MobiliX - UniX with Mobile Computers Daemon Writing Tux
IntroHardwaresurveyMiscellaneous

Linux on the HP Omnibook 800

Preface

HP OmniBook 800 This is a survey about Linux related hardware features of the HP OmniBook 800. I don't try to explain the usual installation details (for instance, which partitions I have made). If you like to know more about Linux and laptops take a look at my Linux-Mobile-Guide. If you want to buy a HP OmniBook 800 you sometimes find a good offer at Warehouse .

Manufacturer Support

The hotline of Hewlett-Packard in Europe (Amsterdam) works very obliging in providing information about the hardware used in the HP OmniBook 800. Though it was only possible to get exact information about the graphic chip. Therefore I give the specifications I found in other sources below.

The HP-(Linux)-HOWTO v0.91 (in French) currently contains only minor information about laptops.

From the access statistics I found out that 8% of the visitors are from HP now :) though the support of HP in providing infos or other kinds of sponsorship wasn't much :( I wrote to HP twice but didn't get any answer yet.

You may also read the German issue (deutsche Fassung), but it's not up to date anymore.

Installation

On my HP Omnibook 800 CT I have installed the Deutsche Linux-Distribution DLD-5.2 with update to kernel 2.0.35-7. The installation from CD was quite easy. Exception: PCMCIA service was not detected automatically. I used an external SCSI CD drive (SCSI controller is Symbios-Logic ncr53c8xx). I didn't use the original HP CD drive (to expensive, to slow). Unfortunately the SCSI port has a proprietary plug. You have to buy a special HP SCSI cable (approx. 50 US$). This is available in two sorts, the cheaper one (Product-Nr. F1182A) is good enough. According to your SCSI device you need a gender changer (SCSI-2: <HP-OmniBook> MiniDS50Female to DR50Male <Your SCSI device>) , too. Probably it is also possible to install with PCMCIA via network, if you change the PCMCIA parameters as mentioned below during the installation process. I use Linux and Window$95 together, using the boot manager LILO without any problems.

Recently I changed to Debian/GNU Linux 2.2 aka Potato, which I like very much, though I couldn't get the SCSI controller to work yet (hadn't much time to fiddle this out). Hence I use a custom 2.2.7 kernel anyway, this is no problem.

Jason Theis reported some infos for SuSE and Redhat:

Clean install of Suse 8.0 (only): Create the bootdisk, modules1, modules2, and modules3 using rawrite. Stick the bootimage disk in the drive. Follow the instructions until it asks you to put CD #1 one. From there everything installs like a dream. Even installed Ximian-gnome and everything works great!

Clean install of Redhat 7.2 (only): Create boot.img using rawrite. Boot the machine with the disk in the floppy. Automatically detects your scsi cdrom (if you have the power on) and installs the valid drivers. From there it will start in text mode but then moves into graphical mode after you put the first cd in. The install works so nice. install packages seem bigger than suse but it depends on how you setup your machine (of course). Haven't done ximian but I am positive that will install without any difficulties. Special notes: Make sure the cdrom is plugged in and the power is on (my cdrom didn't automatically turn on) when you turn the machine on with the boot disk in.

Hardwaresurvey

General Hardware Data

586CPU/100MHz/16MB/810MB, HP Omnibook 800CT (TFT), 39.42 BogoMIPS (2.0.35)

I suppose these data are also valid in appropriate form for the HP OmniBook 2000, 5500 and 5700.

I use the machine nearly two years now and I'm quite satisfied by the quality of the laptop. Though I encountered three exceptions:

  • After eight weeks the cover for the peripheral ports at the backside broke.
  • After half a year of usage the keyboard controller broke. HP repaired it in a short time (3 weeks) without complaint.
  • The floppy drive broke after 18 months but was changed by HP with no problems.

Harddisk

General Information

TOSHIBA MK1926FCV, 777MB, w/128kB cache, CHS=789/32/63

Toshiba Support (Jumper Settings)

It is very easy to change the harddisk (up to 2GB are possible). The five according screws at the backside have to be openend with a TORX-6 screwdriver, than turn the machine, open the lid and lift the keyboard at the side of the display. Also this process is explained in the manual very clearly.

Speed Tuning

hdparm -d busmaster DMA doesn't work. hdparm -c1 /dev/hda enables I/O 32-bit and the harddisk seems to work faster and reliable. Here is the output from the benchmark bonnie (slightly modified to fit into the page):

32 Bit
 -------Sequential Output-------- ---Sequential Input----Random
 -Per Char- --Block--- -Rewrite-- -Per Char- --Block-----Seeks-
K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec %CPU
1311  43.3  1260 12.1   616 15.3  1256 43.6  1425 13.6  25.8  2.3
16 Bit
 -------Sequential Output-------- ---Sequential Input----Random
 -Per Char- --Block--- -Rewrite-- -Per Char- --Block-----Seeks-
K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec %CPU
1295  43.2  1268 14.2   596 20.1  1182 49.3  1433 24.7  25.8  4.0

For details about making your harddisk faster see c't 22/98, S. 216: EIDE-Festplatten voll ausnutzen (German)

