Difference between revisions of "Fedora 8 Linux"
Marc Saric (talk | contribs) |
Marc Saric (talk | contribs) |
||
(18 intermediate revisions by the same user not shown) | |||
Line 52: | Line 52: | ||
PPPoE could be configured on the commandline with | PPPoE could be configured on the commandline with | ||
− | < | + | <syntaxhighlight lang="bash"> |
pppoe-setup | pppoe-setup | ||
− | </ | + | </syntaxhighlight> |
Depending on your setup, you also need to disable the conventional eth0/eth1 autoconfig, or the boot-up time will increase by the time needed to complete the timeout for the DHCP-configuration. | Depending on your setup, you also need to disable the conventional eth0/eth1 autoconfig, or the boot-up time will increase by the time needed to complete the timeout for the DHCP-configuration. | ||
Slightly stupid, but... | Slightly stupid, but... | ||
+ | |||
+ | === Ethernet problems with Davicom adapters === | ||
+ | |||
+ | After installing Fedora 8 on my old 700 MHz ADM Duron, I failed to get the Ethernet card up and running. It is a cheap Davicom based model: | ||
+ | |||
+ | <syntaxhighlight lang="bash"> | ||
+ | [root@hardy ~]# lspci | ||
+ | ... | ||
+ | Ethernet controller: Davicom Semiconductor, Inc. 21x4x DEC-Tulip compatible 10/100 Ethernet (rev 31) | ||
+ | </syntaxhighlight> | ||
+ | |||
+ | http://ubuntuforums.org/showthread.php?t=186430, https://bugzilla.redhat.com/show_bug.cgi?id=236928 and http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=405203 have hints, what might be wrong here. I did as advised and blacklisted the tulip-driver but to no effect. | ||
+ | |||
+ | On startup, the <tt>dmfe</tt> driver still could not initialize the card. | ||
+ | |||
+ | Looking at <tt>dmesg</tt> showed the following entries: | ||
+ | |||
+ | <syntaxhighlight lang="bash"> | ||
+ | ... | ||
+ | dmfe: Davicom DM9xxx net driver, version 1.36.4 (2002-01-17) | ||
+ | ... | ||
+ | dmfe: Change Speed to 100Mhz half duplex | ||
+ | ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready | ||
+ | ... | ||
+ | </syntaxhighlight> | ||
+ | |||
+ | |||
+ | It does work after a manual <tt>ifdown</tt>/<tt>ifup</tt>, so it seems to be two problems, one was the tulip driver messing things up, the second beeing the dmfe driver having problems with the network card switching modes (to 100 MBit half duplex actually). | ||
+ | |||
+ | The error message on startup reads: | ||
+ | |||
+ | <syntaxhighlight lang="bash"> | ||
+ | ...failed; no link present. Check cable? | ||
+ | </syntaxhighlight> | ||
+ | |||
+ | This can be found in <tt>/etc/sysconfig/network-scripts/ifup-eth</tt>, which uses <tt>check_link_down</tt> in <tt>network-functions</tt> in the same directory. | ||
+ | |||
+ | To solve the problem, increase the delay until the function fails: | ||
+ | |||
+ | <syntaxhighlight lang="bash"> | ||
+ | check_link_down () | ||
+ | { | ||
+ | if [ -x /sbin/mii-tool -o -x /sbin/ethtool ]; then | ||
+ | if ! LC_ALL=C ip link show dev $1 2>/dev/null| grep -q UP ; then | ||
+ | ip link set dev $1 up >/dev/null 2>&1 | ||
+ | fi | ||
+ | timeout=0 | ||
+ | # delay=10 | ||
+ | delay=30 | ||
+ | [ -n "$LINKDELAY" ] && delay=$(($LINKDELAY * 2)) | ||
+ | while [ $timeout -le $delay ]; do | ||
+ | ... | ||
+ | </syntaxhighlight> | ||
== nVidia driver == | == nVidia driver == | ||
Line 68: | Line 121: | ||
</pre> | </pre> | ||
− | == nVidia driver for the nVidia GeForce 8800GT == | + | === nVidia driver for the nVidia GeForce 8800GT === |
Recently I bought a new rig, which got the (at the time of writing) brand-new nVidia GeForce 8800GT graphics board. nVidia states on their website, that this chip is supported under Linux by their newest graphics driver. Unfortunately, the livna.org package of that driver does not recognize the chip and fails to start the X-server. | Recently I bought a new rig, which got the (at the time of writing) brand-new nVidia GeForce 8800GT graphics board. nVidia states on their website, that this chip is supported under Linux by their newest graphics driver. Unfortunately, the livna.org package of that driver does not recognize the chip and fails to start the X-server. | ||
