn Mayer]

> On  8 Aug 03 at 17:45, Juergen Rose wrote:
> ı I tried on my PC with a Matrox-G450 several kernel and boot options.
> ı Every time when the console should work with matrox framebuffer,
> linux
> ı was crashed. With 2.6.0-test2 and 2.6.0-test2-bk7 I had the
> following
> ı warning performing ''make modules_install''
> ı WARNING:
> ı
> /lib/modules/2.6.0-test2[-bk7]/kernel/drivers/video/matrox/matroxfb_crtc2.ko
> ı needs unknown symbol matroxfb_enable_irq This WARNING disapears for
> 
> I'm not able to get through Linus's mail filters for past three weeks.

I can boot a Athlon based PC with a Matrox-G450. It runs well.
The frame buffer is broken, and so is X and overlay buffer,
but it doesn't crash at all.
I also booted another kernel on the VGA console.
The console is OK, X and overlay buffer are still broken,
but the machine euns well.
I have two 2.6.0-test2, one with the matroxfb,
the other one just with the VGA console.

> 
> You can also try <a
> href="ftp://platan.vc.cvut.cz/pub/linux/matrox-latest/matroxfb-2.5.72.gz">ftp://platan.vc.cvut.cz/pub/linux/matrox-latest/matroxfb-2.5.72.gz</a>
> (if 2.6.0-test2-mm5 does not complain about unexported matrox_*_irq
> it will probably not apply cleanly to it). It restores fbcon subsystem
> to the 2.4.x's version.
>                                             Petr

I will check this, I'd like to have more than a VGA console :-)


-- 
Jocelyn Mayer <l_indien@magic.fr>

Re: 2.6.0-test2 does not boot with matroxfb

[Jocelyn Mayer]

On Sat, 2003-08-09 at 13:40, Jocelyn Mayer wrote:
> > On  8 Aug 03 at 17:45, Juergen Rose wrote:
> > ı I tried on my PC with a Matrox-G450 several kernel and boot options.
> > ı Every time when the console should work with matrox framebuffer,
> > linux
> > ı was crashed. With 2.6.0-test2 and 2.6.0-test2-bk7 I had the
> > following
> > ı warning performing ''make modules_install''
> > ı WARNING:
> > ı
> > /lib/modules/2.6.0-test2[-bk7]/kernel/drivers/video/matrox/matroxfb_crtc2.ko
> > ı needs unknown symbol matroxfb_enable_irq This WARNING disapears for
> > 
> > I'm not able to get through Linus's mail filters for past three weeks.
> 
> I can boot a Athlon based PC with a Matrox-G450. It runs well.
> The frame buffer is broken, and so is X and overlay buffer,
> but it doesn't crash at all.
> I also booted another kernel on the VGA console.
> The console is OK, X and overlay buffer are still broken,
> but the machine euns well.
> I have two 2.6.0-test2, one with the matroxfb,
> the other one just with the VGA console.
> 
I played with my PC this week-end.
First I recompiled XFree up to version 4.3.0. It fixed nothing.
I found out that the agpgart/dri drivers failed to init:

Clock  Driver v1.11 Linux agpgart interface v0.100 (c) Dave Jones
[drm:drm_init] *ERROR*  Cannot  initialize  the  agpgart  module.
Uninitialised  timer!   This is just a warning.  Your computer is
OK function=0x00000000, data=0x0 Call Trace:
 [<c0121a80>] check_timer_failed+0x40/0x50
 [<c0121da5>] del_timer+0x15/0x80
 [<c020e318>] mga_takedown+0x68/0x340
 [<c0211d4d>] mga_stub_unregister+0x2d/0x39
 [<c03d8ca7>] drm_init+0xd7/0x260
 [<c03ca6dd>] do_initcalls+0x3d/0x90
 [<c03ca749>] do_basic_setup+0x19/0x20
 [<c010508f>] init+0x2f/0x180
 [<c0105060>] init+0x0/0x180
 [<c0106f9d>] kernel_thread_helper+0x5/0x18
from dmesg

So, I reccompiled my kernels without agpgart.
Then, I noticed that X was ran in 32 bits mode
but that xwininfo said it was in 8 bits mode.
I relaunched X in 16 bits modes and it runs fine, now,
including overlay video.
This is quite strange, because it used to run OK on 2.4 kernels
and I got no idea of the source of the problem.
I now run a VGA console kernel without agpgart with a 16bps X.