Installing a Bigger Hard Disk

By Tom Kennedy : "Although it's not officially supported, I've been using a 4GB drive with my HP OmniBook 800 for over a year now without problems (using the newest BIOS off the HP web page)."

hda: IBM-DTCA-24090, 3909MB w/468kB Cache, CHS=993/128/63
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14

"Note that I don't speak for HP, and I don't work on any OmniBook products :) But I do use a HP OmniBook 800 with Debian :)"

Cedric Adjih reported to me: The/my Omnibook 800CT (810) can use actually some hard-drive of 3 GB. I mention it, because the hot-line was a bit frightening "ATTENTION, it is not supported, we cannot at all guarantee that bigger hard-drive will work at all". Here more information, I've been using it for 6 months, with no problems:

hda: FUJITSU MHD2032AT, 3102MB w/0kB Cache, CHS=788/128/63
hdparm -t (with hdparm -u 1, hdparm -c 1 before)

[...]
Model=FUJITSU MHD2032AT, FwRev=1812
Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% }
RawCHS=6304/16/63, rkSize=0, SectSize=0, ECCbytes=4
[...]

See also the page by Michael Ashley, which discusses his experiences, when he installed a large (12GB) hard disk in his HP800CT, also including some hints for configuring Linux (and MS-Windows98).

Tony Godshall wrote: "I have successfully installed IBM 20GB hard drives on my HP OB800 CT and have been running this config on two OB800's for about three months. These are at a great price point right now: ~$120USD (see http://www.pricewatch.com). (My system is Debian/GNU-Linux stable + some testing). IBM's OnTrack boot diskette (see IBM Hard Disk Utils) did the trick. I ran them with default config for a month and then switched them to 32bit, short timeout via hdparm."

BIOS

System Soft BIOS for Eagle II Version 1.01 (Software Version 2.00.00). The latest version is 2.01.00 and works without problems. The available BIOS settings you may find a the HP support page. A survey about the IRQs, DMAs, and I/O addresses you my find at another HP support page.

Y2K Hype aka Millenium Bug Hype

I checked the BIOS with the following dates and it seems to work alright:

        **)
BIOSDD012829282929 !!29
BIOSMM01020202020202
BIOSYYYY2000200020002004200420032040
date SatMonTueSatSunSatThu
       1 Mar 2003 !!1 Jan 1970

**) beyond 32 Bit border, see http://www.th-phys.rug.nl/~schut/gnulist.html for more information

Source: Linux-Magazin 11/98 p. 72

Memory (RAM)

I checked memory and cache with CTCM16N.EXE ftp://ftp.ix.de/pub/ct/pcconfig/ctcm16n.zip , which gives this output (style slightly changed):

PROZESSOR- und CACHE-INF     c't 5/98/ Andreas Stiller V1.6n
Prozessor Timing  : Pentium,Pentium-MMX
Prozessor CPUID   : GenuineIntel Typ=00 Fam=05 Mod=07 Rev=00 Feat=000001BF
Prozessor Name    : PentiumVRT mA4-Step
Aktueller Takt    :  98.788 MHz,  gemäss Pentium Timer:98.820 MHz
Primär-Cache(L1)  :   8 KByte,2fach assoziativ
Sekundär-Cache(L2): 256 KByte,direct mapped
Code Cache (L1)   :   8 KByte,2fach assoziativ
Hauptspeicher     : 16 MByte,keine Memory holes gefunden
Cacheable Area L1 : 16MByte, keine noncacheable Areas gefunden
Cacheable Area L2 : 16MByte, keine noncacheable Areas gefunden
Write Strategie L1: Write Back, no Write Allocation, linear Fill
Write Strategie L2: Write Back, no L2 Flush (wbinvd)
Dirty Tag L2      : ok
 Datenfluss- und Bus Performance (Hauptspeicher: 00111000h) :
Beste Zeit für  8K MOVSD Cache/Page Hit :  20.8 æs => 393.1 MByte/s
mittlere " für  8K MOVSD (Miss + Hit)   : 158.5 æs =>  51.7 MByte/s
mittlere " für  8K MOVSD (L2 clean)     : 205.5 æs =>  39.9 MByte/s
mittlere " für  8K MOVSD (L2 dirty)     : 287.3 æs =>  28.5 MByte/s
schlechteste "  8K MOVSD (misses)       : 389.5 æs =>  21.0 MByte/s
via FPU         8K       (misses)       : 389.4 æs =>  21.0 MByte/s
im Mittel bei 256 KB L2-Cache /DOS(640K): 173.3 æs =>  47.3 MByte/s
im Mittel bei 256 KB L2-Cache /Win(4M  ): 215.9 æs =>  37.9 MByte/s

It is very easy to upgrade the memory (up to 48MB). The access to the memory slot is quite simple. The process is explained in the manual very clearly. Norman Peterson -ncphd_AT_bellatlantic.net- reported a working memory module from PNY technologies and Peter Sprenger uses a 64MB module from Kingston. AFAIK the main memory module is not removable.

