No GUI from the getgo

[PabloTwo]

I did my 1st install of Linux (FC5) the other day from a DVD that accompanied a book checked out of the library. When Anaconda (the installer program) presented the graphical interface to begin the install process, the video screen turned to total garbage (nothing legible). I had to resort to a text only mode install, which was successful.

Once booted and logged into FC5, attempting to call up anything "graphical" in nature only gives me another garbage video screen. Is this worth attempting to fix from CLI mode? If so, what would be the steps to take? Or should I not waste time and just seek out a newer distro? Is this a known issue? I've scoured the forum in HARDWARE and GENERAL for any reference to this particular problem and found none.

Please keep any "do this" replies mindful of the fact I'm a pretty fresh noob in Linux and still feeling my way around the operating system.

----------------------------------------------
Asus A8N5X; 1GB DDR; nvidia 6800GS
Dual-boot: W2K/FC5
----------------------------------------------

Thanks, PabloTwo

[marko]

Fedora 6 would be a better choice, it's been out since October 2006
Even Fedora 7 is coming out May 24,2007.

Is the monitor very new? Xorg is the Xserver that does
the graphics in Fedora, it tries to match the monitor to
a database of known resolutions and frequencies it has
and/or does some other tests on the monitor to figure out
what resolution/refresh to use. Maybe your monitor post dates
that Xorg.

Another good option with Fedora 6 is a 're-spin' since
they are composed of the official FC6 release + any updates
up to that time of the "spin".

See http://fedoraunity.org/re-spins

Mark

[PabloTwo]

Thank you for you comments Marko

I realized after the fact that I should have made my original post in the INSTALLATION section, since I have since found more related posts there and found some things to try, so far without success. To address you primary point, my monitor is a Samsung SyncMaster 753DF, pretty mainstream stuff, and something less than 5 years of age.

I tried running the installer again with the cmd line args: mode=vesa resolution=800x600

Doing that at least enabled me to watch Anaconda up to the point it finished interogating the video card and monitor, at which point the screen then turned to gibberish.

I also peeked into the xorg.conf file and the video card and monitor entries look fairly plain vanilla, it at least recognized my video card as having an nvidia chipset and set the driver as "nv". It did not figure out what monitor I have though, wherein probably lays the problem. If I knew which entries to modify in xorg.conf, I could probably get the GUI to fly with intelligible video.

I may take your advice and send off for the latest 're-spin' of the FC6 release and wait for it to arrive in the mail. Downloading it is not an option for me with 28.8Kps dial-up.... hooboy!

Thanks again,
PabloTwo
Sevile, FL

[marko]

Pablo:

"nv" is the open, free software driver for the nvidia cards, since it was
partly reversed engineered with little help or guidance from
Nvidia it does have some problems (usually concerning
poorer performance than the closed proprietary driver)

you might want to try the closed nvidia driver:
see the instructions (FC5 specific) at:

http://www.mjmwired.net/resources/mj...c5.html#nvidia

There's a price to pay for the closed driver,it makes it a bit
less convenient to update the kernel. The nvidia driver
version has to be the same as the kernel (ie, the repository had
to have built the video driver with the same kernel version and set of headers
as your PC is using). This is why that URL above has clearly stated on it

"For nvidia kernel driver you MUST MATCH YOUR KERNEL"

This is pretty easy now in FC6 and later
because the package manager handles that pretty well but
I'm not sure FC5 would be as nice.

One good thing, since the nvidia.ko file is placed under the
/lib/modules/<kernelversion>/extra/nvidia directory, once
you install a driver you can still switch between kernels from
the grub menu at will without losing your Xwindows support.
This is because each kernel then has its own driver to call on.
I think a long time back nvidia put the driver in a single place
so only one kernel would work at a time (bad engineering design).

Mark

[PabloTwo]

Thank you once again Mark-

Obtaining and installing newer Nvidia drivers for my GeForce6800GS sounded like a swell idea. I followed the link you gave me in your previous post. Besides the nvidia drivers section there, there are bunches of other great "how to" stuff for FC5. But I'm running into the "chicken before the egg" kanundrum... in a big way. And it looking like I'm gonna have to kill a lot of chickens before I can get to that egg (installed nvidia drivers on my Linux OS).