So, there seems to be two issues:
- broken matrox fb: I lose the synchro when switching from X to fb.
  fbset configuration is lost when switching consoles.
- X problems maybe related to agpgart bad initialisation for
  some modes.

Please tell me if I should do more tests to find out where the problems
come from.

Regards.

Here's my /proc/pci output, if this can help to find out issues for
agpgart sutffs:
PCI devices found:
  Bus  0, device   0, function  0:
    Host bridge: Advanced Micro Devic AMD-760 [IGD4-1P] Sy (rev 19).
      Master Capable.  Latency=32.  
      Prefetchable 32 bit memory at 0xc0000000 [0xcfffffff].
      Prefetchable 32 bit memory at 0xd6002000 [0xd6002fff].
      I/O at 0xc000 [0xc003].
  Bus  0, device   1, function  0:
    PCI bridge: Advanced Micro Devic AMD-760 [IGD4-1P] AG (rev 0).
      Master Capable.  Latency=32.  Min Gnt=14.
  Bus  0, device   7, function  0:
    ISA bridge: VIA Technologies, In VT82C686 [Apollo Sup (rev 64).
  Bus  0, device   7, function  1:
    IDE interface: VIA Technologies, In VT82C586/B/686A/B PI (rev 6).
      Master Capable.  Latency=32.  
      I/O at 0xc400 [0xc40f].
  Bus  0, device   7, function  2:
    USB Controller: VIA Technologies, In USB (rev 26).
      IRQ 12.
      Master Capable.  Latency=32.  
      I/O at 0xc800 [0xc81f].
  Bus  0, device   7, function  3:
    USB Controller: VIA Technologies, In USB (#2) (rev 26).
      IRQ 12.
      Master Capable.  Latency=32.  
      I/O at 0xcc00 [0xcc1f].
  Bus  0, device   7, function  4:
    SMBus: VIA Technologies, In VT82C686 [Apollo Sup (rev 64).
      IRQ 9.
  Bus  0, device   8, function  0:
    Ethernet controller: VIA Technologies, In VT86C100A [Rhine] (rev 6).
      IRQ 12.
      Master Capable.  Latency=32.  
      I/O at 0xd000 [0xd07f].
      Non-prefetchable 32 bit memory at 0xd6000000 [0xd600007f].
  Bus  0, device   9, function  0:
    Ethernet controller: Realtek Semiconducto RTL-8139/8139C/8139C (rev
16).
      IRQ 10.
      Master Capable.  Latency=32.  Min Gnt=32.Max Lat=64.
      I/O at 0xd400 [0xd4ff].
      Non-prefetchable 32 bit memory at 0xd6001000 [0xd60010ff].
  Bus  0, device  11, function  0:
    Multimedia video controller: Brooktree Corporatio Bt878 Video
Capture (rev 17).
      IRQ 12.
      Master Capable.  Latency=32.  Min Gnt=16.Max Lat=40.
      Prefetchable 32 bit memory at 0xd6003000 [0xd6003fff].
  Bus  0, device  11, function  1:
    Multimedia controller: Brooktree Corporatio Bt878 Audio Capture (rev
17).
      IRQ 12.
      Master Capable.  Latency=32.  Min Gnt=4.Max Lat=255.
      Prefetchable 32 bit memory at 0xd6004000 [0xd6004fff].
  Bus  0, device  13, function  0:
    FireWire (IEEE 1394): VIA Technologies, In IEEE 1394 Host Contr (rev
70).
      IRQ 11.
      Master Capable.  Latency=32.  Max Lat=32.
      Non-prefetchable 32 bit memory at 0xd6005000 [0xd60057ff].
      I/O at 0xd800 [0xd87f].
  Bus  0, device  15, function  0:
    SCSI storage controller: Tekram Technology Co TRM-S1040 (rev 1).
      IRQ 11.
      Master Capable.  Latency=32.  
      I/O at 0xdc00 [0xdcff].
      Non-prefetchable 32 bit memory at 0xd6006000 [0xd6006fff].
  Bus  0, device  17, function  0:
    Multimedia audio controller: Creative Labs SB Live! EMU10k1 (rev 7).
      IRQ 5.
      Master Capable.  Latency=32.  Min Gnt=2.Max Lat=20.
      I/O at 0xe000 [0xe01f].
  Bus  0, device  17, function  1:
    Input device controller: Creative Labs SB Live! MIDI/Game P (rev 7).
      Master Capable.  Latency=32.  
      I/O at 0xe400 [0xe407].
  Bus  1, device   5, function  0:
    VGA compatible controller: Matrox Graphics, Inc MGA G400 AGP (rev
130).
      IRQ 5.
      Master Capable.  Latency=64.  Min Gnt=16.Max Lat=32.
      Prefetchable 32 bit memory at 0xd0000000 [0xd1ffffff].
      Non-prefetchable 32 bit memory at 0xd2000000 [0xd2003fff].
      Non-prefetchable 32 bit memory at 0xd3000000 [0xd37fffff].

