FC5: kern and nvidia mod - no longer playing nice

Discussion in 'Kernel Questions' started by jsabarese, Oct 12, 2006.

  1. jsabarese

    jsabarese New Member

    hi. i'm having an issue w/ my 2.6.17-1.2187_FC5 kernel and nvidia geforce card.

    there's a lot to the story, and it pains me so to be brief, but i'll do my best for the sake of humanity!

    from start to finish, it's basically like this:
    - fresh 120GB drive > install FC5 (5 CD iso set)
    - using the Stanton-Finley.net guide, step by step, i follow the updates including the nvida module.
    - YumEx updates the kernel (after i've installed the nvidia modue as above) from 2.6.17-1.2157_FC5 to 2.6.17-1.2187_FC5

    the system had been working fine all along. one bit of software, which i believe played a bit part in this tragedy, was one i enjoy for image sorting-- Picasa, a non-native linux app. i had an XP drive slaved from which i was grabbing a bunch of photos using Picasa-- the linux version (which i believe runs via Wine).

    for some reason, i decided many days later that i wasn't sure if i had ever installed the nvidia module, by way of yum -y --enablerepo livna install kmod-nvidia, which i've learned from stanton-finley.net. so, as root, i did [ yum -y --enablerepo livna install kmod-nvidia ] (again? i'm not sure, but i think yes, this was in fact the second time, and recall-- post-kernel update).

    an unusual dialogue appeared later during this desktop session, i'm pretty sure it was "Picasa", which requested i CHMOD 666 as ROOT two files, which i recall contained the name 'nvidia' themselves. the request was much like "cannot [ do something w/ Picasa ] because [some-file-name] is [ locked / unwritable / being uncooperative ]. please CHMOD 666 the files aBc-nvidia and nvidia-Xyz". so, i proceeded to CHMOD 666 as root.

    it was not until i finally ended this desktop session and rebooted that i realized i would not be able to boot back into x-server w/ the exisiting configuration. grasping at straws, i decided to try the 'old' kernel which, due to the stanton-finley configuration, remained visible for a few seconds at the begining of the boot process.

    booting into this older 2.6.17-1.2157_FC5 kernel, the x-server came to life. however, it's not right:
    Code:
    [@localhost ~]$ locate /lib/modules/2.6.17-1.2157_FC5/*/nvidia
    [@localhost ~]$ locate /lib/modules/2.6.17-1.2187_FC5/*/nvidia
    /lib/modules/2.6.17-1.2187_FC5/extra/nvidia
    /lib/modules/2.6.17-1.2187_FC5/extra/nvidia/nvidia.ko
    i have experienced some 'instability' in the video (screen flicker, etc., depending upon what i'm doing w/ what video intensive stuff) and i just know it's not right.

    what can i do to get back to the proper, up-to-date kernel matched w/ the proper nvidia module? (remember, although it's there, i can't boot into KDE or Gnome on 2.6.17-1.2187_FC5)
    throughout my efforts to correct the problem, i recall doing the opposite of the nvida mod install [yum -y remove kmod-nvidia ]

    what should be my first step to correcting this problem?

    thanks !!
     
    Last edited: Oct 12, 2006
  2. falko

    falko Super Moderator Howtoforge Staff

    I'd try to play around with the permissions of the two files you chmod'ed. Maybe you also find some hints in your log files.
     
  3. jsabarese

    jsabarese New Member

    hey, thanks, Falko! yours is a 'new' approach to this nvidia module issue, vs other communities i've solicited w/ this problem. it's nice to hear a less single-minded viewpoint.

    where would i want to look for which logs? ( not expecting you to have an exact PATh of course... but, are we talking about Yum log, KDE log... etc., which folder(s) of the Filesystem might begin a search to find 'em, etc)

    unfortunately, i have no recollection of what were the true names of those CHMOD'd files... it was a bad call on my part-- i should have recorded those names-- at least via screencapture...

    let's assume that i reach a dead-end w/ these logs.
    the situation, as it stands: i have an 'older' Kernel. i have an unsupporting / unavailable nvidia module
    must i get rid of that module first, or isn't it going to play a part in my application of a supporting module (nvidia module supporting the kernel).

