On Sunday 07 September 2003 09:15, Curtis Rey wrote:

On Sunday 07 September 2003 03:55, Jan Elders wrote: // CUT

...

Hi all, After upgrading my kernel (including the source) with YOU to 2.4.20-100, I have downloaded the NVIDIA Installer for Linux IA32. Subsequently, I followed the instructions in www.suse.de/de/private/download/xfree86/index.html section “SuSE Linux 8.2 with update kernel or self-compiled kernel”. - went to runlevel 3 (by restart in Safe Mode) - did the following : cd /usr/src/linux make cloneconfig && make dep cd /windows/e/Linuxdata/Downloads/nvidia/ (= my download directory) export IGNORE_CC_MISMATCH=yes sh NVIDIA-Linux-x86-1.0-4496-pkg2.run - thus ran the NVIDIA Installer - then I changed /etc/X11/XF86Config section Device: replaced Driver “nv” by Driver “nvidia” (according to the NVidia installation instructions) - went to runlevel 5 --> No more display ! (remains black) !!!! - I solved this by running SaX2 (in Safe Mode) : set Monitor model = LCD 1024 x 768 @ 60 Hz set Color & Resolution = 1024 x 768 - 24 bit After this I had my display back, but no 3D !!!! Looking in XF86Config shows that Driver is set back to “nv” (apparently by SAX2).

So .... it seems that setting the driver to "nvidia" ruins my system ? Or , in other words, I think the new NVidia driver is not working correct here. BTW, I'm using a laptop with a 1024x768 LCD screen. What could be the problem ?

Have a look at this page on the SBD, it might help for place to start.

http://sdb.suse.de/sdb/en/html/wessels_blackinst.html

Yes, I studied this and tried some things, but it seems that the basic problem is setting the driver to 'nvidia' always leads to a black screen (on my system), while 'nv' works all-right. Does everybody else work OK under SuSE 8.2 kernel 2.4.20-100 with the 'nvidia' driver ? Greetings, -- Jan Elders Nuenen the Netherlands

On Monday 08 September 2003 21:48, Curtis Rey wrote:
> On Monday 08 September 2003 15:11, Jan Elders wrote:
> > On Sunday 07 September 2003 09:15, Curtis Rey wrote:
> > > On Sunday 07 September 2003 03:55, Jan Elders wrote:
> > > > Hi all,
> > > > After upgrading my kernel (including the source) with YOU to
> > > > 2.4.20-100, I have downloaded the NVIDIA Installer for Linux IA32.
> > > > Subsequently, I followed the instructions in
> > > > www.suse.de/de/private/download/xfree86/index.html section “SuSE
> > > > Linux 8.2 with update kernel or self-compiled kernel”.
// CUT
> > > > - went to runlevel 5 after installation
> > > >    --> No more display ! (remains black) !!!!
> > > > - I solved this by running SaX2 (in Safe Mode) :
> > > >    set Monitor model = LCD  1024 x 768 @ 60 Hz
> > > >    set Color & Resolution = 1024 x 768 -  24 bit
> > > > After this I had my display back, but no 3D !!!!
> > > > Looking in XF86Config shows that Driver is set back to “nv”
> > > > (apparently by SAX2).
> > > > So .... it seems that setting the driver to "nvidia" ruins my system
> > > > ? Or , in other words, I think the new NVidia driver is not working
> > > > correct here.
> > >
> > > Have a look at this page on the SBD, it might help for place to start.
> > > http://sdb.suse.de/sdb/en/html/wessels_blackinst.html
> >
> > Yes, I studied this and tried some things, but it seems that the basic
> > problem is setting the driver to 'nvidia' always leads to a black screen
> > (on my system), while 'nv' works all-right.
> > Does everybody else work OK under SuSE 8.2 kernel 2.4.20-100 with the
> > 'nvidia' driver ?
>
> Ok, now one of the problems I have encountered before is bad or incomplete
> symlinks and/or the incomplete removal of old files/symlinks that kludge
> the install.  Have a look at the nvidia readme, down about the middle of
> the doc. You'll see something that refers to check the files/symlinks.  I
> wouldn't be surprised is you have files that are not being replaced
> properly and somewhere in your system a file is reporting an older version
> -- i.e. 4232 verses the current version.
>
> What you could do is got through the NVIDIA readme and perform some of the
> checks it suggests paste it into a follow up email if your still not having
> any luck.
OK, I went through this excercise.
Not all appears to be well :
1. Suggested check with 'xdpyinfo' does NOT show the "GLX", "NV-GLX" and 
"NVIDIA-GLX" extensions ! However I do have the statement   Load  "glx"  in 
the Module section of my XF86Config. 
Also in /var/log/XFree86.0.log I see the message :
(EE) [GLX]: Failed to add GLX extension (NVIDIA XFree86 driver not found)
This must be the cause of the problem, but what to do about it ?
The driver installation process went all-right as far as I can judge.
2. All symlinks to be checked (as mentioned in Appendix C of the NVidia 
README) are correct.
3. The check for correct libraries being used, initially didn't work, because 
I have no /usr/X11R6/bin/gears as mentioned in the README.
However, I saw that I do have a "glxgears" in /usr/X11R6/bin.
Running the command ldd /usr/X11R6/bin/glxgears gave me a different result 
from what is shown in the README.
I get :
        libGL.so.1 => /usr/lib/libGL.so.1 (0x40027000)
        libXext.so.6 => /usr/X11R6/lib/libXext.so.6 (0x40091000)
        libX11.so.6 => /usr/X11R6/lib/libX11.so.6 (0x400a0000)
        libpthread.so.0 => /lib/libpthread.so.0 (0x4016e000)
        libm.so.6 => /lib/libm.so.6 (0x401be000)
        libc.so.6 => /lib/libc.so.6 (0x401e0000)
        libGLcore.so.1 => /usr/lib/libGLcore.so.1 (0x40316000)
        libdl.so.2 => /lib/libdl.so.2 (0x407ee000)
        /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)