Re: 2.6.0-test2 does not boot with matroxfb

[Dave Jones]

On Mon, Aug 11, 2003 at 08:01:06PM +0200, Jocelyn Mayer wrote:
 > I played with my PC this week-end.
 > First I recompiled XFree up to version 4.3.0. It fixed nothing.
 > I found out that the agpgart/dri drivers failed to init:
 > Linux agpgart interface v0.100 (c) Dave Jones

Did you also compile in any of the AGP chipset drivers?
You should see another AGP line following the above message.
If you built them as modules, make sure you put amd-k7-agp or the like
in your /etc/modules to make sure it gets loaded.

Greg, I'm getting quite a few mails which has been people getting
bitten by this. We discussed this briefly at OLS, what's the missing
piece of the puzzle here, hotplug userspace scripts iirc ?

		Dave

-- 
 Dave Jones     http://www.codemonkey.org.uk

Re: 2.6.0-test2 does not boot with matroxfb

[Greg KH]

On Mon, Aug 11, 2003 at 07:07:03PM +0100, Dave Jones wrote:
> On Mon, Aug 11, 2003 at 08:01:06PM +0200, Jocelyn Mayer wrote:
>  > I played with my PC this week-end.
>  > First I recompiled XFree up to version 4.3.0. It fixed nothing.
>  > I found out that the agpgart/dri drivers failed to init:
>  > Linux agpgart interface v0.100 (c) Dave Jones
> 
> Did you also compile in any of the AGP chipset drivers?
> You should see another AGP line following the above message.
> If you built them as modules, make sure you put amd-k7-agp or the like
> in your /etc/modules to make sure it gets loaded.
> 
> Greg, I'm getting quite a few mails which has been people getting
> bitten by this. We discussed this briefly at OLS, what's the missing
> piece of the puzzle here, hotplug userspace scripts iirc ?

Yeah, putting a /sbin/hotplug and the agp modules in initramfs so that
when the pci device is found during boot, the module will be loaded.  I
think you might be able to do this with initrd too.

