[Edgy] 5 white blinking stripes when booting on i810 laptop

sorry for assigning to xorg... I guess this happens due to frame buffer?

Please also consider that black background boot splash suggestion.
Should I file another bug or something?

On 10/9/06, towsonu2003 <email address hidden> wrote:
>
> sorry for assigning to xorg... I guess this happens due to frame buffer?
>
> ** Changed in: Ubuntu
> Sourcepackagename: None => xorg
>
> ** Changed in: xorg (Ubuntu)
> Sourcepackagename: xorg => fbset
>
> --
> [Edgy] 5 white blinking stripes when booting on i810 laptop
> https://launchpad.net/bugs/64666
>

--
Sincerely yours
Huang Bo

> Should I file another bug
probably yes, to the usplash package I believe.

thanks for the report :)

I'm seeing identical problems.

Here its affecting two Sony Vaios: PCG-SRX41 and PCG-SRX51 (only difference is CPU clock and RAM) which are equivalent to the SRX99.

They have the Intel 82815 CGC rev 11 chipset with a native 1024x768 panel.

This is the default Edgy 6.10 (2.6.17-10) kernel install.

The GRUB splash image loads without a problem. The issue seems to be when the kernel is loading there is no graphical framebuffer.

Display is limited to a central rectangle of the screen and when the usplash screen displays all we see is vertical bands of white blocks that flicker (obviously as the splash screen progress indicator moves along the interpretation of the video memory in text mode causes the flicker).

When booting no graphical framebuffer appears to have been loaded. Neither vesafb nor i810fb load even when the GRUB boot command-line is edited to include:

video=vesfab:
video=i810fb:

(I've tried many different versions of those lines for differing resolutions, etc., as per the vesafb.txt and intel810.txt documenation).

The notebook will only offer TEXT modes.

Any use of "vga=xxx" where xxx represents a vaild graphics setting results in "You passed an undefined mode number" and offering of the text modes 0xF00-0xF07 (80x25 to 80x60)

From what I've read it appears that the initrd should contain the FB drivers because it is built with the "most" option by default.

After Gnome has started, running "usplash -c" from an xterm brings up the splash screen perfectly on tty7.

Is there a way to log or check what is going in with initrd ?

According to this Debian bug-report

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=398948

the issue affects the 2.6.17 and some later kernel builds.

Booting with the default kernel 2.6.17 and checking loaded modules suggests the kernel has correctly loaded i810fb:

$ lsmod | grep fb
i810fb 33852 0
vgastate 10240 1 i810fb
agpgart 34888 4 drm,i810fb,intel_agp
fbcon 41504 0
tileblit 3840 1 fbcon
font 9344 1 fbcon
bitblit 7168 1 fbcon
vesafb 9244 0

So my question now is, is i810fb loaded from initrd or later?

I'm trying to detect whether it is actually available to use the GRUB kernel command-line options - if it isn't then it makes sense that the splash screen fails AND that any attempt to use "vga=791" results in the "...undefined mode number" error.

I found comments in a Linux Kernel mailing list message talking about changing the way the framebuffer code initialises.

http://lkml.org/lkml/2004/9/3/429

This led me to examine the source for 2.6.17-10 to see if I could see anything that would explain the behaviour.

In drivers/video/i810/i810_main.c the call to fb_get_options() (in drivers/video/fbmem.c) only occurs if the driver is statically linked into the kernel, not if its a module:

#ifndef MODULE
static int __devinit i810fb_init(void)
{
 char *option = NULL;

 if (fb_get_options("i810fb", &option))
  return -ENODEV;
 i810fb_setup(option);

 return pci_register_driver(&i810fb_driver);
}
#endif

This means that the function that is responsible for reading the boot-time parameters, i810fb_setup(), isn't available at boot-time.

This would seem to explain why the boot-time options aren't having any effect.

It appears that i810fb has to be built into the kernel to be used in conjunction with usplash?

Can someone with knowledge confirm this is the explanation?

Building the i810fb driver and its dependencies (agpgart, drm, intel_agp) into the kernel fixed this issue on these notebooks, but created another one:

Now, xorg can't load it's i810 driver (they argue over which one has control, I guess).

So it seems that the kernel driver isn't reading the kernel command line when loaded as a module.

See my posts on this in the similar debian bug for more info (towards the end of the bug report):

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=398948

Pay particular attention to the entry timed "Sat, 20 Jan 2007 21:10:37 +0000" that refers to how drivers should implement kernel command-line parsing in module and static-link variations.

I've reported the knock-on bug that kills X if kernel i810fb driver is loaded at:

https://bugs.launchpad.net/ubuntu/+source/xorg/+bug/80792

Thank you for taking the time to report this bug and helping to make Ubuntu better. You reported this bug a while ago and there hasn't been any activity in it recently. We were wondering is this still an issue for you? Can you try with latest Ubuntu release? Thanks in advance.

We are closing this bug report because it lacks the information we need to investigate the problem, as described in the previous comments. Please reopen it if you can give us the missing information, and don't hesitate to submit bug reports in the future. To reopen the bug report you can click on the current status, under the Status column, and change the Status back to "New". Thanks again!