So apparently phenomenon 1 and maybe 3 above are the cause of my black screen 
problem ? But what can I do about it ? I already tried twice to do the driver 
installation, carefully following the instructions both times. I probably did 
miss something though, but what ?

Cheers,
-- 
Jan Elders
Nuenen
the Netherlands

On Tuesday 09 September 2003 20:56, Curtis Rey wrote:

...

OK, I went through this excercise. Not all appears to be well : 1. Suggested check with 'xdpyinfo' does NOT show the "GLX", "NV-GLX" and "NVIDIA-GLX" extensions ! However I do have the statement Load "glx" in the Module section of my XF86Config. Also in /var/log/XFree86.0.log I see the message : (EE) [GLX]: Failed to add GLX extension (NVIDIA XFree86 driver not found) This must be the cause of the problem, but what to do about it ? The driver installation process went all-right as far as I can judge. // CUT

Ok, run this command and mail the output.

la /usr/lib/libGL*

This will show the files for the GL libs and then we can see if the symlinks are established. From time to time this will kludge and then it has to be manually be made. An example would be the libGL.so will point to nowhere, or it may point to libGL.so.1 but libGL.so.1 isn't there or points somewhere else. Here is my output:

lrwxrwxrwx 1 root root 10 2003-08-13 10:42 /usr/lib/libGL.so - -> libGL.so.1 lrwxrwxrwx 1 root root 17 2003-08-13 10:42 /usr/lib/libGL.so.1 -> libGL.so.1.0.4496 - -rwxr-xr-x 1 root root 423476 2003-08-13 10:42 /usr/lib/libGL.so.1.0.4496 lrwxrwxrwx 1 root root 13 2003-05-10 08:12 /usr/lib/libGLU.so.1 -> libGLU.so.1.3 - -rwxr-xr-x 1 root root 702207 2003-03-17 08:40 /usr/lib/libGLU.so.1.3 lrwxrwxrwx 1 root root 21 2003-08-13 10:42 /usr/lib/libGLcore.so.1 -> libGLcore.so.1.0.4496 - -rwxr-xr-x 1 root root 5060036 2003-08-13 10:42 /usr/lib/libGLcore.so.1.0.4496

So, this is one of the steps to see if things are setup correctly. Making a symlink is really rather simple if need be. So let's check this out and see if it needs fixing or we need to move to the next step. Oh and by the way, what version are you running again?

I am running 2.4.20-100 Output of la /usr/lib/libGL* is : lrwxrwxrwx 1 root root 10 2003-09-08 22:23 /usr/lib/libGL.so -> libGL.so.1 lrwxrwxrwx 1 root root 17 2003-09-08 22:23 /usr/lib/libGL.so.1 -> libGL.so.1.0.4496 -rwxr-xr-x 1 root root 423476 2003-09-08 22:23 /usr/lib/libGL.so.1.0.4496 lrwxrwxrwx 1 root root 13 2003-04-20 17:49 /usr/lib/libGLU.so.1 -> libGLU.so.1.3 -rwxr-xr-x 1 root root 702207 2003-03-17 15:40 /usr/lib/libGLU.so.1.3 lrwxrwxrwx 1 root root 21 2003-09-08 22:23 /usr/lib/libGLcore.so.1 -> libGLcore.so.1.0.4496 -rwxr-xr-x 1 root root 5060036 2003-09-08 22:23 /usr/lib/libGLcore.so.1.0.4496 This looks perfect (and identical to your output) to me. BTW, did you notice my remark about /var/log/XFree86.0.log ?? In there, I see the message : (EE) [GLX]: Failed to add GLX extension (NVIDIA XFree86 driver not found) I think this must be the cause of the problem, but what to do about it ? The driver installation process went all-right as far as I can judge. Cheers, -- Jan Elders Nuenen the Netherlands