But in reality, it's a user config error :)

thanks,

greg k-h

Re: 2.6.0-test2 does not boot with matroxfb

[Jocelyn Mayer]

On Mon, 2003-08-11 at 20:14, Greg KH wrote:
> On Mon, Aug 11, 2003 at 07:07:03PM +0100, Dave Jones wrote:
> > On Mon, Aug 11, 2003 at 08:01:06PM +0200, Jocelyn Mayer wrote:
> >  > I played with my PC this week-end.
> >  > First I recompiled XFree up to version 4.3.0. It fixed nothing.
> >  > I found out that the agpgart/dri drivers failed to init:
> >  > Linux agpgart interface v0.100 (c) Dave Jones
> > 
> > Did you also compile in any of the AGP chipset drivers?
> > You should see another AGP line following the above message.
> > If you built them as modules, make sure you put amd-k7-agp or the like
> > in your /etc/modules to make sure it gets loaded.
> > 
Here's the agpgart config I used:
all is in the kernel, nothing in a module....

dules in the kernel, not in modules,
as I used to do in 2.4 kernel, which works well for me
The configuration I tested was (from my .config file):
CONFIG_AGP=y
# CONFIG_AGP_ALI is not set
# CONFIG_AGP_AMD is not set
# CONFIG_AGP_AMD_8151 is not set
# CONFIG_AGP_INTEL is not set
# CONFIG_AGP_NVIDIA is not set
# CONFIG_AGP_SIS is not set
# CONFIG_AGP_SWORKS is not set
CONFIG_AGP_VIA=y
CONFIG_DRM=y
# CONFIG_DRM_TDFX is not set
# CONFIG_DRM_GAMMA is not set
# CONFIG_DRM_R128 is not set
# CONFIG_DRM_RADEON is not set
CONFIG_DRM_MGA=y

> > Greg, I'm getting quite a few mails which has been people getting
> > bitten by this. We discussed this briefly at OLS, what's the missing
> > piece of the puzzle here, hotplug userspace scripts iirc ?
> 
> Yeah, putting a /sbin/hotplug and the agp modules in initramfs so that
> when the pci device is found during boot, the module will be loaded.  I
> think you might be able to do this with initrd too.
> 
> But in reality, it's a user config error :)
> 
I guess not, as I don't use modules (and don't want to) for this feature
andI don't use any initrd (and never want too)...


> thanks,
> 
> greg k-h
What am I missing here ?

Regards.

-- 
Jocelyn Mayer <l_indien@magic.fr>

Re: 2.6.0-test2 does not boot with matroxfb

[Dave Jones]

On Mon, Aug 11, 2003 at 08:55:45PM +0200, Jocelyn Mayer wrote:

 > > > Did you also compile in any of the AGP chipset drivers?
 > # CONFIG_AGP_AMD is not set

That'll be 'no' then 8-)
Fix this, and everything should start magickally working.

		Dave

-- 
 Dave Jones     http://www.codemonkey.org.uk

Re: 2.6.0-test2 does not boot with matroxfb

[Jocelyn Mayer]

On Mon, 2003-08-11 at 20:56, Dave Jones wrote:
> On Mon, Aug 11, 2003 at 08:55:45PM +0200, Jocelyn Mayer wrote:
> 
>  > > > Did you also compile in any of the AGP chipset drivers?
>  > # CONFIG_AGP_AMD is not set
> 
> That'll be 'no' then 8-)
> Fix this, and everything should start magickally working.
> 
> 		Dave
Ooops... So sorry, I've been caught like a newby, this time !
I though that was VIA chipset... I just checked what I used to
do on 2.4 kernels, and I did use CONFIG_AGP_AMD...
So I got no excuse...

Many thanks to you !

-- 
Jocelyn Mayer <l_indien@magic.fr>

Re: 2.6.0-test2 does not boot with matroxfb

[Petr Vandrovec]

On 11 Aug 03 at 21:00, Jocelyn Mayer wrote:
> > On Mon, 2003-08-11 at 20:29, Petr Vandrovec wrote:
> > On 11 Aug 03 at 20:01, Jocelyn Mayer wrote:
> > > I now run a VGA console kernel without agpgart with a 16bps X.
> > > 
> > > So, there seems to be two issues:
> > > - broken matrox fb: I lose the synchro when switching from X to fb.
> > >   fbset configuration is lost when switching consoles.
> > 
> > It is feature, not a bug. fbset does not work on 2.6.x kernels
> > anymore. Apply 
> > ftp://platan.vc.cvut.cz/pub/linux/matrox-latest/matroxfb-2.6.0-t3-c1149.gz
> > if you prefer 2.4.x behavior.
> > 
> In fact, it works untill I switch to another console...
> What is the right way to change the console configuration at run-time,
> now ? Or where should I look for this info ?
> Did it become impossible (would be a severe bug, at least for me...) ?