All the instructions given assume one has a working internet connection on their Linux system. Mine does not. But I do have access from the W2K OS on the same machine. Should be simple, eh? Download the file(s) you need via Windows, transfer them over to your Linux install and have at it.

Since the nvidia driver file (I opted to get them from Nvidia as an .rpm package) is rather largish, almost 14MB, I also downloaded the .rpm file that would allow me mount an NTSF partition(s) on my specific version of Linux. I could copy that to a FD from Windows then load it into Linux from the FD.

After a bit of head bashing with failed attempts to "mount" my floppy, I finally tried sticking some media into the FD drive and tried again. Success. First chicken killed. Then carefully following instructions and typing, in the dir containing the .rpm file, and as root user:

rpm -ivh kernel-module-ntfs-2.6.15-1.2054_FC5-2.1.26-0.rr.10.0.i686.rpm

Gives me the error msg: :not an rpm mpackage (or package manifest):

I also tried copying the file from the floppy using the "mcopy" cmd, with the same results. Without being able to install this file, as far as I know, I can't mount the NTSF partition where the nvidia driver resides to import it into Linux. In the aftermath of the above, I was able to mount my CD/DVD drive and get it working.

Does passing the .rpm file through an NTSF partition then to a VFAT (floppy) corrupt the original file?
Would burning it onto a CD_ROM from the NTSF partition then into Linux from there, bypassing the floppy, prove any better? My knifes still sharp... gotta lotta chickens to kill here.

Paul
Seville FL

[clearer]

When it comes to nVidia drivers, you should really opt for the manual things here.
BUT -- there is no need to do all this just to get the graphical output working. If you just need graphics working (no 3d acceleration and it's stuck at 60Hz -- so find your eyes), you can simply edit /etc/X11/xorg.conf to use the vesa driver instead of the nv driver.

This will give you a 1024x768 resolution which should give some killed chickens. If your eth0 is recognized, setting it up should be easy after this.

Once done, make sure you have gcc installed and get the shell script drivers from nvidia.com -- no rpms from anywhere. Kill X (gah! now that it just got working) and install the drivers. Remember to actually switch to init 3, and not just kill X.

[marko]

getting the internet connection working on the original FC5 installation doesn't
need graphical support working at all. I'd think the easier way to go would be
to just get the internet connection configured, then use yum to install the kmod-nvidia
rpm (the proprietary nvidia driver) (yum is a command line CLI tool).
Mark

[PabloTwo]

Once again, thanks for everyones input here.

Clearer, yes, setting the driver to "vesa" in xorg.conf did get me a legible graphical screen when calling up "system-config-display" or "startx". That certainly helped to find my way around the various configuration options easier without having to figure which "file" to edit to change something. It only gives me a max resolution of 800x600 at 61hz refresh... sorta reminds me of the days back in Windows for WorkGroups 3.1.

As for getting my Linux install on the internet, eth0 is recognized, but that's not my path to internet connectivity. I have dial-up with a Winmodem. Yes, I've already researched and found a driver that would/should work with my Conexant chipset modem, but haven't downloaded and installed it yet.... another chicken or two kill there before I can use YUM.

And as for the rpm package not recognized problem stated in an earlier post: operator error. Downloaded that with the "downloader" in FireFox. It said I had the file 100% and closed the downloader window. It lied, and not for the first time. I only had about 25% of the file. When I downloaded it using a standalone download client I got the whole file, which installed just fine, which then allowed me to mount the NTSF partition and I can now easily import a file of any size into Linux after a download via Windows.

As for the Nvidia driver package I download, it clearly warned me that I needed a newer version kernel than I have to work right, but just for ha ha's, I ran the package anyway to see what color smoke would come out: "sh NVIDIA-Linux-x86-1.0.9755-pkg1.run". It quickly told me it couldn't find the binary utility "ld", that I needed to include it in the "path" (I understand that- I come from the days of DOS) and that the "binutils" package may not be installed, which it obviously isn't.

I'm going to keep hacking away at this, for nothing else as a "learning" experience, but if the obstacles keep piling up too much preventing me from getting from point A to point B with this Linux install, I can certainly see a shiny new DVD arriving in the mail with the latest version of FC6 in the not too distant future.

Thanks again
Paul
Seville FL

[marko]

Paul,

Back before I had broadband I was a fan of cheapbytes.com for
getting a disk in the mail. Just search "fedora" in the search tool
there.