On Wednesday 10 September 2003 19:56, Curtis Rey wrote:
// CUT
>
> Ok, what type of monitor do you have.  I have had this with sort of things
> happen with a monitor - mod line or v and h sync problems. So sometimes you
> have to tweak the X settings. 
I have a laptop with a 1024x768 LCD screen.

> Though the problem does seem to stem around
> the GLX drivers.  You might try reinstalling the drivers again.  Did you
> use the nvidia installer?  Or did you use the source code.  
Yes, I have used the nvidia installer, and just did it again for the 3rd time.
I followed the instructions in
www.suse.de/de/private/download/xfree86/index.html section “SuSE Linux 8.2
with update kernel or self-compiled kernel”.
- went to runlevel 3 (by restart in Safe Mode)
- did the following :
   cd /usr/src/linux
   make cloneconfig 
   make dep
   cd  /windows/e/Linuxdata/Downloads/nvidia/ (= my download directory)
   export IGNORE_CC_MISMATCH=yes
   sh NVIDIA-Linux-x86-1.0-4496-pkg2.run
Thus ran the NVIDIA Installer.
During the process the installer tried to contact the nvidia ftp-site (I had 
an Internet connection available - and working) and it told me :
"No matching precompiled kernel interface was found on the NVidia ftp site,       
this means the installer will need to compile a kernel interface for your    
kernel."

After this, I hit OK.
After some processing it told me :
"Installation of the NVidia Accelerated Graphics Driver for Linux-x86 (version 
: 1.0-4496) is now complete. On SuSE Linux/United Linux please use SAX2 now 
to enable the NVidia driver."
After this, I hit OK again and the installer exits.
So far this looks good to me.

Then I give the command : sax2
and ................ I have a black screen !

CTRL-ALT-DEL has no effect, so I have to cut power !  ;-(

Next, I reboot to runlevel 3.
I restore my "old" XF86Config file (which I had saved before).
Then I go to runlevel 5 -> all is OK (using the "old" nv driver).
In Shell Konsole, I go to sax2.
In Desktop I have :
- Monitor = VESA 1024x768@60Hz
- Graphics card = NVidia GeForce4 488 Go
    after "finish" it shows "nvidia" at the end of the information-bar about 
the configured graphics card.
- Color & Resolution = Desktop 1024x768 with 16 bit
- 3D accelleration is off, but I activate it.
Then I do "finalize"  --> "test"  
and ................ I have a black screen again !

This time CTRL-ALT-DEL does work (it shuts down and reboots).
I boot to runlevel 3 and restore my "old" XF86Config file again.
Then I go to runlevel 5 and all is OK again (but using the "old" nv driver).

> If You still
> have troubles then try to download the drivers using the installer and then
> make the source package.  It's covered in the nvidia README as far as
> options are concerned.  Also there's the flag for the installer "--help"
> that list a set of commands, one of these is also an advanced set of
> commands "--adv" IIRC.
>
> If you  download the source binaries move them to /usr/src/package/SOURCES
> and then untar them and do a make install for the GLX drivers. 
Ehh, all of the above is something I never did before. It is not totally clear 
to me what -exactly- I must do (e.g. how do I do a "make install for the GLX 
drivers ?)

 > If they
> fail it might provide some useful output.  If not then it should work and
> if that fails then we have some deeper digging to do.  I have had a couple
> of instances where this happens.  Sometimes you have to uninstall and
> reinstall the drivers.  
Haven't done this yet. Do you recommed this ?

> ...........  Just be hang in
> there, I know all too well how frustrating this can be.
You are so very right. I am starting to get fed-up by all this.
Black screens, cutting power, unclean filesystems.