"stty cols XXX rows YYY" sets size. How you'll set pixclock, color depth
or other features is unknown to me - I use my patch above and others
should ask James Simmons...

> > Also if you are using DRI, even latest XFree mga driver I found still 
> > reprograms hardware even if XFree server is not on a foreground, so you 
> > must use same color depth and vxres for both X and console, and even in 
> > this configuration display corruption on console may occur from time to
> > time...
> >                                             Petr Vandrovec
> >        
> OK, you confirm exactly the symptom I have seen.
> So it seems to be an X driver problem...
> But X is still buggy when it doesn't use DRI,
> at least in 32 bits modes. I know it's quite certain that it's an X bug,
> but I really would like to understand why it works with 2.4 kernels
> and not with 2.6 ones. The X problem is there even if I use the VGA
> console (I mean with a kernel with fb support _not_ activated).

In 2.4.x kernels dri refused to work on couple of systems I have, while
in 2.6.x it is willing to work even on more systems - if we are talking
about console.

I do not know why you have problems with X - 2.6.0-test3-bk-current
works for me with G200, G450 and G550 with Debian's unstable XFree,
(16/16bpp on G550 system and 24/32bpp on others).
                                                Petr
                                                

Re: 2.6.0-test2 does not boot with matroxfb

[Petr Vandrovec]

On 11 Aug 03 at 20:01, Jocelyn Mayer wrote:
> I now run a VGA console kernel without agpgart with a 16bps X.
> 
> So, there seems to be two issues:
> - broken matrox fb: I lose the synchro when switching from X to fb.
>   fbset configuration is lost when switching consoles.

It is feature, not a bug. fbset does not work on 2.6.x kernels
anymore. Apply 
ftp://platan.vc.cvut.cz/pub/linux/matrox-latest/matroxfb-2.6.0-t3-c1149.gz
if you prefer 2.4.x behavior.

Also if you are using DRI, even latest XFree mga driver I found still 
reprograms hardware even if XFree server is not on a foreground, so you 
must use same color depth and vxres for both X and console, and even in 
this configuration display corruption on console may occur from time to
time...
                                            Petr Vandrovec
                                            

Re: 2.6.0-test2 does not boot with matroxfb

[Jocelyn Mayer]

> On Mon, 2003-08-11 at 20:29, Petr Vandrovec wrote:
> On 11 Aug 03 at 20:01, Jocelyn Mayer wrote:
> > I now run a VGA console kernel without agpgart with a 16bps X.
> > 
> > So, there seems to be two issues:
> > - broken matrox fb: I lose the synchro when switching from X to fb.
> >   fbset configuration is lost when switching consoles.
> 
> It is feature, not a bug. fbset does not work on 2.6.x kernels
> anymore. Apply 
> ftp://platan.vc.cvut.cz/pub/linux/matrox-latest/matroxfb-2.6.0-t3-c1149.gz
> if you prefer 2.4.x behavior.
> 
In fact, it works untill I switch to another console...
What is the right way to change the console configuration at run-time,
now ? Or where should I look for this info ?
Did it become impossible (would be a severe bug, at least for me...) ?

> Also if you are using DRI, even latest XFree mga driver I found still 
> reprograms hardware even if XFree server is not on a foreground, so you 
> must use same color depth and vxres for both X and console, and even in 
> this configuration display corruption on console may occur from time to
> time...
>                                             Petr Vandrovec
>        
OK, you confirm exactly the symptom I have seen.
So it seems to be an X driver problem...
But X is still buggy when it doesn't use DRI,
at least in 32 bits modes. I know it's quite certain that it's an X bug,
but I really would like to understand why it works with 2.4 kernels
and not with 2.6 ones. The X problem is there even if I use the VGA
console (I mean with a kernel with fb support _not_ activated).

