Weird problem with my blocking browser: this has been a bad problem: (some/all) pages that use flashplayer plugins will block the browser to the point it will need to be killed.

Linux (RedHat 7.3/9.0, Fedora) on the Dell Inspiron 8200

Important note: out of the box this laptop has two major (fixable) problems with redhat 7.3: clock runs slow and the disk I/O slows down the machine beyond anybody's patience. Read on for the perhaps not-so-obvious fixes. Oddly enough they are both due to the fancier desktop managers Gnome/KDE and would not have appeared with e.g. fvwm, and should in general be distribution independant. The first notes below are sometimes specific to redhat7.3, some more redhat9 specific notes are below. I've put my fedora specific comments someplace else. machine arrived Sept 5 
	FIRST IMPRESSIONS:  XP slow, battery quite short, good sound,
                            nice crispy and bright screen, noisy fans

  - Win XP is slow, things take a long time, and a lot of disk cranking going
    on. Also at reboot still the case.
    Could not get the wireless to connect, whereas got that in linux
    in a snap. Go figure, must be me.
  - fans come on a lot, and are quite noisy. in linux they sometimes won't
    really stop.  [solution: hit Fn-Z, it will then stop]
  - there is no mute button on this machine? XP is very noisy a start and stop
  - battery is only 3:00 hr, instead of 3:20 for my old Dell5k, but much worse
    from the 5:00hr for a Dell8k!!! With 2 batteries loaded, this is a big 
    difference if you fly to Japan.....
  - the CMOS clock is terrible, i easily loose around an hour per day, probably
    need to get ntp running (which isn't easy on rh73) 
	solution: kill the battery applet (found Nov 17)
  + much better sound (volume wise :-) than the Dell5k
  + BIOS setup accessible from within a running operating system, very useful
    though it can muck up your clock.

  - Linux notes for install (details below)
    - partition the disk with PartitionMagic 7.0
	(mandrake 9.1 now comes with NTFS resizers in linux!)
	(but on another machine i was able to loose XP with PM7!!)
    - patch with the NVidia drivers, and get a good XF86Config-4 file
	(the new NVidia drivers make it (near?) impossible to patch the
	nv.c file to solve the suspend problem!!!)
    - eth0=wire  eth1=wireless   : both ok and work out of the box
      (for the wireless need to put the right stuff  in /etc/pcmcia/wireless.opts)
      though some issues after coming back out of suspend
    - lilo was not done properly, need to use 'lba32' option instead of 'linear'
      XP was also not (properly?) added to the menu, had to do that manually
    - manual sound volume does not seem to work (it was so nice in the end 
      this would work for the Dell5k)
	(i8k utils might solve this, didn't try it yet)
    - coming out of suspend the time is messed up
	(is that a UT vs. civil-time problem??)

  - My Favorite RedHat Patches:

      -	/etc/inittab:      '/sbin/mingetty --noclear tty1'   to prevent the login:
	prompt to clear the screen and see the end of the boot process

      - /etc/sysctl.conf:   kernel.cores_uses_pid = 0       was introduced around rh73,
        it would otherwise write lots of core.* files, instead of one. Tends to fill your
        disk too quickly if you do a lot of buggable program development and you forget
        to clean them. However, can be indispensible for debugging multithreaded apps.


  - Weird:

    - wireless comes on all the time at reboot/un-suspend....
	(ok, solved initially, but still given me problems)

    - sometimes suspend didn't work, some USB module kept being re-spawned and kept it from
      going into suspend. apmd then claimed   "suspend vetoed!"
	(happened in the beginning a few times, now seems to be gone)

    - time re-set at un-suspend needs some fix other than 'rdate -s earth.astro.umd.edu'
        and also don't run the battery applet, it slows the clock down due
        to high interrupt rate to the APM
	(i'm not running the clock in UT time mode, seems to work fine)

    - very bad disk I/O, hdparm gives fluctuating results 8 - 13 MB/s,
      should be a consistent 21.
	** the first time the disk I/O problem went away due to the patched NVidia
      driver was a fluke. upon a reboot i got back into slow I/O mode, very
      very unworkable. E.g. starting mozilla takes nearly a minute.
	** the 2nd time i noticed  the problem went away is when I mounted
      a CD... (CD/DVD is in /dev/hdb). umounting the CD immediately got the
      problem back. Mounting it again, problem gone..... go figure
	When the CD stopped spinning, problem back. Make it spin
	(e.g. (u)mount), problem gone.
      Some more trials revealed that twm and a fresh gnome user didn't
      have this problem, so it must be related to my gnome starting up
      some task pounding on the disk whenever the CD is not 'on'...
      what could that be :-)
      On GNOME's menu
            Programs | Settings | Peripherals | CD Properties
      there is a checkbox when an audio CD is inserted, to use gtcd to play
      the CD..... if it's checked, it goes bananas when no CD is mounted.
      The task at fault is called 'magicdev' (in Gnome, in KDE it would
      be called autorun; fvwm and twm don't suffer from the problem)

    - some startx's invocations result in crashing panels, and badly behaving
      GNOME. I mostly attribute this to the slowness of startup (see previous
      item) because of the I/O problem.