    why wouldnt i just go back to the other Kernel?
    at this point, it won't book to that kernel, it's true-- but what i mean is-- why not simply "plug-in" the proper configuration so i CAN go back to that kernel / nvidia module? i just don't understand... if i was able to muck-it-up so easily-- why not un-muck it just as easy too? i have difficulty with it being so "permanently changed" that i can't just reverse it more easily.
    i realize i'm speaking from ignorance, so please feel free to educate me about any of that. URL's are always appreciated too. ;)

    thanks!!!
     
    Last edited: Oct 13, 2006
  4. jsabarese

    jsabarese New Member

    Update:

    discovery!
    i know now which of the files i CHMOD to 666.

    i tried launching Picasa (for investigative, not leisurely practice), which again (just as before) shows
    does this mean anything to anyone here-- particularly in regard to why my video module installation would cause this CHMOD request from Picasa?

    NOTE: What i keep getting from all of this is that i need to either :
    1). install the software for the '2.6.17-1.2187_FC5' kernel, and (however it is done, i don't know, but...) then get back to that kernel. i'm currently in 2.6.17-1.2157_FC5
    - OR -
    2.) install the software for the 2.6.17-1.2157_FC5 kernel -- which i've had difficulty doing because 'yum' and such seem to only install for the .2187 kernel.
    (grrr... why is this so impossible?!?)

    also, note the other steps i've taken since i last updated this situation here at howtoforge.com . this represents my actions based on suggestion from advice i found here and there this evening at other on-line resources. i still haven't been able to fix this problem. my most recent attempts have been as follows, in order:
    Code:
    Removing:
     xorg-x11-drv-nvidia     i386       1.0.8774-1.lvn5  installed          12 M
    Removing for dependencies:
     kmod-nvidia             i686       1.0.8774-1.2.6.17_1.2187_FC5  installed         4.9 M
    to which i replied 'Y'
    ...and after that successful action, :
    Code:
    Reading repository metadata in from local files
    Parsing package install arguments
    Resolving Dependencies
    --> Populating transaction set with selected packages. Please wait.
    ---> Downloading header for kmod-nvidia-legacy to pack into transaction set.
    kmod-nvidia-legacy-1.0.71 100% |=========================|  15 kB    00:00
    ---> Package kmod-nvidia-legacy.i586 0:1.0.7184-2.2.6.17_1.2187_FC5 set to be installed
    --> Running transaction check
    --> Processing Dependency: kernel-i586 = 2.6.17-1.2187_FC5 for package: kmod-nvidia-legacy
    --> Processing Dependency: nvidia-legacy-kmod-common >= 1.0.7184 for package: kmod-nvidia-legacy
    --> Restarting Dependency Resolution with new changes.
    --> Populating transaction set with selected packages. Please wait.
    ---> Downloading header for kernel to pack into transaction set.
    kernel-2.6.17-1.2187_FC5. 100% |=========================| 166 kB    00:05
    ---> Package kernel.i586 0:2.6.17-1.2187_FC5 set to be installed
    ---> Downloading header for xorg-x11-drv-nvidia-legacy to pack into transaction set.
    xorg-x11-drv-nvidia-legac 100% |=========================|  22 kB    00:00
    ---> Package xorg-x11-drv-nvidia-legacy.i386 0:1.0.7184-2.lvn5 set to be updated
    --> Running transaction check
    --> Populating transaction set with selected packages. Please wait.
    ---> Package kernel.i686 0:2.6.17-1.2187_FC5 set to be erased
    --> Running transaction check
    
    Dependencies Resolved
    
    =============================================================================
     Package                 Arch       Version          Repository        Size
    =============================================================================
    Installing:
     kmod-nvidia-legacy      i586       1.0.7184-2.2.6.17_1.2187_FC5  livna             1.3 M
    Removing:
     kernel                  i686       2.6.17-1.2187_FC5  installed          37 M
    Installing for dependencies:
     kernel                  i586       2.6.17-1.2187_FC5  updates            14 M
     xorg-x11-drv-nvidia-legacy  i386       1.0.7184-2.lvn5  livna             3.6 M
    