Thanks for the infos.

Regards.
                                
-- 
Jocelyn Mayer <l_indien@magic.fr>

Re: 2.6.0-test2 does not boot with matroxfb

[Juergen Rose]

Hi Petr,

On Fri, 2003-08-08 at 18:16, Petr Vandrovec wrote: 
> On  8 Aug 03 at 17:45, Juergen Rose wrote:
> > I tried on my PC with a Matrox-G450 several kernel and boot options.
> > Every time when the console should work with matrox framebuffer, linux
> > was crashed. With 2.6.0-test2 and 2.6.0-test2-bk7 I had the following
> > warning performing ''make modules_install''
> > WARNING:
> > /lib/modules/2.6.0-test2[-bk7]/kernel/drivers/video/matrox/matroxfb_crtc2.ko
> > needs unknown symbol matroxfb_enable_irq This WARNING disapears for
> 
> I'm not able to get through Linus's mail filters for past three weeks.
> 
> > 2.4.22-pre6-ac1             |         |               |
> > CONFIG_FB=y                 | VGA=6   |  VGA=775      | vesa:0x31C
> > CONFIG_VIDEO_SELECT=y       |         |               |
> > --+--+--+--
> > 2.6.0-test2                 |         |crashes        |crashes        
> > CONFIG_FRAMEBUFFER_CONSOLE=y|  VGA=6  |after          |after          
> > CONFIG_VIDEO_SELECT=y       |         |BIOS DATA CHECK|BIOS DATA CHECK
> 
> What means this? No useful data here... Oopses? Reboot?
> Endless loop? And did you built vgacon into the kernel? And I'm
> not sure that 2.6.0 can have both vesafb & matroxfb compiled in (well,
> you can compile in both, but I'm not sure what happens).

Sorry, that I did not explain more detailed. "crashes after BIOS DATA
CHECK" means, that "BIOS DATA CHECK" was the last, what I have seen,
then I got a blank screen. For some configurations I still waited
several minutes to see, if it is possible to ping from another PC to the
PC with 2.6.0-test2. But I did not get a connection, so I had to perform
a hard reset with 2.6.0-test2-PC.

> And from your message above it looks like that you are building some 
> vital matroxfb portions as a module. Please build them into the kernel 
> unless you are really sure that you do not want console on them - while 
> 2.4.x's console could use modular fbdevs, 2.6.x's console subsytem 
> cannot do that. 

'grep -i matrox amd4GBaic7xxx_ideHPT_noacpi_noapic_ess1371.cfg' (which
was my .config) gives
CONFIG_FB_MATROX=m
# CONFIG_FB_MATROX_MILLENIUM is not set
# CONFIG_FB_MATROX_MYSTIQUE is not set
CONFIG_FB_MATROX_G450=y
CONFIG_FB_MATROX_G100=y
CONFIG_FB_MATROX_I2C=m
CONFIG_FB_MATROX_MAVEN=m
# CONFIG_FB_MATROX_PROC is not set
CONFIG_FB_MATROX_MULTIHEAD=y

'grep -i "fb.*=[my]" amd4GBaic7xxx_ideHPT_noacpi_noapic_ess1371.cfg'
gives besides MATROS lines:
CONFIG_FB=y
CONFIG_FB_VESA=y
...
CONFIG_FBCON_ADVANCED=y
CONFIG_FBCON_MFB=m
CONFIG_FBCON_CFB2=m
CONFIG_FBCON_CFB4=m
CONFIG_FBCON_CFB8=y
CONFIG_FBCON_CFB16=y
CONFIG_FBCON_CFB24=y
CONFIG_FBCON_CFB32=y
CONFIG_FBCON_VGA=y
CONFIG_FBCON_HGA=m
CONFIG_FBCON_FONTS=y

So I will set CONFIG_FB_MATROX and CONFIG_FB_MATROX_I2C to yes, and
build a new kernel at monday.