> Oh and last but not by anymeans least, do an "lsmod" and see if the kernel
> drivers are actually loaded - I have had this happen before as well.
>
> it should be something like:
>
> # lsmod
> Module                  Size  Used by    Tainted: P
> snd-pcm-oss            50144   1  (autoclean)
> snd-mixer-oss          15704   0  (autoclean) [snd-pcm-oss]
> videodev                6176   0  (autoclean)
> agpgart                38656   3  (autoclean)   *************
> nvidia               1628576  11  (autoclean)  *************
>
> The ones marked "****" are often needed in conjunction with each other.
Well, if I do this then - of course - it does not show "agpgart" and "nvidia" 
because I am running with the "old" configuration which gives me a visible 
screen.
I would love to do the lsmod in the black screen situation and see what is 
listed. Wouldn't that be a miracle ?  ;-)

-- 
Jan Elders
Nuenen
the Netherlands

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Thursday 11 September 2003 15:16, Jan Elders wrote:

On Wednesday 10 September 2003 19:56, Curtis Rey wrote: // CUT

...

Ok, what type of monitor do you have. I have had this with sort of things happen with a monitor - mod line or v and h sync problems. So sometimes you have to tweak the X settings.

I have a laptop with a 1024x768 LCD screen.

...

Though the problem does seem to stem around the GLX drivers. You might try reinstalling the drivers again. Did you use the nvidia installer? Or did you use the source code.

Yes, I have used the nvidia installer, and just did it again for the 3rd time. I followed the instructions in www.suse.de/de/private/download/xfree86/index.html section “SuSE Linux 8.2 with update kernel or self-compiled kernel”. - went to runlevel 3 (by restart in Safe Mode) - did the following : cd /usr/src/linux make cloneconfig make dep cd /windows/e/Linuxdata/Downloads/nvidia/ (= my download directory) export IGNORE_CC_MISMATCH=yes sh NVIDIA-Linux-x86-1.0-4496-pkg2.run Thus ran the NVIDIA Installer. During the process the installer tried to contact the nvidia ftp-site (I had an Internet connection available - and working) and it told me : "No matching precompiled kernel interface was found on the NVidia ftp site, this means the installer will need to compile a kernel interface for your kernel."

After this, I hit OK. After some processing it told me : "Installation of the NVidia Accelerated Graphics Driver for Linux-x86 (version

: 1.0-4496) is now complete. On SuSE Linux/United Linux please use SAX2 now

to enable the NVidia driver." After this, I hit OK again and the installer exits. So far this looks good to me.

Then I give the command : sax2 and ................ I have a black screen !

CTRL-ALT-DEL has no effect, so I have to cut power ! ;-(

Next, I reboot to runlevel 3. I restore my "old" XF86Config file (which I had saved before). Then I go to runlevel 5 -> all is OK (using the "old" nv driver). In Shell Konsole, I go to sax2. In Desktop I have : - Monitor = VESA 1024x768@60Hz - Graphics card = NVidia GeForce4 488 Go after "finish" it shows "nvidia" at the end of the information-bar about the configured graphics card. - Color & Resolution = Desktop 1024x768 with 16 bit - 3D accelleration is off, but I activate it. Then I do "finalize" --> "test" and ................ I have a black screen again !

This time CTRL-ALT-DEL does work (it shuts down and reboots). I boot to runlevel 3 and restore my "old" XF86Config file again. Then I go to runlevel 5 and all is OK again (but using the "old" nv driver).

...

If You still have troubles then try to download the drivers using the installer and then make the source package. It's covered in the nvidia README as far as options are concerned. Also there's the flag for the installer "--help" that list a set of commands, one of these is also an advanced set of commands "--adv" IIRC.

If you download the source binaries move them to /usr/src/package/SOURCES and then untar them and do a make install for the GLX drivers.

Ehh, all of the above is something I never did before. It is not totally clear to me what -exactly- I must do (e.g. how do I do a "make install for the GLX drivers ?)

...

If they

fail it might provide some useful output. If not then it should work and if that fails then we have some deeper digging to do. I have had a couple of instances where this happens. Sometimes you have to uninstall and reinstall the drivers.

Haven't done this yet. Do you recommed this ?

...

........... Just be hang in there, I know all too well how frustrating this can be.

You are so very right. I am starting to get fed-up by all this. Black screens, cutting power, unclean filesystems.

...

Oh and last but not by anymeans least, do an "lsmod" and see if the kernel drivers are actually loaded - I have had this happen before as well.

it should be something like:

# lsmod Module Size Used by Tainted: P snd-pcm-oss 50144 1 (autoclean) snd-mixer-oss 15704 0 (autoclean) [snd-pcm-oss] videodev 6176 0 (autoclean) agpgart 38656 3 (autoclean) ************* nvidia 1628576 11 (autoclean) *************

The ones marked "****" are often needed in conjunction with each other.