    Transaction Summary
    to which i also responded 'Y'
    which turned out:
    Code:
    =============================================================================
     Package                 Arch       Version          Repository        Size
    =============================================================================
    Installing:
     kmod-nvidia-legacy      i586       1.0.7184-2.2.6.17_1.2187_FC5  livna             1.3 M
    Removing:
     kernel                  i686       2.6.17-1.2187_FC5  installed          37 M
    Installing for dependencies:
     kernel                  i586       2.6.17-1.2187_FC5  updates            14 M
     xorg-x11-drv-nvidia-legacy  i386       1.0.7184-2.lvn5  livna             3.6 M
    
    Transaction Summary
    =============================================================================
    Install      3 Package(s)
    Update       0 Package(s)
    Remove       1 Package(s)
    Total download size: 19 M
    Is this ok [y/N]: y
    Downloading Packages:
    (1/3): kernel-2.6.17-1.21 100% |=========================|  14 MB    04:58
    (2/3): kmod-nvidia-legacy 100% |=========================| 1.3 MB    00:15
    (3/3): xorg-x11-drv-nvidi 100% |=========================| 3.6 MB    00:46
    Running Transaction Test
    Finished Transaction Test
    
    
    Transaction Check Error:   package kernel-2.6.17-1.2187_FC5 is already installed
    yum list n* shows:
    Code:
    nvidia-x11-drv.i386                      1.0.8774-3             freshrpms
    yum list x* shows:
    Code:
    xorg-x11-drv-i810-devel.i386             1.4.1.3-3.1            core
    xorg-x11-drv-tek4957.i386                1.0.0.1-1.2            core
    xorg-x11-drv-via-devel.i386              0.1.33.2-2             core
    
    [root@localhost ~]# yum info kmod-nvidia*
    Loading "installonlyn" plugin
    Setting up repositories
    dries                                                                [1/7]
    macromedia                                                           [2/7]
    rpmforge                                                             [3/7]
    core                                                                 [4/7]
    updates                                                              [5/7]
    freshrpms                                                            [6/7]
    extras                                                               [7/7]
    Reading repository metadata in from local files
    [root@localhost ~]#  
    and finally my last transaction at the time of posting this was...
    Code:
    Resolving Dependencies
    --> Populating transaction set with selected packages. Please wait.
    ---> Package xorg-x11-drv-nvidia-legacy.i386 0:1.0.7184-2.lvn5 set to be updated
    --> Running transaction check
    --> Processing Dependency: nvidia-legacy-kmod >= 1.0.7184 for package: xorg-x11-drv-nvidia-legacy
    --> Restarting Dependency Resolution with new changes.
    --> Populating transaction set with selected packages. Please wait.
    ---> Downloading header for kmod-nvidia-legacy to pack into transaction set.
    kmod-nvidia-legacy-1.0.71 100% |=========================|  15 kB    00:00
    ---> Package kmod-nvidia-legacy.i686 0:1.0.7184-2.2.6.17_1.2187_FC5 set to be updated
    --> Running transaction check
    
    Dependencies Resolved
    
    =============================================================================
     Package                 Arch       Version          Repository        Size
    =============================================================================
    Installing:
     xorg-x11-drv-nvidia-legacy  i386       1.0.7184-2.lvn5  livna             3.6 M
    Installing for dependencies:
     kmod-nvidia-legacy      i686       1.0.7184-2.2.6.17_1.2187_FC5  livna             1.3 M
    
    Transaction Summary
    =============================================================================
    Install      2 Package(s)
    Update       0 Package(s)
    Remove       0 Package(s)
    Total download size: 4.8 M
    Is this ok [y/N]: y
    Downloading Packages:
    (1/1): kmod-nvidia-legacy 100% |=========================| 1.3 MB    00:13
    Running Transaction Test
    Finished Transaction Test
    Transaction Test Succeeded
    Running Transaction
      Installing: kmod-nvidia-legacy           ######################### [1/2]
    FATAL: Could not open '/boot/System.map-2.6.17-1.2187_FC5': No such file or directory
      Installing: xorg-x11-drv-nvidia-legacy   ######################### [2/2]
    FATAL: Module nvidia not found.
    