> You can also try ftp://platan.vc.cvut.cz/pub/linux/matrox-latest/matroxfb-2.5.72.gz
> (if 2.6.0-test2-mm5 does not complain about unexported matrox_*_irq
> it will probably not apply cleanly to it). It restores fbcon subsystem
> to the 2.4.x's version.
>                                             Petr
>                                             
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/


	Juergen

-- 
Juergen Rose <rose@rz.uni-potsdam.de>
University of Potsdam

Re: 2.6.0-test2 does not boot with matroxfb

[Petr Vandrovec]

On  8 Aug 03 at 17:45, Juergen Rose wrote:
> I tried on my PC with a Matrox-G450 several kernel and boot options.
> Every time when the console should work with matrox framebuffer, linux
> was crashed. With 2.6.0-test2 and 2.6.0-test2-bk7 I had the following
> warning performing ''make modules_install''
> WARNING:
> /lib/modules/2.6.0-test2[-bk7]/kernel/drivers/video/matrox/matroxfb_crtc2.ko
> needs unknown symbol matroxfb_enable_irq This WARNING disapears for

I'm not able to get through Linus's mail filters for past three weeks.

> 2.4.22-pre6-ac1             |         |               |
> CONFIG_FB=y                 | VGA=6   |  VGA=775      | vesa:0x31C
> CONFIG_VIDEO_SELECT=y       |         |               |
> ----------------------------+---------+---------------+---------------
> 2.6.0-test2                 |         |crashes        |crashes        
> CONFIG_FRAMEBUFFER_CONSOLE=y|  VGA=6  |after          |after          
> CONFIG_VIDEO_SELECT=y       |         |BIOS DATA CHECK|BIOS DATA CHECK

What means this? No useful data here... Oopses? Reboot?
Endless loop? And did you built vgacon into the kernel? And I'm
not sure that 2.6.0 can have both vesafb & matroxfb compiled in (well,
you can compile in both, but I'm not sure what happens).

And from your message above it looks like that you are building some 
vital matroxfb portions as a module. Please build them into the kernel 
unless you are really sure that you do not want console on them - while 
2.4.x's console could use modular fbdevs, 2.6.x's console subsytem 
cannot do that. 

You can also try ftp://platan.vc.cvut.cz/pub/linux/matrox-latest/matroxfb-2.5.72.gz
(if 2.6.0-test2-mm5 does not complain about unexported matrox_*_irq
it will probably not apply cleanly to it). It restores fbcon subsystem
to the 2.4.x's version.
                                            Petr
                                            

2.6.0-test2 does not boot with matroxfb

[Juergen Rose]

[-- Attachment #1: Type: text/plain, Size: 2871 bytes --]

I tried on my PC with a Matrox-G450 several kernel and boot options.
Every time when the console should work with matrox framebuffer, linux
was crashed. With 2.6.0-test2 and 2.6.0-test2-bk7 I had the following
warning performing ''make modules_install''
WARNING:
/lib/modules/2.6.0-test2[-bk7]/kernel/drivers/video/matrox/matroxfb_crtc2.ko needs unknown symbol matroxfb_enable_irq
This WARNING disapears for 2.6.0-test2-mm5. But also 2.6.0-test2-mm5
does not work with framebuffer. I tested the following combinations of
parameters:

