[2.5.72-mm1] Matroxfb behaves buggy

[2.5.72-mm1] Matroxfb behaves buggy

[Kimmo Sundqvist]

Relevant .config

# CONFIG_FB_VGA16 is not set
# CONFIG_FB_VESA is not set
CONFIG_VIDEO_SELECT=y
CONFIG_FB_MATROX=y
# 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_MULTIHEAD is not set
CONFIG_VGA_CONSOLE=y
# CONFIG_MDA_CONSOLE is not set
CONFIG_DUMMY_CONSOLE=y
CONFIG_FRAMEBUFFER_CONSOLE=m
CONFIG_PCI_CONSOLE=y
# CONFIG_FONTS is not set
CONFIG_FONT_8x8=y
CONFIG_FONT_8x16=y

I used the same append="video:matrox..." line as with Debian 2.4.20 kernel, 
which defines 1280x1024@60Hz for my TFT

append="elevator=as video=matrox:vesa:0x31B,pixclock:8028,    \
left:32,right:32,upper:21,lower:21,hslen:432,vslen:10"

I tried both MATROX_G450 and MATROX_G100 drivers, first G450 and then G100.  I 
find it odd that both are configured, since selecting G450 hides G100 in 
"make menuconfig", so I have all the time assumed it is disabled when it's 
hidden.  Might the problem be caused by including both?  I'll go and see, but 
I think it's not the only problem.

What's the difference between them?  .config help ought to tell this briefly.

When 2.4.20 switches to 1280x1024, the screen goes black with 2.5.72, then 
some seconds pass and 1. vertical dark gray lines appear, and 2. some dark 
blue areas between them.  Some seconds pass and X comes up as if nothing was 
wrong.

I logged in to the console as root, not seeing anything.  The TFT display 
reported 640x480@60Hz in it's menu.  I typed "fbset 1280x1024" for my custom 
1280x1024@60Hz mode.  Oddly, X display appeared at 1280x1024 resolution, mode 
parameters being something between the fbset and X modes, but I could do 
nothing, so I pushed the reset button.

syslog had a screenful of "^@^@^@^@^@^@^@^@^@^@^@" between time before the 
lockup and next boot, as did kern.log

Additionally, syslog had another screenful of 
"<9C><B5><EC><FF>^N^NNz<FC><FF>^S<C4>^?<A9><C7><FF><AB>  \
<B1><82><DE>^_\<CB>W<C0><CE><CE><DE>      \
<CE><F5><93>!<FF>^ZT<FC>fI^?td<F4>"

  Bus  1, device   0, function  0:
    VGA compatible controller: Matrox Graphics, Inc MGA G400 AGP (rev 4).
      IRQ 16.
      Master Capable.  Latency=16.  Min Gnt=16.Max Lat=32.
      Prefetchable 32 bit memory at 0xd2000000 [0xd3ffffff].
      Non-prefetchable 32 bit memory at 0xd4000000 [0xd4003fff].
      Non-prefetchable 32 bit memory at 0xd5000000 [0xd57fffff].

SMP and pre-empt enabled.

-Kimmo Sundqvist

Re: [2.5.72-mm1] Matroxfb behaves buggy

[Petr Vandrovec]

On 19 Jun 03 at 13:49, Kimmo Sundqvist wrote:
> I tried both MATROX_G450 and MATROX_G100 drivers, first G450 and then G100.  I 
> find it odd that both are configured, since selecting G450 hides G100 in 
> "make menuconfig", so I have all the time assumed it is disabled when it's 
> hidden.  Might the problem be caused by including both?  I'll go and see, but 
> I think it's not the only problem.
> 
> What's the difference between them?  .config help ought to tell this briefly.

As far as I can tell, it is as clear as it could be... First option
is named 'G100/G200/G400/G450/G550' support, while second is named
'G100/G200/G400'.

> When 2.4.20 switches to 1280x1024, the screen goes black with 2.5.72, then 
> some seconds pass and 1. vertical dark gray lines appear, and 2. some dark 
> blue areas between them.  Some seconds pass and X comes up as if nothing was 
> wrong.
> 
> I logged in to the console as root, not seeing anything.  The TFT display 
> reported 640x480@60Hz in it's menu.  I typed "fbset 1280x1024" for my custom 
> 1280x1024@60Hz mode.  Oddly, X display appeared at 1280x1024 resolution, mode 
> parameters being something between the fbset and X modes, but I could do 
> nothing, so I pushed the reset button.

fbset is not supported on 2.5.x kernels for setting console resolution.
If you want same environment you had in 2.4.x, get patch from
ftp://platan.vc.cvut.cz/pub/linux/matrox-latest.
 
> syslog had a screenful of "^@^@^@^@^@^@^@^@^@^@^@" between time before the 
> lockup and next boot, as did kern.log

No surprise... But I doubt that it is matroxfb's problem.
                                                        Petr