Activity Log - here are the steps I took for the Dell Inspiron 8200

Activity Log - here are the steps I took for the Dell Inspiron 8200

Top Specs Log References

First Steps

Thu Feb 27 17:41:59 EST 2003

Booted up under Windows XP SP1 (5.1.2600) - no network connection. Had to answer anoying questions.

Defragged drive. Thought this might help in the next part.

Examined control panel to learn about the drivers

System froze while looking at the control panel. Could not move mouse. Could not use Control-Alt-Delete. Had to power cycle.

Windows is suppose to be the most stable version of the OS, and it freezes while listing the devices in the system. Sheesh. I hope this isn't a hardware problem.

I wonder if the comments in Michael Jennings's Paper are true, especially those where Microsoft purposely cripples Windows XP unless you give up your privacy.

Rebooted

Examined hardware list and documented it here

Booted up RedHat (Really Pink Tie)

Had to change the BIOS to allow this.

Got to the point where I was reformating the disk. I wanted Dual Boot, and while I could wipe out Windows XP and reinstall, I looked for other options. I grabbed ntfsresize (rev v1.7.1s), untarred it, and did
mcopy ntfsresize a:
to put it on a floppy on another system. Then I went back to my new system, did a [Control][Alt][F2] to get a prompt. I then typed
cd /tmp
mcopy a:ntfsresize .
chmod +x ntfsresize
./ntfsresize -i /dev/hda2
It told me I could shrink the disk partition to 3.7GB. I decided to shrink it to 8 GB. (Correction added March 18 2003: Szakacsits Szabolcs told me I must have made a mistake in my notes. It's possible because I could not cut and paste. This is what I thought I did) I typed
./ntfsresize -n -s 8G /dev/hda2
to do a test run, then
./ntfsresize -s 8G /dev/hda2
It said
Calculating smallest shrunken size supported.... The volume end is fragmented, this case is not supported. Defragment it (windows 2000, SP, and .NET have built in defragmentation tool) and try again.

Since I had just defragmented the disk, I don't understand why it says I have to defragment it again. I did notice that when I did this, the upper end of the disk showed a bar on the graph and it is color codes as "unmovable files.". Perhaps Windows XP purposely places a marker at the end of the partition?
I guess I will have to repartion the disk and re-install windows. Or else buy Partiton Magic. To be honest, I bought an earlier version of Partition Magic, found it unusable because it didn;t support FAT32 file systems, and it took 3 years before they added the functionality needed. I'm not real excited about this product. Since this is a brand-new machine, I might just reformat the disk.

(Here is the correction - March 17, 2003 - . Szaka said the error I got was impossible. The other mistake I did was to run the defragmenter before I ran ntfsresize. Szaka said the documentation will be updated to show this change. The suggested procedure is to try things in this order:
ntfsresize
defrag
ntfsresize
fdisk
I hope this helps. Check Szaka's web site site for updates in documentation.) One other thing - the Linux-Dell-Laptop's FAQ mentioned the need of a partition for the Suspend-to-Disk function. They say that either phdisk.exe of mks2d.exe is used under Windows. I was unable to find either file. But I did notice that there is no partition dedicated to the Suspend-to-disk function. Perhaps when I re-partition the system, I might add a partition for this.
I should also document that the system has two partitions:

  • 31 MB Fat Partition(EISA Configuration)
  • 55.85 GB NTFS

    Well, I think I'm going to call it quits for tonight. Tomorrow I'll reformat the disk and wipe out Windows XP.

    Fri Feb 28 07:48:36 EST 2003

    Added page counter to the top page. Had to re-write major parts of this page to do so. Is it worth it?

    Fri Feb 28 23:13:19 EST 2003

    Later that day I booted into Windows XP. If I'm going to wipe it out, I want to make sure I've gotten the info I need. I also searched for the Suspend-2-Disk application. Whiling the system was searching, Windows XP - that most stable OS yet, froze again. Sigh. I guess I won't mind wiping it out.

    Besides - everyone should know how to re-install windows. SHOW NO FEAR

    Partitioning the disk

    Now I have to decide on the partitioning. The current partition is (according to Disk Druid) is:

    /dev/hda (Geom: 7296/255/63) (Model Fujitsu MHS2060AT)
    /dev/hda1vfat31MBstart: 1, End: 4
    /dev/hda2NTFS57192 MBstart: 5, End: 7295
    FREEFree Space8 MBstart: 7296, End: 7296

    I'm going to switch to FDISK, so I can define the mount points.