Well, if I do this then - of course - it does not show "agpgart" and "nvidia" because I am running with the "old" configuration which gives me a visible screen. I would love to do the lsmod in the black screen situation and see what is listed. Wouldn't that be a miracle ? ;-)

Ok, this sounds like the ever popular monitor fun with sax2. I unfortunately don't have as much experience with LCDs. I might suggest that you could try the command SaX2 -r This will reinit the database. sax2 --help ...[ -r | --reinit ] remove detection database and re-init the hardware database... I strongly suspect that this is all about the monitor/video card/drivers not being set up in the correct fashion (through no fault of yours). I situations like this I have had to hand modify the XF86Config file in the v-sync/h-sync do lower levels, and also at time tinker with the modelines. I am downloading some ISOs right now but should be done before too long (if the SuSE server doesn't drop the bandwidth to a lower throughput). As far as the modelines, there are a couple of ways to try to see if they're reconfiged. One is to use sax2 -x [ -x | --xmode ] use server build in Modelines The other methods is from the XF86.org site and tells how to manually probe the modeline timings for each resolution. They also have sections that explains setup and troubleshooting LCDs. LCDs can be a specific pain in the ass (nvidia readme also has a section for LCDs). You should be able to put back in the NVIDIA specific XF86Config file and at boot up in the command section (in lilo/grub) issue the command "init 3". Then you boot to a non-X mode but if the nvidia kernel driver is installed it should load the modules. You can issue the "lsmod" and see if nvidia and agpgart modules are loaded. Then you can copy the XF86config file back again. You should have two files. One with "nv" as the driver and another as "nvidia" as the driver, i.e. XF86Config-nv and XF86Config-nvidia. Then issue the command in the /etc directory "cp XF86Config-nvidia XF86Config" to replace the nv file, and "cp XF86Config-nv XF86Config" to put it back to a working config file. So you actually have 3 XF86Config files. One that is labelled XF86Config, one labelled XF86Config-nv, and a 3rd named XF86Config-nvidia. When you cp (copy) the *-nv or *-nvidia to the working XF86Config file you keep the nv/nvidia config files and just make the working file with either the nv or nvidia driver. This way you can have the nvidia driver in the X file, reboot to init 3, issue the lsmod to see if the driver is loaded, then cp the *-nv file to the working X file, issue init 5 and go to a non-3D driver and X will start without a black screen. Also you need to look in /etc/module.conf file and see if the line "alias char-major-195 nvidia". If it's not then this explains why you get a black screen. I will look on the XF86.org pages for LCD stuff when I get a chance after dl'ing my ISOs (I have access now and don't want to stop since I might have to wait another 1/2 day or more before I get another chance), and you can try the same by looking at both the NVIDIA README and on the XF86.org site to see if you can't find a fix. I really think that this is all about the LCD and something specific to the monitor. If you have the specs for the monitor (i.e. the notebook manual) with the v-sync/h-sync/refresh rates then you might be able to check this against the XF86Config file. Keep trying, we'll nail this yet. :) Curtis. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2-rc1-SuSE (GNU/Linux) iD4DBQE/YVR6iqnGhdjCOJsRAjHmAJ0e1OtRx6TB+ICCbMMDh2YsyFrWuACVH5YS YcJIoOFFMCSUIQ5U6CbbpA== =Agkq -----END PGP SIGNATURE-----

The 03.09.12 at 23:20, Jan Elders wrote:

* cancel sax2 * goto init 5 (so, with the nvidia-XF86Config) just to try -> black screen

Instead of that, just do "startx" from the terminal (better as user, not root). Errors get printed on the console, and it's easier to kill. And, "startx fwmn" or similar is faster than kde.

NOW HERE IS A STRANGE THING : When (in init 5 mode) I start sax2 from within a terminal, then I get a normal - good - sax screen. Then I cancel sax2, go down to init 3, and start sax2 in command mode, and then I get the fuzzy screen again I mentioned before.

I think that is related to this text (sax2 --help) that I posted here the other day, and that I said I do not fully understand : [ -s | --sysconfig ] this option tell SaX2 to import the system wide config file even if SaX2 was started from a textconsole which normaly will import the SaX2 HW detection data See? It seems that sax2 uses different data if started from a console or inside X.

BTW, someone told me that with TFT screens (like mine) there is no vsync/hsync. He said there is only a 'reaction time'. Is that not true ?

