Bug 168618

Created attachment 79571 [details]
YaST log directory (tar.bz2)

Created attachment 79572 [details]
screenshot of SaX2 during upgrade

sorry for the photo, but I couldn't take a "real" screenshot in any way and this is better than any description. This looks just like manually started "SaX2 -r" during a clean install, except in the later case the visible fonts were bigger.

Created attachment 79573 [details]
screenshot of garbled KDE desktop

Same goes for this attachment - this is how the default KDE  desktop looked the first time X started successfully. (It didn't directly after installation - blank screen, frozen computer, even Shift-Lock didn't work any more).

This is what X11 reports upon startup (just starting "Xorg") - I caught this using a remote ssh session to start up X11, because otherwise you don't see it.

X Window System Version 6.9.0
Release Date: 21 December 2005
X Protocol Version 11, Revision 0, Release 6.9
Build Operating System: SuSE Linux [ELF] SuSE
Current Operating System: Linux schleppmich 2.6.16-28-default #1 Wed Apr 19 15:38:34 UTC 2006 i686
Build Date: 18 April 2006
        Before reporting problems, check http://wiki.X.Org
        to make sure that you have the latest version.
Module Loader present
Markers: (--) probed, (**) from config file, (==) default setting,
        (++) from command line, (!!) notice, (II) informational,
        (WW) warning, (EE) error, (NI) not implemented, (??) unknown.
(==) Log file: "/var/log/Xorg.0.log", Time: Mon Apr 24 00:20:37 2006
(==) Using config file: "/etc/X11/xorg.conf"
(WW) ****INVALID MEM ALLOCATION**** b: 0xffe80000 e: 0xffeffff0 correcting
The XKEYBOARD keymap compiler (xkbcomp) reports:
> Warning:          Type "ONE_LEVEL" has 1 levels, but <RALT> has 2 symbols
>                   Ignoring extra symbols
Errors from xkbcomp are not fatal to the X server
Synaptics DeviceInit called
SynapticsCtrl called.
Synaptics DeviceOn called
Could not init font path element /usr/X11R6/lib/X11/fonts/local, removing from list!
Could not init font path element /usr/X11R6/lib/X11/fonts/CID, removing from list!
ADVANCE_LP_RING: outring (0x27) isn't on a QWord boundaryADVANCE_LP_RING: outring (0x2f) isn't on a QWord boundaryADVANCE_LP_RING: outring (0x37) isn't on a QWord boundaryADVANCE_LP_RING: outring (0x3f) isn't on a QWord boundaryADVANCE_LP_RING: outring (0x47) isn't on a QWord boundaryADVANCE_LP_RING: outring (0x4f) isn't on a QWord boundarySynaptics DeviceOff called
ADVANCE_LP_RING: outring (0x7) isn't on a QWord boundary

Additional information:

Ubuntu seems to have this problem to - same display corruption with i810 driver:
http://ubuntuforums.org/showthread.php?t=154774

They've also opened a bug report and declared it as a showstopper:
https://launchpad.net/distros/ubuntu/+source/xserver-xorg-driver-i810/+bug/29880

The upstream bugreport is probably the more interesting link...
https://bugs.freedesktop.org/show_bug.cgi?id=5443

CC'ing Egbert Eich as he is CCd there also ;-)

Created attachment 79610 [details]
Xorg.0.log

That is true, however I'm not getting the signal 11 errors the other people are getting. I'm just getting a non-working X server with these errors at the bottom of the Xorg log file:

ADVANCE_LP_RING: outring (0x27) isn't on a QWord boundary(WW) I810(0): Fixing display offsets.
ADVANCE_LP_RING: outring (0x2f) isn't on a QWord boundary(WW) I810(0): Fixing display offsets.

I'm also getting this:
(WW) ****INVALID MEM ALLOCATION**** b: 0xffe80000 e: 0xffeffff0 correcting^G
(WW) I810(0): Bad V_BIOS checksum

(WW) I810(0): Disabling XVideo output because Pipe B is in double-wide mode.
(II) I810(0): direct rendering: Failed

I'm reassigning this to Stefan. Maby this should reported/fixed upstream. Stefan: Please decide what should be done.

One idea: If the old driver worked, why don't we just keep it for 10.1 until there is a fix?

It works with

Option "noaccel"
Option "SWcursor"

Correct?

Created attachment 79678 [details]
Xorg logfile with "SWcursor" and "noaccel" options

Yes. It also works withough "SWcursor" but then the mouse pointer is invisible. :)

Could you send me the output of "hwinfo --gfxcard"? Thanks.

Created attachment 79680 [details]
hwinfo --gfxcard output

Here you are.

Addendum to Comment 11:

With "noaccel", moving windows and screen updates in general are unbearably slow. You can actually see the window being repainted. I didn't know there was a frame rate for scrolling a web site. ;)

Vendor: pci 0x8086 "Intel Corporation"
  Device: pci 0x3582 "855 GM"
  SubVendor: pci 0x1584 "Uniwill Computer Corp"
  SubDevice: pci 0x9500 

For 10.1 you'll need to live with Noaccel/SWcursor. I'm sorry.

