Created attachment 104037 [details]
just before X starts

Created attachment 104038 [details]
strange X display

Just boot with resolution 800x600, it works for me.

Hardware info:
Intel D815EEA, 352MB RAM.

I experienced this error when trying the install in graphical mode
from the CDROMs for FC3T2. The screenshot shows the distortion that
happens when trying the installation.
To exit the installer when this error happens, pressing alt-sysreq-b
allows an immediate break from the installer.

I didn't try passing 800 x 600 resolution to the installer before
attempting the GUI installation. If the need for designating the
specific resolution at boot is needed, then adding this behavior to
the installer would probably be a wise choice. not many people
attempting installations would think to specify a screen resolution
before atempting an installation.

The effected graphics version is:
VGA compatible controller: Intel Corp. 82815 CGC [Chipset Graphics
Controller] (rev 02) 

I've updated my system to current development tree, after that this
distortion back again causes console lockup. However, I can login
remotely and do all my things as normal.

Here is interesting part of Xorg log:

(II) I810(0): [drm] DRM interface version 1.2
(II) I810(0): [drm] created "i810" driver at busid "pci:0000:00:02.0"
(II) I810(0): [drm] added 8192 byte SAREA at 0x1897e000
(II) I810(0): [drm] mapped SAREA 0x1897e000 to 0xf6f93000
(II) I810(0): [drm] framebuffer handle = 0xf8000000
(II) I810(0): [drm] added 1 reserved context for kernel
(II) I810(0): [drm] Registers = 0xffa80000
(II) I810(0): [agp] dcacheHandle : 0x0
(II) I810(0): [agp] GART: no dcache memory found
(II) I810(0): [agp] Bound backbuffer memory
(II) I810(0): [agp] Bound depthbuffer memory
(II) I810(0): [agp] Bound System Texture Memory
(II) I810(0): [agp] GART: Allocated 4K for mouse cursor image
(II) I810(0): [agp] GART: Allocated 16K for ARGB mouse cursor image
(II) I810(0): Adding 768 scanlines for pixmap caching
(II) I810(0): Allocated Scratch Memory
(II) I810(0): [dri] Buffer map : 3fb000
(II) I810(0): [drm] added 256 4096 byte DMA buffers
(II) I810(0): [drm] Init v1.4 interface.
(II) I810(0): [drm] dma control initialized, using IRQ -1007
(II) I810(0): [dri] visual configs initialized.
(==) I810(0): Write-combining range (0xf8000000,0x4000000)
(II) I810(0): vgaHWGetIOBase: hwp->IOBase is 0x03d0, hwp->PIOOffset is
0x0000
pgetbl_ctl: 0x15010001 pgetbl_err: 0x0
ipeir: 0 iphdr: cc9d8b00
LP ring tail: 48 head: 8 len: f001 start 2300000
eir: 0 esr: 1 emr: 3d
instdone: ff7a instpm: 0
memmode: 4 instps: 10
hwstam: 9ac7 ier: 0 imr: 9ac7 iir: 0

Fatal server error:
Active ring not flushed

Please try a text mode install, so you can get to a system prompt
after installation, and then once you've rebooted into the new
OS install, log in as root and run "system-config-display --reconfig".

Now run "startx" and report back what the results are.  Once
you've tested this, please attach the X server log, config file,
and /var/log/messages to the report as individual uncompressed
file attachments using the link below and we will review the
information and try to troubleshoot the issue further.

After you've tested that, please try disabling Red Hat Graphical
Boot (rhgb) by modifying your grub.conf or modifying the kernel
commandline inside grub.  Then reboot the system and let it
start up without rhgb.  When rhgb does not start, does the problem
go away at all?

Setting bug status to "NEEDINFO"

- I installed FC3t3 with the textmode installer
- Everything seems to work (rhgb, firstboot, X) with 800x600 resolution
- Selected 1024x768 screen resolution
- restarted X, everything worked
- updated system to rawhide, reboot
- X (at gdm time) failed to start, rhgb worked though
- system-config-display --reconfig
- selected tft panel 1024x768
- chose resolution 1024x768
- reboot
- disabled rhgb @ boot
- working @ 1024x768 with i810 driver
- I'll try using it for a time and see what happens

Attached dmesg, xorg.conf, Xorg.0.org (there is a drm warning at the
end of dmesg...)