That must be true. On a glass CRT, there is an electron beam that is swept left to right, up to down, so as to "illuminate" every dot (pixel) on the screen. Parameters such as vsync/hsync, frequencies, times, and such relate to the times needed to "move" that electron beam, and how to synchronize it to the signals (ie, to know where the beam is anytime). They have a very physical meaning. A TFT screen is a type of a matrix display, with electronics activating whatever they activate (leds, LC orientation, whatever), in some kind of matrix, indexing or whatever (I lost track of the actual technologies long ago). It could be better used with a bus, instead of the VGA connector, and a specific video card tailored to that type of addressing (perhaps with no memory). To use the VGA connector they must use intermediate electronics that "read" the signals on the VGA cable and "convert" it to the matrix needed to activate the dots on the display. This is rough sketch of an explanation, and probably incorrect on some places and details, but it will suffice. The thing is that the frequencies supported are more or less arbitrary, probably not really very much related to the real refresh rate of the screen, and thus limited, much more limited that those available on a glass CRT. The more frequencies available, more money.

Anyway, I could not find any information about vsync/hsync in the documentation of my notebook.

Makes sense. Probably has a fixed frequency(s), supplied as VESA info. For example, mine (a cheap Proview monitor) is running at H:60KHz/V:75Hz, 1024*768 (15"), and absolutely no flicker whatsoever, even at lower frequencies. A CRT at this frequency would tire my eyesight.

Also, modelines is a new phenomenon to me. I'm afraid I'll have some studying to do before I know how to manipulate this.

I'm uncertain of the effect they would have there. Anyway, on mine I don't bother: it has an "auto" button that automatically adjusts everything fast. When I use my bigger and "better" CRT I sorely miss it :-) -- Cheers, Carlos Robinson

The 03.09.13 at 09:48, Joe Morris (NTM) wrote:

ctl-alt-f1, login and go to init 3, then start sax2, which DID auto detect the new monitor, and setup everything relating to the monitor perfectly. HTH.

Yes, that was my case also. -- Cheers, Carlos Robinson

I don't have as severe a problem, but with "nvidia", when I go CTRL-ALT-F1, I just have different length lines like ============ with long blue, then short red, short white, long blue. With "nv", it's fine, except I need GLX for other stuff, so I live with it. I've just noticed that tty2 tty3 etc. have the dreaded BLACK screen, but kdm is OK on tty's 7, 8,9,11, so I can use X. Regards Sid. Jan Elders wrote:

On Saturday 13 September 2003 01:15, Carlos E. R. wrote:

...

The 03.09.12 at 23:20, Jan Elders wrote:

...

* cancel sax2 * goto init 5 (so, with the nvidia-XF86Config) just to try -> black screen

Instead of that, just do "startx" from the terminal (better as user, not root). Errors get printed on the console, and it's easier to kill. And, "startx fwmn" or similar is faster than kde.

Hmmm, I followed your advice. After having made changes in my troublesome XF86Config in console mode, I did startx as user, and ....... got a black screen again. I could not find any other way to kill this than cutting power (which I hate to do). What do you mean by "easier to kill" ? Is there a method I'm not aware of ? CTRL-ALT-DEL doesn't work !

...

...

NOW HERE IS A STRANGE THING : When (in init 5 mode) I start sax2 from within a terminal, then I get a normal - good - sax screen. Then I cancel sax2, go down to init 3, and start sax2 in command mode, and then I get the fuzzy screen again I mentioned before.

I think that is related to this text (sax2 --help) that I posted here the other day, and that I said I do not fully understand :

[ -s | --sysconfig ] this option tell SaX2 to import the system wide config file even if SaX2 was started from a textconsole which normaly will import the SaX2 HW detection data

See? It seems that sax2 uses different data if started from a console or inside X.

Well, here the -s parameter does not give any relief. Things are going worse now. I tried the sax2 -s in console mode but when I do sax2 -s or just sax2, I don't even get the 'fuzzy' sax2 screen anymore. Now I just get some green and white scramble displayed and the system doesn't react to anything anymore. So : a) the -s parameter doesn't seem to make any difference. b) I lost the 'fuzzy' screen and have garbage now. This is maybe caused by the fact that I have tried to use the NvAGP option in my XF86Config ? (without success) See my earlier posting this evening. But after that I have re-activated my 'good' nv-based XF86Config, so where would sax2 get its erroneous information from ?

BTW : sax2 from terminal within X still works normal..

<snip>

I'm getting pretty puzzled by all these nvidia driver problems. ;-(

Cheers,

Jan Elders wrote:

On Wednesday 10 September 2003 19:56, Curtis Rey wrote: // CUT

...

Ok, what type of monitor do you have. I have had this with sort of things happen with a monitor - mod line or v and h sync problems. So sometimes you have to tweak the X settings.

I have a laptop with a 1024x768 LCD screen.