Line 74: | Line 127: | ||
To install a working graphics driver for the nVidia GeForce 8800GT you neet to enable livna-testing: | To install a working graphics driver for the nVidia GeForce 8800GT you neet to enable livna-testing: | ||
− | <bash> | + | <syntaxhighlight lang="bash"> |
cd /etc/yum.repos.d | cd /etc/yum.repos.d | ||
vi livna-testing.repo | vi livna-testing.repo | ||
− | </ | + | </syntaxhighlight> |
edit the line | edit the line | ||
− | <bash> | + | |
+ | <syntaxhighlight lang="bash"> | ||
enabled=0 | enabled=0 | ||
− | </ | + | </syntaxhighlight> |
to | to | ||
− | <bash> | + | <syntaxhighlight lang="bash"> |
enabled=1 | enabled=1 | ||
− | </ | + | </syntaxhighlight> |
after that, install the testing nVidia driver from livna as usual, and everything should work. | after that, install the testing nVidia driver from livna as usual, and everything should work. | ||
+ | |||
+ | === nVidia driver for legacy nVidia cards (GeForce 2MX, NV11) === | ||
+ | |||
+ | Old nVidia cards are supported through legacy drivers (see http://www.nvidia.com/object/unix.htm for details). | ||
+ | |||
+ | To install the legacy driver for the latest Fedora 8 kernel, do the following: | ||
+ | |||
+ | * Enable the livna repository as above | ||
+ | * Run the following command as root | ||
+ | |||
+ | <syntaxhighlight lang="bash"> | ||
+ | yum install kmod-nvidia-legacy | ||
+ | </syntaxhighlight> | ||
== CD-RW/DVD+-RW == | == CD-RW/DVD+-RW == | ||
Line 109: | Line 176: | ||
http://register.vmware.com/content/registration.html | http://register.vmware.com/content/registration.html | ||
* Install (I used the rpm) | * Install (I used the rpm) | ||
− | <bash> | + | <syntaxhighlight lang="bash"> |
sudo rpm -Uhv VMware-server-1.0.4-56528.i386.rpm | sudo rpm -Uhv VMware-server-1.0.4-56528.i386.rpm | ||
sudo rpm -Uhv VMware-server-console-1.0.4-56528.i386.rpm | sudo rpm -Uhv VMware-server-console-1.0.4-56528.i386.rpm | ||
− | </ | + | </syntaxhighlight> |
+ | |||
* The standard setup-script fails to compile the kernel module, so you need to use the following (unsupported) patch: | * The standard setup-script fails to compile the kernel module, so you need to use the following (unsupported) patch: | ||
− | <bash> | + | <syntaxhighlight lang="bash"> |
wget http://knihovny.cvut.cz/ftp/pub/vmware/vmware-any-any-update115.tar.gz | wget http://knihovny.cvut.cz/ftp/pub/vmware/vmware-any-any-update115.tar.gz | ||
tar xvzf vmware-any-any-update115.tar.gz | tar xvzf vmware-any-any-update115.tar.gz | ||
cd vmware-any-any-update115 | cd vmware-any-any-update115 | ||
./runme.pl | ./runme.pl | ||
− | </ | + | </syntaxhighlight> |
The rest is the standard procedure for a VMware installation. | The rest is the standard procedure for a VMware installation. | ||
* And voila, it works. | * And voila, it works. | ||
Line 140: | Line 208: | ||
== Java 6 does not work any more == | == Java 6 does not work any more == | ||
− | One of the recent update orgies fucked up all -previously working- | + | One of the recent update orgies fucked up all -previously working- Java GUI apps. |
− | Upon startup, you will get a stacktrace stating something like this: | + | Upon startup, you will get a stacktrace stating (among a lot of other stuff) something like this: |
− | < | + | <syntaxhighlight lang="bash"> |
fedora 8 java: xcb_xlib.c:82: xcb_xlib_unlock: Assertion `c->xlib.lock' failed | fedora 8 java: xcb_xlib.c:82: xcb_xlib_unlock: Assertion `c->xlib.lock' failed | ||
− | </ | + | </syntaxhighlight> |
+ | |||
+ | |||
This is due to an updated <tt>libxcb</tt>, which somehow worked before. | This is due to an updated <tt>libxcb</tt>, which somehow worked before. | ||
Line 155: | Line 225: | ||
− | <bash> | + | <syntaxhighlight lang="bash"> |