    (Press Back)

    The Suspend-To-Disk partition, which doesn't seem to exist, should be /dev/hda1 starting on cylinder 1, according to Linux-Dell-Laptop's FAQ. Shrug. Since I'm deleting all of the partitions, I might as well create one. It should be larger than memory + video RAM. I have 128MB now, but I may expand to the max of 1 GB. So I'll use 1.1GB for the dize of the first partition.

    The next partition will be Windows. I'll reserve 8 GB for that.

    Swap will be 2GB - twice max memory.

    /boot is recommended - to protect the file system while problems booting up.

    /var will be 1GB. I'm using a separate partition because if a denial of service attack fills up my logs, the rest of the system will still run. And vice versa.

    / - this should be big enough for any current and future RedHat distribution. I'll select 10 GB.

    /spare - I'll make this 10 GB just to have a spare partition for CFS.

    /home - this is very important. Many suggest a single partition for root and user files. I consider this idiotic. I've started a RedHat upgrade, and the upgrade failed. I was left with an unbootable system. The only choice I had was to re-install RedHat, ignoring the upgrade option. This wipes everything out. But because I had a separate /home partition, all of my files were still there. All I had to do is to re-configure the OS.

    Along those lines, I create a symbolic link of /local -> /home/local - and all of the software I install goes into /local. Note that this is separate from /usr/local - which is where many RPM's install files. I want to have a separate space for my libraries and software. I can always use /usr/local/lib, but I am not forced to do so.

    Using fdisk, I list the current partitions

  • 255 heads
  • 63 sectors
  • 7296 cylinders
  • Units=cylinders of 16065*512 bytes
  • /tmp/hda1 start:1 End: 4 Blocks: 32098+ ID: de Dell Utility
  • /tmp/hda2 start:5 End: 7295 Blocks: 58564957+ ID: 7 HPFS/NTFS

    Well, now it's time to commit. When I do this, I can't go back.

    Each cylinder is 8032.5K or 8.0325 MB. (16065*512/1024)

    When selecting partitions, it makes sense to make them a multiple of the cylinder size to make sure there are no wasted blocks at the end of a partition.

    I wrote a perl program to calculate the size for every even cylinder boundary:

    The Perl Program


    #!/local/bin/perl
    my $i;
    my $cyl = (16065*512)/1024;
    printf("%s\t%s\t%s\n", "Cyl", "MB", "GB");
    for ($i = 1; $i<=7296; $i++) {
    printf("%d\t%6.1f\t%6.4f\n",$i,($i*$cyl/1024), ($i*$cyl)/(1024*1024));
    }

    Maping Gigabytes to cylinders

    Here is an extract of the results: 64
    CylMBGB
    502.00.4903
    1281004.10.9805
    1321035.41.0112
    1441129.61.1031
    2622055.22.0070
    10468205.18.0128
    130610244.610.0045
    So I want partitions of size
    S2d 144
    NTFS 1046
    /boot 64
    / 1306
    SWAP 262
    /var 144
    /spare 1306
    /home - the rest

    Partition Plan

    Let's see how I do it. I've got this:
    /tmp/hda1 1-143 OS/2 Hidden C: Drive
    /tmp/hda2 144-1189 NTFS Volume Set
    /tmp/hda3 1190-1253 Linux /boot
    /tmp/hda4 1254-7296 Extended
    /tmp/hda5 1254-2559 Linux /
    /tmp/hda6 2560-2821 Linux Swap
    /tmp/hda7 2822-2965 Linux /var
    /tmp/hda8 2966-4271 Linux /spare
    /tmp/hda9 4272-7296 Linux /home

    Well, I completed the partitioning. I then continued to Disk Druid where I named the partitions. I also said to check for bad blocks. I then selected the packages, (most of them. I can always delete them later) and told Red Hat to go crazy.

    NOTE: I should have put my swap partition near the beginning of the disk for better performance. However, I might still to that. I can see if I can exchange the swap and suspend partitions.

    This will take a while. That's good enough for tonight.

    Installing RedHat

    Sat Mar 1 07:30:21 EST 2003

    Booted RedHat

    Yahoo! Xwindows, and sound seems to work.

    There was an error during the boot process in ./drivers/block/xd.o: init_module: Operation not permitted.