[Pruned]

...

Oh and last but not by anymeans least, do an "lsmod" and see if the kernel drivers are actually loaded - I have had this happen before as well.

it should be something like:

# lsmod Module Size Used by Tainted: P snd-pcm-oss 50144 1 (autoclean) snd-mixer-oss 15704 0 (autoclean) [snd-pcm-oss] videodev 6176 0 (autoclean) agpgart 38656 3 (autoclean) ************* nvidia 1628576 11 (autoclean) *************

The ones marked "****" are often needed in conjunction with each other.

Well, if I do this then - of course - it does not show "agpgart" and "nvidia" because I am running with the "old" configuration which gives me a visible screen. I would love to do the lsmod in the black screen situation and see what is listed. Wouldn't that be a miracle ? ;-)

Jan, Do what most of us appear to have done which is to throw the new 4496 nVidia in the toilet, where they belong, and use the previous version (4363). After all the troubles 4496 has caused to people one would have thought that nVidia would have pulled there finger out and fixed the driver but No, not a stir out of them. I check their site every night but there isn't even an acknowledgement that there may be problem with the 4496 driver. Save your sanity- go back to the 4363 driver. Cheers. -- Understanding only begins with the act of perception.

Gerhard den Hollander wrote:

* Basil Chupin (Fri, Sep 12, 2003 at 03:36:08PM +1000)

...

Do what most of us appear to have done which is to throw the new 4496 nVidia in the toilet, where they belong, and use the previous version (4363).

...

After all the troubles 4496 has caused to people one would have thought that nVidia would have pulled there finger out and fixed the driver but No, not a stir out of them. I check their site every night but there isn't even an acknowledgement that there may be problem with the 4496 driver.

...

Save your sanity- go back to the 4363 driver.

I missed out a large portion of this discussion, but Im curious to hear what issues people have with the 4496 driver.

I've installed 4496 on a vairety of machines and cards, running either suse 73 or suse82 (Geforce2, Geforce4, Geforce 5200, quadro 2 and 4 and the 2go/4go flavours for laptops) and never had an issue with them

[well, if you have a custom kernel, you might have to set the IGNORE__CC_MISMATCH (or similar) variable before installing the latest driver]

So, could you please let me know what issues there are with the 4496 ?

Apart from the problem outlined in the messages in this thread I, for one, get a black screen when I press CTRL-ALT-F2 to F6; and get a multi-coloured, blinking screen when I press CTRL-ALT-F1 and also when I either restart ot shutdown. If I don't do any of these then everything is fine - the driver works. But should some application lock up and I have to KILL it I can only do it by going to another Desktop, start a Console, login as root, do 'ps aux' and then KILL the app. Cheers. -- Indecision is the key to flexibility.

Jan Elders wrote:

On Friday 12 September 2003 07:36, Basil Chupin wrote: // cut

...

Jan, Do what most of us appear to have done which is to throw the new 4496 nVidia in the toilet, where they belong, and use the previous version (4363).

After all the troubles 4496 has caused to people one would have thought that nVidia would have pulled there finger out and fixed the driver but No, not a stir out of them. I check their site every night but there isn't even an acknowledgement that there may be problem with the 4496 driver.

Save your sanity- go back to the 4363 driver.

Hi Basil, It seems that I should follow your advice. I hope you can help me with one more question though : Must I first de-install 4496 and then install 4363 by running NVIDIA-Linux-x86-1.0-4363.run ? Or can I just install 4363 right away without de-install of 4496 ?

Also : can I just do the .run, or must I also preceed this with the commands make cloneconfig && make dep export IGNORE_CC_MISMATCH=yes like required for 4496 ?

TIA and greetings,

Hi Jan, Yes, uninstall 4496 first, and you will have to make cloneconfig dep (no need for the export ... part though) to be able to install 4363. After you have done the previous then just sh NVIDIA.....run and 4363 will be installed. Cheers. -- If logic ruled the World, men would ride side-saddle.

On Friday 12 September 2003 11:40, Marshall Heartley wrote:

<snip> I haven't been following this too closely but I know that my wife's machine requires a parameter in her XF86Config file in order for the nvidia drivers to work without a blackscreen.

I had to add the parameter Option "NvAGP" "1" to get it to work. This makes the system use nvidia's AGP support instead of agpgart. If the number stated doesn't help, you can try putting a 2 there instead of the 1. The 2 will tell the system to load the agpgart module. Be sure to switch to runlevel 3 then to 5 so the system will load up the appropiate module.

Just something to try.