    Installed: xorg-x11-drv-nvidia-legacy.i386 0:1.0.7184-2.lvn5
    Dependency Installed: kmod-nvidia-legacy.i686 0:1.0.7184-2.2.6.17_1.2187_FC5
    Complete!
    [root@localhost ~]#
    it looks to me like it keeps erring due to the *.2187_FC5 kernel
    mine is (uname -r)2.6.17-1.2157_FC5

    what am i missing / doing wrong? it's been recommended to me elsewhere that i
    but, after investigating, all i could find was my current run level, and the utmp file, which i couldn't open in gedit, so i don't know how to get into runlevel 3.

    any other suggestions? i really appreciate all the help here.
    thanks!!!
     
    Last edited: Oct 14, 2006
  5. jsabarese

    jsabarese New Member

    i'm back to the 2.6.17-1.2187_FC5 kernel now
    all it took was rebooting, failing the x-server login, but the 'automatic' process (requiring my root p/w at commandline) restarted the x-server upon 'trying' w/ a new config.

    once the x-server started, i was prompted w/ a GNOME-ish display-settings screen, into which i plugged 1280x1024 (up from 1024x768) and from 'thousands' of colors, i bumped it up to millions.

    but, it didn't boot into KDE as such. i can't seem to get my 1280x1024 resolution back. it's not showing up in the KDE display config (in the 'Control Center').

    suggestions?
     
  6. jsabarese

    jsabarese New Member

    looks like everything is "okay" now. i've got the 1280 x 1024 (after another re-boot and x fail)
    this time, when i got the Display settings prompt (before loading into KDE / the desktop manager), instead of just setting the 1280 x 1024, i took a closer look and realized that my driver was set to vesa instead of 'nv', so i modified that to 'nv', and also set the colors to 'millions of colors'.

    KDE booted, but gave me back the 1024 x 768. i immediately went into the Control Center and swithced the settings there to 1280 x 1024. everything seems okay now.

    :)
     
  7. falko

    falko Super Moderator Howtoforge Staff

    That's good to hear! :)
     
  8. jsabarese

    jsabarese New Member

    well-- so much for my wishful thinking. from what i learned in the meantime, 'nv' isn't the driver i want-- it's the open source 'generic nvidia driver'. not to mention, i've also since realized that nothing i've done so far will allow the system to boot on its own. every time, i must ultimately enter my p/w and modify that same setting to 'nv' because the nvidia driver-- from what i understand-- will not open (i even chmod 766 the directory just to see if that would fix it, but there was no visible change)

    the x-server continues to crash when loading (trying to load?) xorg.conf at boot-time due to:
    Code:
    could not open /lib/modules/2.6.17-1.2187_FC5/extra/nvidia/nvidia...
    it's no wonder selecting 'nvidia' at the crash-recovery x-server configuration dialogue also continues to fail. repeated attempts of selecting 'nvidia', w/ different display-resolution selections, 'always' fail, yet 'nv' always boots directly.

    do we know if the module is failing to open as a direct result of, or independent of other variables (ie. is it failing because of reasons other than what kmod* and xorg*nvida module/drivers i've installed)? ... and whatever other bits are related... aw fiddlesticks!, i don't even think i ever did know for sure what all i'm looking for here! ha! not to mention, i'm a little burned out on it by now, and frankly, beginning to lose faith (hmph! well... i guess it does still 'work', technically...)

    going from what we know of the x-server boot failure, and what i've got installed, what options do i have?

    other than what i've listed below, what do i need to determine about the existing state of my system before i can solve this problem?