    - wireless claims to be 3c59x (also in modules.conf), whereas it
      should be wvlan_cs. Perhaps this explains the weirdness with wireless
      upon suspend.
	In rh9 it works just fine with orinoco_cs

    - left mouse button of PS2 mouse gets reset in a weird way after being
      online on a external display unit.
      When using USB this was still the case, but sort of resolved itself
      after some usage. Didn't have to restart X.

    - when starting a 2nd X, e.g. with
	startx -- :1 -screen Screen1
      to go to lower resulution there are two serious problems:
		- the USB mouse doesn't work anymore
		- it won't go into 16 or 24 bit mode, drops to 8 bit

- if you don't ifdown, close lid, and open. it won't get it back.
  You can also see this from wvlan, or by checking
  /proc/net/wireless

  Best is to:
    pcmcia stop
    pcmcia start

  You can also turn
    PCMCIARESTART=yes
  in the /etc/sysconfig/apmd script 



- in the middle (?) of playing a DVD:  time was reset and
  the mouse left button acts like it did when the Epson
  display was added for the talk.

  ** could not repeat the DVD / mouse problem ** but with
  the display unit it was repeatable

-  sometimes coming out of suspend you have:
	- a dead display, need to reboot (bless ext3!!!)
	  ....No know solution.....
	- fans that are spinning at max rate, and won't stop/spin down
	  ...Possibly the i8tools will fix this
 	  ** also have the impression things are sluggish when this happens **
	  Apparent solution: hit  Fn-Z and it should stop or slow the fans

Patch Software

Apart from your favorite distribution, you probably want the following software to patch your linux:

Installation: 

 - (optional) boot up the machine, do your WinXP registration thing
 - the disk comes as 1 partition, so you need something like PartitionMagic,
   since Dell didn't give us an option to get XP on a FAT32 partition.
   One perhaps useful suggestion to save time in PartitionMagic (it's slow)
   is to "split" the disk in an NTFS (for WinXP) and leave the tail end
   "as is" (free). I choose the route to allow ext2, which caused PM to stage 
   the process in 6 steps. On a 60GB drive this took 22 minutes. The 1 step
   process took 70 seconds. (later i learned mdk91 can do this very
   quick, probably they are using ntfsresize
    (the knoppix CD also has this program)
	
 - boot from CD, do the usual thing, configure the remainder of the disk for
   your favorite distribution.  I wound up with the following scheme.

Disk /dev/hda: 255 heads, 63 sectors, 7296 cylinders
Units = cylinders of 16065 * 512 bytes

   Device Boot    Start       End    Blocks   Id  System
/dev/hda1             1         4     32098+  de  Dell Utility         ?apm?
/dev/hda2   *         5       730   5831595    7  HPFS/NTFS            winxp
/dev/hda3           731      7296  52741395    f  Win95 Ext'd (LBA)    PM
/dev/hda5           731       991   2096451   82  Linux swap           swap
/dev/hda6           992      1373   3068383+  83  Linux                /
/dev/hda7          1374      7296  47576466   83  Linux                /home


 - after the installation lilo used the default 'linear' instead of 'lba32', 
   which is needed here.. Had to rescue boot and fix this manually. Also
   WinXP wasn't installed, had to add that manually too:

	other=/dev/hda2
	  label="WinXPPro"

 - everything basically worked, networking, sound etc. There are still some
   things to be tinkered with, mostly when coming back out of suspend. 
   Something with sounds is not perfect either.
   The fans drive me crazy. Under some circumstances this machine has the
   fans on *all* the time in linux.
Here's my latest /etc/X11/XF86Config-4 file. I use it with a logitech (optical) wheel mouse. Just make sure you have it plugged in before X starts, but don't let Kudzu at boot modify your PS2 setup, or you will loose your touchpad. Between suspends the external mouse adapts just fine.