                            |VGA = 6  |  VGA = 775    |
                            |         |               |video=matrox:
                            |         |               |   vesa:0x31C 
----------------------------+---------+---------------+---------------
2.4.22-pre6-ac1             |         |               |
CONFIG_FB=y                 | VGA=6   |  VGA=775      | vesa:0x31C
CONFIG_VIDEO_SELECT=y       |         |               |
----------------------------+---------+---------------+---------------
2.6.0-test2                 |         |crashes        |crashes        
CONFIG_FRAMEBUFFER_CONSOLE=y|  VGA=6  |after          |after          
CONFIG_VIDEO_SELECT=y       |         |BIOS DATA CHECK|BIOS DATA CHECK
----------------------------+---------+---------------+---------------
2.6.0-test2                 |         |               | 
CONFIG_FRAMEBUFFER_CONSOLE=y| 640x480 |  640x480      |  640x480
CONFIG_VIDEO_SELECT=n       |         |               |  
----------------------------+---------+---------------+---------------
2.6.0-test2                 |         |crashes        |crashes
without console-fb          |  VGA=6  |after          |after
CONFIG_VIDEO_SELECT=y       |         |BIOS DATA CHECK|BIOS DATA CHECK
----------------------------+---------+---------------+---------------
2.6.0-test2                 |         |               |
without console-fb          | 640x480 |  640x480      |  640x480
CONFIG_VIDEO_SELECT=n       |         |               |
----------------------------+---------+---------------+---------------
2.6.0-test2-bk7             |         |crashes        |crashes
with console-fb             | VGA=6   |after          |after
CONFIG_VIDEO_SELECT=y       |         |BIOS DATA CHECK|BIOS DATA CHECK
----------------------------+---------+---------------+---------------
2.6.0-test2-mm5             |         |crashes        |crashes
CONFIG_FRAMEBUFFER_CONSOLE=y| VGA=6   |after          |after
CONFIG_VIDEO_SELECT=y       |         |BIOS DATA CHECK|BIOS DATA CHECK
----------------------------+---------+---------------+---------------

Because I am afraid, that the lines of the table are wrapped by the mail
client, the is alsa attached to this mail. Any hint would very much
appreciated.

	Regards
		Juergen

-- 
Juergen Rose <rose@rz.uni-potsdam.de>
Max-Delbrueck-Zentrum

[-- Attachment #2: 2.6.0-test2_problems_table.txt --]
[-- Type: text/plain, Size: 2143 bytes --]

Results of some kernel and boot parameters:

                            |VGA = 6  |  VGA = 775    |
                            |         |               |video=matrox:
                            |         |               |   vesa:0x31C 
----------------------------+---------+---------------+---------------
2.4.22-pre6-ac1             |         |               |
CONFIG_FB=y                 | VGA=6   |  VGA=775      | vesa:0x31C
CONFIG_VIDEO_SELECT=y       |         |               |
----------------------------+---------+---------------+---------------
2.6.0-test2                 |         |crashes        |crashes        
CONFIG_FRAMEBUFFER_CONSOLE=y|  VGA=6  |after          |after          
CONFIG_VIDEO_SELECT=y       |         |BIOS DATA CHECK|BIOS DATA CHECK
----------------------------+---------+---------------+---------------
2.6.0-test2                 |         |               | 
CONFIG_FRAMEBUFFER_CONSOLE=y| 640x480 |  640x480      |  640x480
CONFIG_VIDEO_SELECT=n       |         |               |  
----------------------------+---------+---------------+---------------
2.6.0-test2                 |         |crashes        |crashes
without console-fb          |  VGA=6  |after          |after
CONFIG_VIDEO_SELECT=y       |         |BIOS DATA CHECK|BIOS DATA CHECK
----------------------------+---------+---------------+---------------
2.6.0-test2                 |         |               |
without console-fb          | 640x480 |  640x480      |  640x480
CONFIG_VIDEO_SELECT=n       |         |               |
----------------------------+---------+---------------+---------------
2.6.0-test2-bk7             |         |crashes        |crashes
with console-fb             | VGA=6   |after          |after
CONFIG_VIDEO_SELECT=y       |         |BIOS DATA CHECK|BIOS DATA CHECK
----------------------------+---------+---------------+---------------
2.6.0-test2-mm5             |         |crashes        |crashes
CONFIG_FRAMEBUFFER_CONSOLE=y| VGA=6   |after          |after
CONFIG_VIDEO_SELECT=y       |         |BIOS DATA CHECK|BIOS DATA CHECK
----------------------------+---------+---------------+---------------