Mark

[PabloTwo]

Mark,

Thanks, I'll check out cheapbytes.com. But the good news is that I'm almost there at getting the Nividia drivers ( NVIDIA-Linux-x86-1.0-9755-pkg1.run ) installed in FC5. FC6 may well be in my future, but I've got so much time invested in working through this video driver problem I'm now determined to either see it through to the end or splat against the brick wall.

In running the above mentioned package, it aborts telling me it couldn't find this or that so I go and fetch this and that and install it, or attempt to install it. Some of the rpm packages I'm trying to install fail due to "dependencies" not being present. So I copy down the list of dependencies and go fetch those. Eventually, I'll have everything installed the Nvidia installer needs to run to completion. If I had installed the Developers Software pkg at initial install of FC5 I probably wouldn't have had to run through this "go fetch" gaumet.

Paul

[PabloTwo]

OK.. spinning my wheels here going nowhere fast. I'm not as "almost" there as I thought. Here's the scenario, after having updated the kernel to 2.6.19-1.288.2.4.fc5.i686:

I "sh NVIDIA-Linux-x86-1.0-9755-pkg1.run"

Get install halt on error: "ld", part of the "binutils" pkg not found.
So I download and install binutils-2.16.91.0.6-5i386.rpm, which installs OK.

Run the NVIDIA driver pkg again

Get install halt on error: "cc", part of "gcc" pkg not found.
So I download and install gcc-4.1.1-51.fc5.rpm, which halts on error: dependencies not found
cpp =4.1.1-51.fc5
glibc-devel >=2.2.90-12
libgcc =4.1.1-51.fc5
libgomp =4.1.1-51.fc5

So I download all four of the above and attempt to install glib-dev-2.2.90-12.rpm first which
halts on error: dependencies not found
glibc =2.4.11
glibc-headers
glibc-headers =2.4-11

Well, you see the pattern here. It's obvious to me there must be a whole package I need to install which contains all this stuff and lots more and that will install via rpm without all the "sorry, can't install this cuz I can't find <whatever>". But I don't know what pkg name it would be.

Be the first to give me the correct package name to download and install and you'll win a free trip to Severodoneck, Russia.

Paul
Seville, FL

[lmo]

I have FC6 and can use yum.
I have installed these package groups and I installed nvidia driver for GeForce2.
yum groupinstall "Development Tools"
yum groupinstall "Development Libraries"
yum groupinstall "X Software Development"
I made sure to the kernel-devel matching the kernel (this has kernel source)

I documented so I could do it again here http://forums.fedoraforum.org/forum/...d.php?t=151907

This probably won't exactly help, unless there is some hidden clue.

yum groupinfo "Development Tools"
Loading "installonlyn" plugin
Loading "downloadonly" plugin
Setting up Group Process
Setting up repositories
comps.xml                 100% |=========================|  67 kB    00:17     

Group: Development Tools
 Description: These tools include core development tools such as automake, gcc, perl, python, and debuggers.
 Mandatory Packages:
   flex
   gcc
   redhat-rpm-config
   strace
   rpm-build
   make
   pkgconfig
   gettext
   automake
   gdb
   bison
   libtool
   autoconf
   gcc-c++
   binutils
 Default Packages:
   ltrace
   diffstat
   cvs
   texinfo
   automake15
   python-ldap
   subversion
   byacc
   frysk
   gcc-gfortran
   elfutils
   pfmon
   rcs
   automake16
   automake17
   automake14
   patchutils
   ctags
   oprofile-gui
   systemtap
   swig
   doxygen
   indent
   valgrind
   pstack
   oprofile
   cscope
 Optional Packages:
   epydoc
   smarteiffel
   dejagnu
   tolua++
   gcc-gnat
   regexxer
   cppunit
   imake
   srecord
   cvsweb
   flasm
   jam
   erlang
   rpmdevtools
   utrac
   Sprog
   darcs
   trac
   oorexx
   mercurial
   lightning
   xfce4-dev-tools
   meld
   colordiff
   clips-doc
   perl-perlmenu
   quilt
   ftnchek
   expect
   translate-toolkit
   tclpro
   sbcl
   yasm
   lua
   ccache
   clisp
   cvsgraph
   gforth
   monotone
   ecl
   asa
   ipython
   tkcvs
   bazaar
   abicheck
   ocaml
   tla
   sysconftool
   phpcs
   mcrypt
   curry
   pgadmin3
   cogito
   cvs2cl
   splint
   ElectricFence
   pscan
   cvs2svn
   cpan2rpm
   git
   gprolog
   kuipc
   yap
   cmucl
   rapidsvn
   pl
   cproto
   plt-scheme
   archmage
   pylint
   crossvc
   tkcon
   sysprof
   gauche
   sunifdef
   clips
   haddock
   rpmlint
   highlight
   cpanspec
   gphpedit
   xbsql
   lush
   nqc
   manedit
   eric
   hugs98
   ucblogo
   clips-xclips
   cmake
   fpc
   nasm
   gcl
   qgit
   scons
   aplus-fsf
   mlton
   mock
   gpsim
   patchy
   bigloo
   alleyoop
   perltidy
   memtest86+
   cvsps
   mach
   bzr
   svn2cl
   pikdev
   gcc-objc
   q
   ghc
   codeblocks
   lcov
   svnmailer
   gputils