Created attachment 105116 [details]
dmesg

Created attachment 105117 [details]
Xorg logfile

Created attachment 105118 [details]
Xorg config file

There are some strange messages shown by dmesg:

[drm:i810_wait_ring] *ERROR* space: 65520 wanted 65528
[drm:i810_wait_ring] *ERROR* lockup
[drm:i810_wait_ring] *ERROR* space: 65512 wanted 65528
[drm:i810_wait_ring] *ERROR* lockup
[drm:i810_wait_ring] *ERROR* space: 65504 wanted 65528
[drm:i810_wait_ring] *ERROR* lockup
[drm:i810_wait_ring] *ERROR* space: 65496 wanted 65528
[drm:i810_wait_ring] *ERROR* lockup
mtrr: base(0xf8000000) is not aligned on a size(0x180000) boundary
[drm] Using v1.4 init.

glxinfo shows DRI enabled, glxgears works (200 fps @ celeron 850)

> - X (at gdm time) failed to start, rhgb worked though
> - disabled rhgb @ boot
> - working @ 1024x768 with i810 driver

This sounds to me like a problem we're observing.  In previous
OS releases, we started rhgb in it's own X server, which then
quit, and started the real X server.  Now, it attempts to start
a second X server while the first is still running, in order
to speed up boot time.  This causes 2 problems:

1) The first X server (the rhgb one) gets DRI support, and the
   second and subsequent X servers get disabled DRI because DRI
   only works on one X server (limitation).

2) Many video drivers were never designed, implemented, and
   rigorously tested to work in systems with multiple X servers
   running.

As a result of #2, we are hitting bugs and limitations in the
X server and video drivers that would not ordinarily surface
for normal users running one X server.

For comment #14, that may or may not be related to this problem,
as that is a common error that occurs in the Intel driver which
seems to have eluded X.Org and XFree86 i810 developers for 3 years
or so now.  ;o)  Does the error go away if you disable rhgb and
reboot, or does it occur always?

Oh, forgot to mention..  We're currently evaluating the ramifications
of starting 2 X servers simultaneously to speed up booting and wether
it is worth the risk of instabilities in some drivers with some users.

We'll hopefully have a solution to this for FC3.  I'll provide status
updates as we progress.

Thanks.

With rhgb disabled the "warnings" disappear. 

It's strange I can't install FC with the graphical installer (fc3t3
installer turns the screen green at about the root password dialog. No
rainbow effect this time) but use X with the i810 driver after the
install. Something changed in the kernel / Xorg packages recently? 

Which warnings disappear specifically?

Woops, I meant the errors reported by dmesg:

[drm:i810_wait_ring] *ERROR* space: 65520 wanted 65528
[drm:i810_wait_ring] *ERROR* lockup
[drm:i810_wait_ring] *ERROR* space: 65512 wanted 65528
[drm:i810_wait_ring] *ERROR* lockup
[drm:i810_wait_ring] *ERROR* space: 65504 wanted 65528
[drm:i810_wait_ring] *ERROR* lockup
[drm:i810_wait_ring] *ERROR* space: 65496 wanted 65528
[drm:i810_wait_ring] *ERROR* lockup

In the meantime I've installed FC3T3 also in text mode, as graphical
installer lockup with black screen. I'd like to mention that adding
Option âNoAccelâ to xorg.conf solves the issue as well.

I had the same problem with an i815 (motherboard GigaByte GA-6OMM7E)
and the final FC3. NoAccel solved it for me as well. Thanks for that hint.

Also, I had to install using the text mode and have turned of rhgb.

*** This bug has been marked as a duplicate of 132267 ***

I can confirm this bug on an i810 video with Fedora Core 3, trying to
upgrade from Fedora Core 2. I'll try to install in text mode tonight
and add the "NoAccel" option to x.org and report it back.

Installing FC3 in text mode and adding "NoAccel" to the i815 device
section of /etc/X11/xorg.conf does indeed fix the problem on FC3.

You really should see bug #132267

All I needed was the "NoAccel" adjustment to fix a crazy X interface
after upgrade from FC2 to FC3. e.g. Windows would open but not
refresh. Text would not scroll properly. Only some menus would drop down.
Thank you Marek Kassur!
Perhaps NoAccel should be default for this driver?

Bug is back in Rawhide 2006-01-30. I tried to install Rawhide on my laptop and
at the moment anaconda formats the harddisk the screen is in the funny rainbow
state as described by this bug.

Changed to 'CLOSED' state since 'RESOLVED' has been deprecated.