yum --enablerepo=development install libxcb.i386 | yum --enablerepo=development install libxcb.i386 | ||
− | </ | + | </syntaxhighlight> |
There is still a stacktrace, but it seems to work... | There is still a stacktrace, but it seems to work... | ||
+ | |||
+ | Great, if one can be shure, that software get's tested even against esoteric fringe language runtimes like SUN Java. | ||
= Conclusions = | = Conclusions = | ||
− | Fedora 8 is a nice and stable system, which installs without problems (almost boring ;-) ). The new design (icons, wallpaper, colour scheme) is also quite nice (except for the window borders I'd say). Great work, Red Hat! | + | Fedora 8 is a nice and stable system, which installs without problems (almost boring ;-) ). The new design (icons, wallpaper, colour scheme) is also quite nice (except for the window borders I'd say). |
+ | |||
+ | Some of the glitches which crept in with updates could have been caught by better testing, but still: | ||
+ | |||
+ | Great work, Red Hat! |
Latest revision as of 21:10, 20 January 2020
Installation notes for Fedora 8 on my AMD Athlon64 3200+.
The most recent version of Fedora looked to good to stay with the old one.
Contents
Download
Via one of the ftp mirrors. Burn to DVD, done.
Basic install
As with my previous installation, I did not do an update but a complete reinstall. I just scraped the old FC6 installation, keeping the home-directories and manually installed software.
There are tons of installation instructions out there, so I will only list the stuff, which did not work out of the box.
Tweaking
Serial mouse
If you use one of the (very ancient) serial mouses, you may be in trouble with the default X-Window installation. Fedora 8 does not recognize it automatically.
Here's a manual config, which should work:
Edit /etc/X11/xorg.conf
Add the following lines
InputDevice "Mouse0" "CorePointer"
to the Section "ServerLayout".
Add the following config block below.
Section "InputDevice" Identifier "Mouse0" Driver "mouse" Option "CorePointer" Option "Device" "/dev/ttyS0" Option "Protocol" "Microsoft" EndSection
afterwards restart your X-Server. It should work now.
Network
Currently, my box runs with a direct connection to a PPPoE-modem, a fact, that was not recognized by the installer (altough I guess it is not that uncommon). Installation had to be finished without online-connection therefore.
PPPoE could be configured on the commandline with
pppoe-setup
Depending on your setup, you also need to disable the conventional eth0/eth1 autoconfig, or the boot-up time will increase by the time needed to complete the timeout for the DHCP-configuration.
Slightly stupid, but...
Ethernet problems with Davicom adapters
After installing Fedora 8 on my old 700 MHz ADM Duron, I failed to get the Ethernet card up and running. It is a cheap Davicom based model:
[root@hardy ~]# lspci
...
Ethernet controller: Davicom Semiconductor, Inc. 21x4x DEC-Tulip compatible 10/100 Ethernet (rev 31)
http://ubuntuforums.org/showthread.php?t=186430, https://bugzilla.redhat.com/show_bug.cgi?id=236928 and http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=405203 have hints, what might be wrong here. I did as advised and blacklisted the tulip-driver but to no effect.
On startup, the dmfe driver still could not initialize the card.
Looking at dmesg showed the following entries:
...
dmfe: Davicom DM9xxx net driver, version 1.36.4 (2002-01-17)
...
dmfe: Change Speed to 100Mhz half duplex
ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
...
It does work after a manual ifdown/ifup, so it seems to be two problems, one was the tulip driver messing things up, the second beeing the dmfe driver having problems with the network card switching modes (to 100 MBit half duplex actually).
The error message on startup reads:
...failed; no link present. Check cable?
This can be found in /etc/sysconfig/network-scripts/ifup-eth, which uses check_link_down in network-functions in the same directory.