That isn't good. This is a very common notebook chipset.
Where exactly is the problem? Is there anything I can do to help fix it?

Wouldn't it be possible to provide the 10.0 driver for 10.1, as a workaround? The 10.0 driver worked perfectly. You are doing the same thing for ATI Radeon users with "radeonold", after all.

Jens

fixed for RC3. We cannot do more at this point. Beta tests begin with Beta1 and not with RC2 ...

I understand that, I was asking for a starting point to try and fix this for myself. Maybe I can get the old driver compiled for 10.1 and use it, even if it's not integrated in the distribution.

Of course, in case of success I'll report here. I just need a few hints as to what would need to be done.

Thanks!

What you can try it the latest Intel driver from X.Org CVS/git. But building is not a trivial task.

(In reply to comment #14)
> For 10.1 you'll need to live with Noaccel/SWcursor. I'm sorry.

(In reply to comment #16)
> fixed for RC3. We cannot do more at this point. 
> Beta tests begin with Beta1 and not with RC2 ...

Does this mean that RC3 will have this bug fixed really or does it mean that you have applied the noaccel/swcursor hack to RC3? 

If the latter, that's really sad. Every release has an almost-showstopper bug -- with 10.0 it was bug #133651, with 10.1 it's going to be this? I think if the bug is really serious you should include the bugfix even with "feature freeze". (Bugfixes are not new features.)

That said, Fedora Core 5 with X 7.0 works on my machine, so SUSE 10.1 with X 6.9 should work, right?

> Does this mean that RC3 will have this bug fixed really or does it mean that
> you have applied the noaccel/swcursor hack to RC3?
The latter.

As one of the affected by this problem, I have to say that this is most annoying. It means that majority of today's notebooks on the market won't install SuSE RC3. Will a real fix be added in the final release? If not, why do you release RC's if bugs do not get fixed properly?

As an importand side note I'd also like to point out that the issue, as it seems, has already been resolved in Xorg - see
https://bugs.freedesktop.org/show_bug.cgi?id=5443
and especially
https://bugs.freedesktop.org/show_bug.cgi?id=6750

>As one of the affected by this problem, I have to say that this is most
>annoying. It means that majority of today's notebooks on the market 
I see exactly one bugreport about this problem on 855. We are using a bunch of 855 notebooks inhouse, and none of them shows this problem.

> won't install SuSE RC3. Will a real fix be added in the final release?
Unfortunately not. :-(

> If not, why do you release RC's if bugs do not get fixed properly?
I would like to invite you to participate in the next beta test program with Beta1. Then we would have had a chance to fix such issues in time ...

> As an importand side note I'd also like to point out that the issue, as it
> seems, has already been resolved in Xorg - see
> https://bugs.freedesktop.org/show_bug.cgi?id=5443
> and especially
> https://bugs.freedesktop.org/show_bug.cgi?id=6750
Unfortunately this is a different issue. :-( The problem is "outring (0x27) isn't on a QWord boundary" and not "****INVALID MEM ALLOCATION****".

What notebooks using i855GM do you have? Do they have the same IDs as reported by lspci and hwinfo? If you search Google you'll find a number of people using i810 having tried the beta and failing as soon as X starts.

Maybe the bug reporting process is not easy/streamlined enough for users to care to actually send bug reports. Maybe we should do something like have an automated feedback process, that automatically collects information from the user (which application crashed, what was done before that), and from the system (like hwinfo, /var/log/YaST2, etc) and puts this into a statistical database so SuSE developers know which bugs _actually_ happen most often, and not just which are most often reported.

btw: I have the "QWord boundary" problems too. Starting "Xorg" on the console also keeps printing "(whatever) isn't on a QWord boundary". It should also be logged in the Xorg.0.log I posted here.

I tried rebuilding Xorg (the RPM) from source with the patches applied but it seems I can't get RPM to work properly (it keeps re-extracting the sources after I patched them, when I tell it to build a binary package). I've asked for assitance on #rpm and #suse and in the Xorg developer's mailing list. We'll see.

Anyway, I'm back to 10.0. In this state, 10.1 is unuseable for me: no working X driver, suspend to disk and RAM doesn't work any more (notebook just wakes up again), beagled-helper background task constantly eats up about 25-50% CPU load and doesn't let my CPU scale down - or my harddisk sleep - even when running on battery, installer starts mdadm by default (I don't have RAID on my notebook!) which then throws errors at me when booting, the installer also detected my DVD-RW drive as an IDE-tape device and then keeps modprobe'ing ide_tape which promptly produces a kernel Oops on every boot, etc ...

I reported most of these as bugs already. I'd like to help fixing them (and I keep the 10.1 partition on my harddisk, although the infamous ReiserFS resizing bug in 10.0 almost destroyed it when I reinstalled) but if the resolution is "wait for 10.2" anyway, then I'm going to do exactly that.

But please don't consider this bug "fixed". It's not. It's "LATER" or "WONTFIX", but not "FIXED". A workaround that just disables anything that doesn't work is not a fix.