I have tried going through that "one thing after another" pattern a few times before, and each time, I eventually gave up, bit the bullet, and started the latest distro.

[PabloTwo]

Thank you Imo-

The one aspect of my situation you missed from earlier in this thread is that I do not have any connection to the internet, yet, with my Linux install. I download whats needed in Windows with a slow dialup connection and put that in an NTFS partition that Linux can see and "cp" the file(s) into Linux from there. Therefore, using "yum" is not possible, so I'm downloading "rpm" files, so far, all from:

http://redhat.download.fedoraproject...pdates/5/i386/

I've looked in other directories there for something that would equate to "Development Tools" and "Development Libraries" but have found no rpm packages as such. It looks like all of the individual components that would be included in such packages are listed individually. I'm guessing I probably wouldn't need the "XSoftware Development" stuff in order to get the nvidia driver shell script installer to run without choking.

If anyone knows of a download source for the rpm packages of the Development Tools and Development libraries, please let me know.

Yes, this has been quite an "one thing after another" os install ordeal, and although I haven't given up on this just quite yet, yesterday, I did... Marko should smile here... order FC6 DVD install from cheapbytes.com.

Thanks all
Paul
Seville FL

[lmo]

Oh, ... I do not know about FC5, but with FC6, (I had a dvd and a cd from a book), I put in the dvd and the cd into /etc/yum.conf as repos something like this:

[main]
cachedir=/var/cache/yum
keepcache=1
debuglevel=2
logfile=/var/log/yum.log
pkgpolicy=newest
distroverpkg=redhat-release
tolerant=1
exactarch=1
obsoletes=1
gpgcheck=1
plugins=1
#metadata_expire=2592000
#metadata_expire=10
#metadata_expire=86400
#60*60*24=86400*30=2592000
#throttle=4.0k

# PUT YOUR REPOS HERE OR IN separate files named file.repo
# in /etc/yum.repos.d
reposdir=/etc/yum.repos.phony

[fc6-dvd]
name=FC6 i386 DVD
baseurl=file:/FC6DVD/
enabled=1
gpgcheck=0

Then, suppose I created a directory /FC6DVD and mounted the DVD on it like:
mkdir /FC6DVD
mount /dev/hdc /FC6DVD

Then ran "yum clean all" and "yum update".
The reposdir=phony stuff kept it from trying to go on the web, and I could yum install all day from the dvd.

However hard it is to get the DVD/CDs set up like a repository, it sure beats going round and around week in and week out.

I am trying to remember a quote JN4OldSchool had posted on fedora forum

JN's quote of the week: "friends did allured me that linux need never reinstall. just put commands in, then all things will get right."

[PabloTwo]

Wow! Very clever. I'll give that try when I have time and see if that also works with FC5 and if my Linux manipulation skill level is up to it when I hit the inevitable obstacles (remember, I'm a Linux noob here).

I tried "looking" at the FC5 install DVD from the GUI last night with something like that in mind but Gnome told me I didn't have "permissions" to "mount" the DVD drive, even though I was logged in as root. In fact, the only type of CD/DVD media Gnome would give me access to was an "audio" CD. All other media types (data CD/DVD; movie DVD; mpeg CD) was locked out with no permissions. Go figure. But I drift from the mission at hand here.

Paul
Seville, FL