    Code:
    $ yum --enablerepo=livna list kmod*
    [color=blue]Installed Packages
    kmod-nvidia-legacy.i686                  1.0.7184-2.2.6.17_1.21 installed[/color]
    [color=green]Available Packages
    kmod-nvidia.i586                         1.0.8774-1.2.6.17_1.21 livna
    kmod-nvidia.i686                         1.0.8774-1.2.6.17_1.21 livna
    kmod-nvidia-kdump.i686                   1.0.8774-1.2.6.17_1.21 livna
    kmod-nvidia-legacy.i586                  1.0.7184-2.2.6.17_1.21 livna
    kmod-nvidia-legacy-kdump.i686            1.0.7184-2.2.6.17_1.21 livna
    kmod-nvidia-legacy-smp.i686              1.0.7184-2.2.6.17_1.21 livna
    kmod-nvidia-legacy-xen0.i686             1.0.7184-2.2.6.17_1.21 livna
    kmod-nvidia-smp.i686                     1.0.8774-1.2.6.17_1.21 livna
    kmod-nvidia-xen0.i686                    1.0.8774-1.2.6.17_1.21 livna[/color]
    $ yum --enablerepo=livna list xorg-x11-drv-n*
    [color=blue]Installed Packages
    xorg-x11-drv-nv.i386                     1.2.0-3.fc5            installed
    xorg-x11-drv-nvidia-legacy.i386          1.0.7184-2.lvn5        installed[/color]
    [color=green]Available Packages
    xorg-x11-drv-nvidia.i386                 1.0.8774-1.lvn5        livna
    xorg-x11-drv-nvidia-devel.i386           1.0.8774-1.lvn5        livna
    xorg-x11-drv-nvidia-legacy-devel.i386    1.0.7184-2.lvn5        livna[/color]
    and the complete contents of my xorg.conf:
     
  9. jsabarese

    jsabarese New Member

    and the bell tolls for me

    looks like my time is up. i have no idea how it happened, but i can't get log into root anymore via ]$ su or ]$ su -
    i tried a few different things, such as moving to the filesystem root and trying again. no luck. time to start over or what?

    the error is that i'm entering the wrong password. how weird is that... hmm... maybe it's a keyboard issue... if ya don't hear from me for a while, i guess you'll know why ;)
     
  10. falko

    falko Super Moderator Howtoforge Staff

    Directories must always be executable by everyone, so 766 is not good. 777 or 755 or 731, etc. is ok, but not 766.
     
  11. jsabarese

    jsabarese New Member

    the story continues. (forget the password thing... maybe i have dust under a key that's causing it... it's intermittent... weird)

    i was ready to close the book on this thing. i made it back to the 2.6.17-1.2187_FC5 kernel, dug around the nvidia settings and FINALLY got the nvidia-legacy driver to work... which i've now been testing by executing the following at the CLI:
    Code:
     ]# nvidia-settings 
    i got into the NV-CONTROL but there wasn't really anything in there i felt i needed to change-- i just wanted to see it open successfully, which it did. i figured, by then-- i was using the newest kernel, and the appropriate nvidia driver was working-- it even booted w/out error! once. (maybe twice).

    that's right-- just a couple boots all the way into the desktop using that configuration (a config which i thought for certain was sufficiently operable). since then, all i've got has been a blank screen (black) where typically the login-screen would appear-- just after the startup info is finished scrolling. i guess you could say, it makes it throught the startup process-- i even see the 'enable nvidia-legacy [OK]' during startup

    so-- here i am, back at the 2.6.17-1.2157_FC5 kernel again-- i figured i could have booted into the newest, but i don't know how to manipulate the process... wait! would that be to put 'nv' in place of 'nvidia' in Xorg.conf? i'll try that-- but it still doesn't explain why it goes fine and then -boom- all of a sudden not working again.

    if you have an extra moment, please take a look at the info i posted in regard to problems i'm having with yum not updating-- hasn't updated properly since sept. whether it's the cause of these problems, i don't know, but i'm certainly suspicious of it.
     
    Last edited: Oct 16, 2006
  12. falko

    falko Super Moderator Howtoforge Staff

    I don't think this problem is related to yum, but I also don't know what's causing it... :confused:
     

Share This Page