Jens


PS: BTW, the DRI (dri.sf.net) driver for 6.9 does not work either. Same symptoms. I don't know if it's the same one, but the Xorg bug report says it's an X server issue and not a driver issue anyway, so that might be the problem.

LATER.

This bug is an absolute showstopper for me. How can I find out whether my laptop would be one of those affected?

I just tried to install SUSE 10.1 on a Stamp brand model 223 (same as the LinuxCertifed LC2100).  The chipset is an Intel 855GM.  I could not get any usable video.  sax2 would not work either.  After starting sax2, only some of the window borders showed, the background was black and there was no cursor.  Any key presses caused sax2 to immediately crash.  CTRL-ALT-Backspace would not kill the broken server.  The only way out of sax2 was CTRL-ALT-Delete to reboot the machine.

I verified that the 855 resolution package was installed and functioning correctly providing the 1280x800 mode I required.  I tried many different modes but I was unable to get the i810 X server to work in *any* mode including basic ones such as 800x600.

For those that want to get a very basic but usable X session going, back up your xorg.conf file and then copy /etc/X11/xorg.conf.install to /etc/X11/xorg.conf.

As it is, I will be forced to go back to 10.0 or try another distro.  I would strongly recommend that users *do not* upgrade to 10.1 if they have a similar chipset.  Perhaps this should be stated more clearly in the release notes.  In fact, it should probably be in the form of a strong and clear warning.

Created attachment 85547 [details]
my SuSE 10.1 login screen

I`am just install too SuSE 10.1 on my office PC "Prime80" (CPU - Intel Celeron D 331 2,6 GHz, RAM - 512 MB PC3200 DDR SDRAM, motherboard - Intel D915GAG with integrated video i915G. After finish installation I got garbled screen, as you can see in attach. Options "SWcursor" and "noaccel" give no result. 
Only when change driver to "vesa", I got usable screen.

Created attachment 85550 [details]
This is virtual console screen after installation

reopen

*** This bug has been marked as a duplicate of 179773 ***

I recently installed Suse 10.1 (Final) on my Averatec 4265-EH1 (Intel Celeron w/ i915GM graphics) and am experiencing the exact same bug.  I have also attempted multiple fixes with different drivers, resolutions, refresh rates, etc, and have been unable to fix totally.  The only workaround I have as of now is to use the vesa driver (slow, no acceleration).  For other looking here, you can apply this fix very easily.  I've used 1 of 2 different methods.

1:  run sax2 in vesa compliant mode.
# sax2-vesa

This will force the use of the vesa driver.  Sax will open properly, and will configure your xorg.conf file to use the vesa driver.

2:  manually enter "vesa" into xorg.conf file.
-Boot into failsafe mode at bootup.
-login as root
-edit /etc/X11/xorg.conf

scroll to:
Section "Device"
  BoardName    "915 GM"
  BusID        "0:2:0"
  Driver       "i810"  
  Identifier   "Device[0]"
  VendorName   "Uniwill Computer Corp"
EndSection

-Chance "i810" to "vesa"
-save
-startx and XWindows should come up.  It will look decent, but perform slowly.

A fix is still needed.

Also, I can't access the "duplicate bug."  Anyone else???  Would like to know what's going on with this.

I have bought a box version of Suse 10.1. I have used Suse 10.0 before that. Guess if  I was suprised when it was not possible to do the HW update without problems in the end of the installation. Sax2 hangs up in an inverted kind of mode. The timer that should grant return to the previous value also hangs. The main switch was the only way to come out. So here I am, I cannot install  a Swedish keyboard which is a big problem for me in Sweden.  Sax2 seems to have been i peace of shit. It works fine in 10.0.  Why?  I have an Intel Pentium4, 2.8 GHz with intergrated videochipset on the mothercard, Intel 865G. The screen is a Sony 19 inch TFT SDM-HS 95 P.
Ive also tried to run in the Sax2-vesa mode but the update frequenses get to high for the screen so it does not work either.  The frequenses get much higher than the menu tells. When I choose e.g 1280x1024, 75 Hz I get 98 Hz??
Is someone working at it or must I change to another distro???

I have a Fujitsu Siemens Amilo 1450G. It works well with Suse 10.0 but with 10.1 I get the same problem that you. In my case, it works With sax2-vesa option suggested by Stephen MacDonald. 

What is the status of this bug?  I note that it has been marked as "resolved/duplicate of Bug 179773".  However, I get an "access denied" message when trying to view Bug 179773.

Will I be able to use my i915-based chipset using the i810 driver, or will I need to use slow VESA driver instead?  I'm currently running SuSE 9.3 and would like to upgrade to 10.1, but not if the video isn't going to work.

perhaps you could try to remove the package 855resolution, for me it helped -> see bug 187794

reopen

*** This bug has been marked as a duplicate of 191987 ***

Now I have solved the graphical problems......I bought a Nvidia mx 4000 graphics card. It works very fine and I got a DVI connection to my screen wich made the picture even better. 
But I still think suse 10.0 got lot higher quality....and it's bad that you cant use the hardware you used for 10.0. 
By the way I also installed the "Smart Package Manager" to be able to update the Yast2 sysem making the Yast2 update system start working!  
"Smart Package Manager" is very very nice.