    I have not connected to the Internet yet. Last time I did a major upgrade, a hacker broke into my system while I was downloading the patches. Therefore I already have the patches on a CD I burned. I'll into patches from there before I connect.

    I went to the RPMS/i386 directory and did a
    rpm -Fvh *.rpm
    I did the same for the other directories as well.

    Dang. I jarred the system (it fell an inch) and the mouse stopped working.

    I hope this doesn't mean the hardware is flaky. perhaps that's why Windows was misbehaving.

    Create First File

    Something I do when I install a new system is to create a file after the install is done. This way I know what files I've modified after the install process. I can use find -newer to identify these files. So I executed the script

    for i in / /boot /var /spare /home

    do

    date >$i/.FIRSTFILE

    done

    Output of DF

    The output of df gives:
    Filesystem1K-blocksUsed Available Use% Mounted on
    /dev/hda5103257482614860718636827%/
    /dev/hda3497861132584588993%/boot
    /dev/hda92391647632872226686921%/home
    none631960631960%/dev/shm
    /dev/hda8103257483283297683961%/spare
    /dev/hda711384287708810035088%/var

    Linux Test

    Sat Mar 1 11:18:13 EST 2003

    Did a quick check of Linux. Ran tuxracer. It was godawful slow. Completely unplayable. Work has to be done.

    Reinstalled Windows

    The nice thing about Dell is that they provide CD's to re-install all or some of the software. My old Thinkpad had a CD that essentially re-wrote the entire disk. There was no way to re-install just the OS. Won't buy another IBM laptop unless I know this is no longer the case.

    Installed Windows XP from CD. I apparently didn't select the right type of partition for the 8GB hda2 partition. I used the boot process to delete and create a partition in the same slot. Windows installed no problem. It did say that WIndows would mess up the boot loaded.

    I then installed the Dell drivers.

    Rebooted using the Linux CDROM, using "linux rescue"
    Got the prompt, and typed
    chroot /mnt/sysimage
    grub-install '(hd0)'

    Rebooted under Linux.

    Removed all unnecessary services, including nfs, portmap, atd

    Connected to network first time

    Grabbed latest copy of

  • the Phoenix browser
  • NVIDIA drivers
  • snort intrusion detection system

    NVIDIA drivers

    grabbed NVchooser.sh. Ran it. According to what it says, I need

    NVIDIA_GLX-1.0-4191.tar.gz

    NVIDIA_kernel-1.0-4191.tar.gz

    and the README file

    Installed Intrusion Detection System

    Also - the Suspend-to-disk worked. I pressed "function-(esc)Suspend" and the system worked. But I hear that the new vidia driver causes problems.

    Rebuild kernel

    Next step is to rebuild the kernel with no changes to see if it rebuilds with no errors. Once this is done, I'll install the patches for the NVIDIA card..

    Typed
    cd /usr/src/linux-2.4
    make mrproper
    cd configs/kernel-2.4.18.i686.config .config
    make dep
    make
    make bzImage
    make modules

    All looks well, so I did a
    make modules_install

    Then if everything builds with no errors, do a (big breath)
    make install

    Then reboot. If this doesn't work, you have problems. Sorry. But grub shows both the new and the old kernel when you book up. So perhaps you can still go back to the old kernel.

    I also installed
    tpconfig
    gkrellm
    i8kutils-1.8
    i8krellm-2.3

    I installed the NVIDIA files from the tar directory. This was easy. I did a make and make install

    I then followed the documentation, changed "vesa" to "nvidia" and deleted the line that says
    Load "dri"

    Do this when you are not in X windows. I restarted the X driver, and the tried tuxdriver and it worked.

    The i8krellm plugin for gkrellm is great. However, you have to specify "/usr/bin/i8kfan" in the GUI options. I thought it would use that as the default. But once I did, the system started running the fan every 5 seconds to cool down the laptop. I was concerned about this because the laptop was HOT and it's still winter. Without the fan control, it may be too hot to use. Gkrell has a temperature strip and so far the temperature is off the scale. It started at 150 degrees Fahrenheit. The second bug I noted is that the default unit is Fahrenheit in the graph, but the temperatures in the GKRELLM plugin are in Centigrade. The temperature ought to change to match the units of the scale. What happened was that while I had selected Fahrenheit as a graph, the fan was always running to get the value down to 50 degrees Fahrenheit. Obviously it never could.

    Had to download lameOnce I did that, I could use grip to rip an album. But xmms doesn't play the tracks ripped.