Daily life with the Dell 8200

My normal usage of the laptop consists of being in two locations, lets call them work and home. Normally there is no need to shutdown and reboot the computer, suspending works just fine for this laptop. You can suspend by closing the lid (as long as it's not on AC power), or by hitting Fn-Esc (irrelevant of power mode). Usually I'm able to work in this mode for days and days. Between home and work the network and printer setup is of course different, and this was normally solved by keeping the relevant configuration files with an underscored location version. For example, /etc/resolv.conf_home and /etc/resolv.conf_work would exist, and scripts like athome and atwork would copy the appropriate one to /etc/resolv.conf. For certain configuration files (wireless, rh73 printing) some system services need to be restarted too. For all previous laptops i used a script called eth_configure, but with 2 ethernet cards in a machine, of which any one could be the active one, this turned out to be in-efficient. 
	...<< suspend/wakeup >>...
	athome
	ifup eth1
	...<< suspend/wakeup >>...
	atwork
	ifdown eth1		(pcmcia always starts up the wireless)
	ifup eth0   (after you put the ethernet cable in)
Occasionally - have not quite figured out when and why - the wireless comes up under eth0, check with ifconfig for those confusing moments. May have to do some combination of ifdown, rmmod.

Note added: the redhat-config-network tool in rh9 allows for multiple profiles, which solves many of the problems i solved with eth_configure. It does not solve the printing problem though. rh9 uses CUPS.

Using it for presentations there are still problems. Not all projector units are the same. Some need you to start X a-new, others adapt just fine, and others adapt fine, but only show a subset of the 1600x1200 display with scrolling edges. Also my mouse (at least PS2) seems totally corrupted (at least some of the left mouse action in Gnome), and forces me to restart X. You need an XF86Config-4 file with multiple screen sections of course. Check mine how this can be done.

Insert the ZiO! smartmedia reader, USB based, and issue 

	mount /mnt/flash
to copy files from the the disk.

To start under another resolution (e.g. your external display unit cannot handle 1600x1200) do: 

	startx -- :1 -screen Screen1
but notice the external USB mouse will now stop working, since somehow it was bound to :0.

Here are some 'benchmarks' on battery usage. You get these from /var/log/messages: 

(2002)
Oct 28 09:41:54 localhost apmd[1132]: Normal Resume after 21:56:31, -14.22%/day (87% 5:15) Battery power
Nov 19 12:30:21 localhost apmd[1212]: Normal Resume after 08:09:34, -14.71%/day (95% 5:44) Battery power
Nov 23 11:06:17 localhost apmd[1106]: Normal Resume after 12:32:59, -13.39%/day (93% 5:35) Battery power
(2003)

Apr  6 22:30:09 localhost apmd[1285]: Normal Resume after 08:39:31, -11.09%/day (96% 5:12) Battery power
Apr  8 09:35:27 localhost apmd[1285]: Normal Resume after 14:48:52, -17.82%/day (89% 4:50) Battery power
Apr  9 08:58:15 localhost apmd[1285]: Normal Resume after 10:35:42, -11.33%/day (74% 3:59) Battery power
Apr 26 05:55:11 localhost apmd[1229]: Normal Resume after 12:36:00, -22.86%/day (88% 4:47) Battery power

recently i've had really short times, like 3 hours, on the two batteries.... it was on
wireless and with high screen brightness, but still.   that last hour or 90 minutes is basically
gone. seems like a dead cell. that has been since before christmas at least. meaning instead of
the promised 6 hours, it now claims 5:23, of which the last 90 minutes is basically gone in a few.

RedHat9 conversion

Overal it works. Doing it in stages, e.g. first living with the stock X server from XFree86, and not using the new Nvidia yet. There were a few surprises for me:

Some hardware issues:

Remember that I've never had a laptop with zero issues, here's a record of my specific ones for this model.

Flashing the BIOS

If you have a laptop with only Linux on it (like me), it used to be cumbersome to upgrade the BIOS. Not anymore! From the Dell website you can either enter your service tag or directly enter your model name and number. On the next page choose under "download category" = "flashBIOS" and "operating system" "Linux" or "Windows XP" (doesn't change much). If you want to flash from Linux you will have to download the I8200A11.exe file (not the 8200_A11.exe file which is for Windows). Get the biosflash utility from John Hull, and follow his instructions to build a boot floppy with the bios on it. You'll then have to boot that floppy to flash your bios.

Some other useful Dell 8200 stories:

The two best known lists on TuxMobil and Linux on Laptops keep a much better tab on references. The ones I found useful in the beginning are:
This page was last modified on 20-Apr-2006 by teuben@astro.umd.edu.