To solve the problem, increase the delay until the function fails:
check_link_down ()
{
if [ -x /sbin/mii-tool -o -x /sbin/ethtool ]; then
if ! LC_ALL=C ip link show dev $1 2>/dev/null| grep -q UP ; then
ip link set dev $1 up >/dev/null 2>&1
fi
timeout=0
# delay=10
delay=30
[ -n "$LINKDELAY" ] && delay=$(($LINKDELAY * 2))
while [ $timeout -le $delay ]; do
...
nVidia driver
Via livna as described on http://www.mjmwired.net. I had to add the noapic option to make it run (as per the nVidia manual) to /etc/grub.conf.
kernel /boot/vmlinuz-2.6.23.1-49.fc8 ro root=LABEL=/ rhgb quiet noapic
nVidia driver for the nVidia GeForce 8800GT
Recently I bought a new rig, which got the (at the time of writing) brand-new nVidia GeForce 8800GT graphics board. nVidia states on their website, that this chip is supported under Linux by their newest graphics driver. Unfortunately, the livna.org package of that driver does not recognize the chip and fails to start the X-server.
To install a working graphics driver for the nVidia GeForce 8800GT you neet to enable livna-testing:
cd /etc/yum.repos.d
vi livna-testing.repo
edit the line
enabled=0
to
enabled=1
after that, install the testing nVidia driver from livna as usual, and everything should work.
nVidia driver for legacy nVidia cards (GeForce 2MX, NV11)
Old nVidia cards are supported through legacy drivers (see http://www.nvidia.com/object/unix.htm for details).
To install the legacy driver for the latest Fedora 8 kernel, do the following:
- Enable the livna repository as above
- Run the following command as root
yum install kmod-nvidia-legacy
CD-RW/DVD+-RW
FC3 and FC6 had issues with my Plextor-PX712A, which are really solved with the new version -finally!
VMware server
If you don't want to fumble around with Xen, VMware Server is a free (but not Open Source) alternative to run a virtualized OS (let's say that thing from Redmond).
Unfortunately, VMware does not support the current kernel version, so here is what I did:
(You need a full gcc software development environment and kernel headers for this to work)
- Download VMware from
http://www.vmware.com/download/server/
- Do not forget to register to obtain a (free) license key:
http://register.vmware.com/content/registration.html
- Install (I used the rpm)
sudo rpm -Uhv VMware-server-1.0.4-56528.i386.rpm
sudo rpm -Uhv VMware-server-console-1.0.4-56528.i386.rpm
- The standard setup-script fails to compile the kernel module, so you need to use the following (unsupported) patch:
wget http://knihovny.cvut.cz/ftp/pub/vmware/vmware-any-any-update115.tar.gz
tar xvzf vmware-any-any-update115.tar.gz
cd vmware-any-any-update115
./runme.pl
The rest is the standard procedure for a VMware installation.
- And voila, it works.
Google Earth font problems
The new Google Earth beta 4.3 seems to have problems with true type fonts. Here's how you can solve that problem:
- Locate the installation directory of your Google Earth installation
- Make a backup of the following files
libQt3Support.so.4 libQtGui.so.4 libQtSql.so.4 libQtCore.so.4 libQtNetwork.so.4 libQtXml.so.4
- Install the qt4 package with sudo yum install qt4
- Link the files listed above from /usr/lib to your Google Earth installation dir.
Now you should have more nice looking font rendering (especially in the Wikipedia popups).
Java 6 does not work any more
One of the recent update orgies fucked up all -previously working- Java GUI apps.
Upon startup, you will get a stacktrace stating (among a lot of other stuff) something like this:
fedora 8 java: xcb_xlib.c:82: xcb_xlib_unlock: Assertion `c->xlib.lock' failed
This is due to an updated libxcb, which somehow worked before.
Read this for some background and this for a sollution, which worked for me.
Which boils down to
yum --enablerepo=development install libxcb.i386
There is still a stacktrace, but it seems to work...
Great, if one can be shure, that software get's tested even against esoteric fringe language runtimes like SUN Java.
Conclusions
Fedora 8 is a nice and stable system, which installs without problems (almost boring ;-) ). The new design (icons, wallpaper, colour scheme) is also quite nice (except for the window borders I'd say).
Some of the glitches which crept in with updates could have been caught by better testing, but still:
Great work, Red Hat!