    I tried suspend-to-disk with the new NVIDIA driver. Failed. I followed the directions for modifying XF86Config. Not good enough. I modified the source of the nv.c code. There was a chunk of code that had "#if 0" - I changed the 0 to a one - no good. So I changed it back, and underneath it I changed the "return 1;" to "return 0;" Suspend to disk seemed to work. I tried both "apm -s" and [Function]-[ESC/Suspend] and this worked. However, I tried it again, and left the machine suspended overnight. It would not unsuspend.

    Mon Mar 3 07:50:57 EST 2003

    I followed the instructions for getting apt setup using Robert C Dowdy's Page

    I then did a apt-get install xmms-mp3

    Tue Mar 4 07:59:42 EST 2003

    Got grip to rip and encode mp3's, so that xmms could play it. The part that needed changing was the calling to LAME. The default filename extension was *.ogg. I had to change this to *.mp3. I tried this as non-root, and I had to make sure my username belonged to group "disk" for cdparanoia to run properly.

    Learned about the wonderful fact that INTEL/Microsoft has undocumented mechanism to slow down a laptop's CPU to extend battery life. Also learned that Microsoft doesn't have it working on Dells with Windows XP. Also Linux kernel 2.4 may also not have it working properly. This is just wonderful. Oe buys 1 2.4GHZ CPU but only gets 1.2GHZ performance. Oh joy.

    Downloaded the unix benchmark "bench" and bonnie.c to test CPU and disk performance.

    Thu Mar 6 07:07:02 EST 2003

    Tried suspend-to-disk at run-level 3 (No X-Windows). System failed to come up. Console was filled with usb-uhci errors.

    I installed CPUfreq - or rather tried to. I did this to find out what the CPU speed was, and if speedstep is causing a problem. However, I was not successful. "make modules" failed because "driver" was not defined.

    Thu Mar 13 22:02:15 EST 2003

    Installed new sendmail and snort. Tried to get cpufreq working. However - it didn't work with kernel 2.4.18 I grabbed 2.4.20 - which is the latest stable kernel. I then build it using

    make xconfig

    make dep

    make

    make bzImage

    make modules
    got an error in building the modules, especially the SCSI driver (fake7.c) script_asm.pl : Illegal combination of registers in line 72 : MOVE CTEST7 & 0xef TO CTEST7
    I installed all of the compat-gcc files, as this error is caused by the gcc-3.2.7 compiler, while the kernel needs the gcc-2.96.xx compiler for stability. So I went to the top Makefile, and changed 'gcc' to 'gcc296' and did make clean. The line in the Makefile was changed from

    CC = $(CROSS_COMPILE)gcc

    to

    CC = $(CROSS_COMPILE)gcc296
    This didn't work.

    Fri Mar 14 22:16:04 EST 2003

    Evenually I debugged the script_asm.pl, and I think it was a bug in perl. I edited /usr/src/linux-2.4.20/drivers/scsi/script_asm.pl and changed the two lines:
    $dst_reg = "\U$1\E";
    $rest = $2;

    and changed them to

    $dst_reg = $1;
    $rest = $2;
    $dst_reg =~ tr/a-z/A-Z;

    The first is suppose to take the matched reserved word $1, which is CTEST7 in this case, and make it upper case, which it already is. But for some strange reason, instead of changing CTEST7 to CTEST7, (duh!) $dst_reg has the new value "COB5^@^@". The ^@ is a null character. I was using perl 5.8.0, which is the latest stable version. I'll have to report this to the perl bug list. Very strange. But once I made this change, "make modules" worked fine.
    cd /usr/src/linux-2.4.20
    make
    make bzImage
    make modules
    make modules_install
    cp arch/i386/boot/bzImage /boot/vmlinuz-2.4.20
    mkinitrd /boot/initrd-2.4.20.img 2.4.20
    cp System.map /boot/System.map-2.4.20

    The add to /etc/grub.conf

    title New Linux (2.4.20)
    root (hd0,2)
    kernel /vmlinuz-2.4.20 ro root=LABEL=/ hdb=ide-scsi
    initrd /initrd-2.4.20.img


    After the new kernel, I had to reinstall the video drivers. Because I built the kernel with gcc296, I have to add
    CC=gcc296
    to both NVIDIA make files. I had to then do a
    make clean
    make
    make install
    in both NVIDIA directories.