Marshall, Where did you put this parameter ? Is it in the "Device" Section ? Cheers, -- Jan Elders Nuenen the Netherlands

On Saturday 13 September 2003 05:01, Curtis Rey wrote:

On Friday 12 September 2003 16:50, Jan Elders wrote:

...

On Friday 12 September 2003 11:40, Marshall Heartley wrote:

...

<snip> I had to add the parameter Option "NvAGP" "1" to get it to work. This makes the system use nvidia's AGP support instead of agpgart. If the number stated doesn't help, you can try putting a 2 there instead of the 1. The 2 will tell the system to load the agpgart module. Be sure to switch to runlevel 3 then to 5 so the system will load up the appropiate module. Just something to try.

Marshall, Where did you put this parameter ? Is it in the "Device" Section ?

Yes, here's what you need to look for:

Section "Device" BoardName "GeForce3 Ti 200" Driver "nvidia" Identifier "Device[0]" Option "CursorShadowAlpha" "64" Option "CursorShadowXOffset" "4" Option "DigitalVibrance" "175" Option "CursorShadow" "1" Option "NvAGP" "3" ******************************** Option "NoBandWidthTest" "1" Option "CursorShadowYOffset" "4" Option "NoLogo" "1" Option "NoMTRR" "1" Screen 0 Option "Rotate" "off" VendorName "NVidia"

Well, I gave this a try (of course). But ......... after having added Option "NvAGP" "3" in console mode, I did startx (as user, as recommended by Carlos Robinson) and ....... got a black screen again. After power cut (sigh !!!), I changed the Option value to "2", did startx again, and ....... got a black screen. So, after power cut, I changed the Option value to "1", did init 5 (for a change), HEY, now I have some green and white scramble in the top third of the screen, but that's all -- no further action. Had to cut power again. After reboot I let it run normally all the way to init level 5 and then .... I have a black screen. ;-((((((((( So finally, I restore my 'good old' nv-version of XF86Config and can work normally again. Any other options to try/use ? This is my Section "Device" of the 'black screen' XF86Config : Section "Device" BoardName "GeForce4 488 Go" BusID "1:0:0" Driver "nvidia" Identifier "Device[0]" Screen 0 Option "NvAGP" "3" (also tried 2 and 1) Option "Rotate" "off" VendorName "NVidia" EndSection Question to Curtis Rey : Did you miss my reply of Yesterday 23:20:28 to your posting ? I was looking forward to one of your very extensive helpful replies (although they have not worked out yet), but didn't see one so far.. Cheers, -- Jan Elders Nuenen the Netherlands

The 03.09.14 at 21:07, Jan Elders wrote:

My special thanks go to Curtis Rey who offered a lot of suggestions to get this solved, and also to Carlos Robinson who joined in later.

Welcome :-)

Nevertheless, I found this solution on my own. ;-)

I saw the thing about AGP somewhere, perhaps on another thread, but I lost track of how many sugestion were given. NVidia is such a pain when it choses to be problematic... -- Cheers, Carlos Robinson

The 03.09.15 at 21:21, Sid Boyce wrote:

Went back to 4363, looked great, but X (and keyboard) lock up with garbled colour full screen after I've left it up for some hours, I then have to issue init 6 by ssh from another box, init 3 does not have any effect. I think I shall try some other values for NvAGP other than 0 and see what happens. Regards Sid.

Er... in this case, as the machine was left alone for a long time, it could be a screen saver that crashed badly. Or something related to sleeping. If it does it again, instead of init 6, enter "top" as root, or have a look at "ps afx|less", and start killing tasks, one by one. Perhaps you could find the culprit. -- Cheers, Carlos Robinson

The 03.09.16 at 22:36, Jan Elders wrote:

Hmmm, in the meantime I found out that unfortunately my "SOLVED" message is not completely correct. ;-(

Ahh...

The 4496 driver works perfect under X, but ................ when leaving X, either by going to console mode with CTRL-ALT-F1 .... F6 or when shutting down/rebooting, then I ALWAYS get the garbled colour full screen that Sid is mentioning.

I have just restored my Nvidia driver, after the kernel update from YOU (-100) and gcc-3.3.1-16. It was tedious (downloading those over a modem is slow), and previously gcc-3.3.1-0 failed. So, after recompiling the kernel, the nvidia install went very smoothly, but I used my existing /usr/share/doc/nvidia/NVIDIA-Linux-x86-1.0-4363.run, not the new driver. I new the old one worked for me, I wanted no experiments there :-)

I wonder why NVidia keeps silent about this; by now they should have recognized these complaints and minimally react in some way. Cheers,

Many companies don't say a thing till solved. It could be that. -- Cheers, Carlos Robinson