    Mon Mar 17 07:19:38 EST 2003

    Back to the cpufreq. Make sure you have the right patch. Use cpufreq-2.4.20-9. Save the file, cd to /usr/src/linux-2.4.20; then do a "patch -p1 Then to a make xconfig.
    afterwards, make sure there is, in .config
    CONFIG_CPU_FREQ=y
    CONFIG_CPU_FREQ_24_API=y
    Then do the make dep; make; etc. again. Don't forget to reapply the NVIDIA patches.
    Problem. Let me try again to make sure. I went to RPMseek to find a suitable kernel in RPM format. I found kernel-2.4.20-2.54.src.rpmkernel-2.4.20-2.54.src.rpm and downloaded this one.

    Sun Mar 23 07:35:34 EST 2003

    RECAP of the cpufreq kernel situation



    I was determined to see if I had a crippled machine caused by speedstep technology. I was concerned about all the problems people were having under Windows. If there are problems under windows, then is is likely problems under Linux. If my 2.4Ghz machine was performing at 1.2Ghz, I wanted to know.
    I grabbed a 2.4.20 kernel (kernel-2.4.20-ac2.2.src.rpm) and compiled it, fine. I added the cpufreq patches finally, got it to compile. However, no matter what I did, the interface to the speedstep info did not show up in the /proc/sys/cpu psuedo-device. That entry in the /prof filesystem was missing. I compiled it several times, used gcc296 and gcc, neighter made a difference.
    So I went to rpmseek.com and searched for a newer kernel. I grabbed kernel-2.4.20-2.54.src.rpm. I did

    rpm -ivh kernel-2.4.20-2.54.src.rpm
    cd /usr/src/redhat/SPECS
    rpmbuild -ba kernel-2.4.spec
    I got several dependancy errors. I did a

    apt-get -f upgrade
    And tried again. This worked. I then typed

    cd /usr/src/redhat/RPMS/i386
    rpm -i kernel-source-2.4.20-2.54.i386.rpm
    I did a "make xconfig" "make dep" and "make" and all went well. When I did a "make modules" I got several strange errors.
    make[2]: Entering directory `/usr/src/linux-2.4.20-2.54/drivers/atm'
    gcc -D__KERNEL__ -I/usr/src/linux-2.4.20-2.54/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -fomit-frame-pointer -pipe -mpreferred-stack-boundary=2 -march=i686 -DMODULE -DMODVERSIONS -include /usr/src/linux-2.4.20-2.54/include/linux/modversions.h -g -nostdinc -iwithprefix include -DKBUILD_BASENAME=eni -c -o eni.o eni.c
    In file included from /usr/src/linux-2.4.20-2.54/include/linux/prefetch.h:13,
    from /usr/src/linux-2.4.20-2.54/include/linux/list.h:6,
    from /usr/src/linux-2.4.20-2.54/include/linux/module.h:12,
    from eni.c:6:
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:60: warning: parameter names (without types) in function declaration
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:60: field `loops_per_jiffy_R_ver_str' declared as a function
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:82: invalid suffix on integer constant
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:82: parse error before numeric constant
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:82: warning: function declaration isn't a prototype
    I tried recompiling using gcc296. I tried several things. The error abouce was in a driver I didn't need, so I changed the driver config option and tried the entire process again.
    A sample output was

    make[2]: Entering directory `/usr/src/linux-2.4.20-2.54/drivers/atm'
    gcc296 -D__KERNEL__ -I/usr/src/linux-2.4.20-2.54/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -fomit-frame-pointer -pipe -mpreferred-stack-boundary=2 -march=i686 -DMODULE -DMODVERSIONS -include /usr/src/linux-2.4.20-2.54/include/linux/modversions.h -g -nostdinc -iwithprefix include -DKBUILD_BASENAME=eni -c -o eni.o eni.c
    In file included from /usr/src/linux-2.4.20-2.54/include/linux/prefetch.h:13,
    from /usr/src/linux-2.4.20-2.54/include/linux/list.h:6,
    from /usr/src/linux-2.4.20-2.54/include/linux/module.h:12,
    from eni.c:6:
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:60: warning: parameter names (without types) in function declaration
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:60: field `loops_per_jiffy_R_ver_str' declared as a functioncp .config
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:82: nondigits in number and not hexadecimal
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:82: parse error before `0657d037'
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:82: warning: function declaration isn't a prototype
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:265: nondigits in number and not hexadecimal
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:265: parse error before `7413793a'
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:265: warning: function declaration isn't a prototype
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:269: warning: parameter names (without types) in function declaration
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:450: missing white space after number `7e9'
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:450: parse error before `7e9'
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:450: `kernel_thread_R_ver_str' declared as function returning a function
    /usr/src/linux-2.4.20-2.54/include/asm/processor.h:450: warning: function declaration isn't a prototype
    In file included from eni.c:6:



    ---
    So I used the -E option and typed
    gcc296 -E -D__KERNEL__ -I/usr/src/linux-2.4.20-2.54/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -fomit-frame-pointer -pipe -mpreferred-stack-boundary=2 -march=i686 -DMODULE -DMODVERSIONS -include /usr/src/linux-2.4.20-2.54/include/linux/modversions.h -g -nostdinc -iwithprefix include -DKBUILD_BASENAME=eni eni.c >/tmp/x

    And the key line was originally

    unsigned long loops_per_jiffy;

    This was changed to

    unsigned long loops_per_jiffy_R_ver_str(ba497f13);

    Obviously "jiffy" was defined somewhere in the preprocessor files


    ./include/linux/modules/ksyms.ver:#define loops_per_jiffy _set_ver(loops_per_jiffy)

    Which has
    #define __ver_loops_per_jiffy _ver_str(ba497f13)
    #define loops_per_jiffy _set_ver(loops_per_jiffy)
    Again the web advice was to do a "make mrproper." So I did
    cp .config /tmp/config
    make mrproper
    cp /tmp/config .config
    make oldconfig # I did not do this before
    make dep
    make
    make modules # and this time it worked!
    I decided to go back to the problem file and redoo the gcc -E option. The first difference between last time and this time was

    994c994
    < unsigned long loops_per_jiffy_R_ver_str(ba497f13);
    ---
    > unsigned long loops_per_jiffy_Rba497f13;
    There were many more differences.
    Next step is to reboot the new kernal and see if it works.
    Hmm. No. Got
    Loading jbd.o module
    /lib/jdb.o: kernel-module version mismatch
    /lib/jbd.o was compiled for kernel 2.4.20
    while this kernel is version 2.4.20-2.54custom
    Ah. I went back and instead of typing

    mkinitrd /boot/initrd-2.4.20.img 2.4.20

    I typed

    mkinitrd /boot/initrd-2.4.20-2.54custom.img 2.4.20-2.54custom
    and changed grub.conf to match
    Success!
    I rebooted, and did a "insmod speedstep.o" and then I was able to read the contents of /proc/sys/cpu/0 file. It shows I was running and 2.4Ghz. However, I was unable to compile/install the NVIDIA driver. I think my .config file was not a traditional one. I seemd to recall that my .config has a high memort (4GB) option, and the NVIDIA errors mentioned high memory. But once I knew it would work, I decided to re-tune it. I copied the configs/kernel-2.4.20.i686.config file into .config, did a "make oldconfig" - then I looked at it, and added the value

    CONFIG_CPU_FREQ_24_API=y
    Problems - during suspend
  • resolution changes - Happens during suspend, or closing lid of laptop can be fixed by Control-Alt-(fcn)+
  • i8k utils breaks sometimes (temperature remains constant after unsuspend. - fan always runs)

    Mon May 5 23:23:37 EDT 2003

    Upgraded to redhat 9.0. Gnome broke. kde works. I also noted that apt-get failed. it gave me the error:
    apt-get: error while loading shared libraries: librpm-4.1.so: cannot open shared object file: No such file or directory
    I also noted nVidia released the new driver that works with the latest kernel used by RedHat 9.0
    I noticed that apm was deleted while installing RedHat. Hmm.
    I re-installed it, and got it to work. But synaptic broke. apt-get says I am missing
    librpm-4.1.so
    librpmdb-4.1.so
    librpmio-4.1.so
    I went to APT-RPM Red Hat repository and added the two lines recommended to my to my /etc/apt/sources.list file to include the RedHat 9 apt repositories. I then did a 'apt-get update' and apt-get 'upgrade' to fix things. It's upgrading eog, glibc, glibc-common, glibc-devel, httpd, krbs-libs, openssl, openssl096b
    while editing the file, I noticed /etc/apt/sources.list.rpmnew and saw that I was also missing references to http://ayo.freshrpms.net
    Since I had problems with gnome, I also found another apt repository here - Matt Hall's.
    After adding these lines I did another apt update and upgrade. Many rpm's were updated.