SCSI

The HP OmniBook 800 is one of the few laptops with an built-in SCSI port (the only other ones I know are the TI Travelmate 5000 series). It works alright with the Symbios-Logic ncr53c8xx PCI-SCSI-Bridge. To play music CD's it is necessary to connect the sound output of the CD player to the sound input of the laptop with a separate cable, but I didn't try this yet.

Unfortunately the SCSI port has a proprietary plug. You have to buy a special HP-SCSI cable (approx. 50 US$). This is available in two sorts, the cheaper one (Product-Nr. F1182A) is good enough. If you buy the HP OmniBook 800 CD drive a SCSI cable is included.

PCMCIA

The PCI-PCMCIA controller TI113x is compatible with i82365sl, it is detected by probe. The modules and cardmgr from the PCMCIA-CS-2.9.5-3 package are loaded fine. cardctl detects the sockets and the inserted cards. In the file /etc/pcmcia.conf I use the following parameters:

PCMCIA=yes
PCIC=i82365
PCIC_OPTS="cmd_time=12"
CORE_OPTS="cis_speed=500"
CARDMGR_OPTS="q"	# quite mode, doesn't beep

A test with a PCMCIA network card (RPTI-EP401 Ethernet Card) was successful. Also CardBus cards work.

With kernel 2.4.0-test8 I used the CardBus option in the BIOS and this /etc/pcmcia.conf file.

PCMCIA=yes
PCIC=yenta_socket
PCIC_OPTS=""
CORE_OPTS="cis_speed=500"
CARDMGR_OPTS="q"	# quite mode, doesn't beep

Excursion: MS-Window$95

The Window$95 driver for this networkcard seems to be one of the most wanted drivers searched for in the WWW. Many people look at this site because the RPTI-EP401 is mentioned. The Window$95 driver you may find at RPTI or better at OLIDATA. Both sites seem down sometimes. Though I couldn't get this driver working yet. Instead I use another driver (NE2000) which I installed as follows:

  • System Panel
  • Network
  • Add: Networkcard
  • Novell/Anthem
  • NE-2000 Compatible Device

Graphic-Chip

NeoMagic NM2070, 896KB, is not detected by SuperProbe. The chip is also known as MagicGraph 128, in MS-Window$95 it is displayed as Neomagic NMX.

XFree86 3.3.6

The NeoMagic chipset now has support in XFree86 (since 3.3.5). It is included in the SVGA driver. This works well up to 800x600 (x65536). A detailed list of parameters is in the according README in the /usr/doc section.

XFree86 4.x

I have installed the according Debian packages, but no time to make the right XF86Config-4.0 yet. Currently the screens scrolls vertically very fast. Usually this indicates a vertical sync pulse, which is way-out-of place (see XFree86-Video-Timing-HOWTO. So I tried different VertRefresh rates, but with no effort yet.

Gregory Mounie <gmounie_AT_cs.vu.nl> wrote: "This is informations on the way I get a working HP 800 (CT) with XFree86-4.0.3 at a resolution of 800x600 for updating the HP 800 page. I did not find this information anywhere save in *one* email of the xfree-neomagic mailing list ("progLCDModeRegs" option seems to be undocumented). The only point to get XFree86-4.0.x working is that you need to set up some options in the section:

Section "Device"
 Driver "neomagic"
# This first option removes the flickering of the screen at 800x600
 Option "progLCDModeRegs"
# Others options remove "ghosts" from the screen (any resolution)
# One way is to remove directly acceleration
 Option "no_accel"
# or to use shadow frame buffer
# (this option seems to disable acceleration too)
 Option "ShadowFB"
# may be some xaa option can do the same ?

Cory Bell <cory.bell_AT_usa.net> "The rolling LCD at 800x600 was killing me, but the "progLCDModeRegs" was just what I needed. I also experienced the same video corruption you mentioned when acceleration was enable, but I found a way to disable just the XAA feature that was causing the problem.

Machine - HP OmniBook 800CT w/NM2070 (896K) XFree86 Version - 4.0.3 (RedHat 7.1)

By running strings /usr/X11R6/lib/modules/libxaa.a | grep XaaNo I got a list of XAA debugging options (used in the "Screen" section of XF86Config), and tried different combinations. If I include *either* the "XaaNoPixmapCache" or the "XaaNoScreenToScreenCopy" options, my display experiences no corruption. Neither one seems to cause noticeable slowdown, so I'm not sure which to pick, but I thought the info might be useful to others of your readers. None of the other options seemed to have any effect. I don't need the "NoAccel" or "ShadowFB" options.

In the XFree86 4.1.0 docs , the options are defined as follows: Option "XaaNoScreenToScreenCopy" Disables accelerated copies of rectangular regions from one part of video memory to another part of video memory. Option "XaaNoPixmapCache" Disables caching of patterns in offscreen video memory."

Andi Gohr the new owner of my former machine provides latest information on XFree86 4.x, he has suggested the framebuffer X-Server.

Resolutions/Color Depths

Possible combinations of resolutions and color depths.

60Hz
 16	256	16bit
640x480		x	x	-
800x600		x	x	x
1024x768		x	-	-

Benchmark

Sorry no x11perf graphic benchmark yet.

NeoMagic X-Server

This NeoMagic X-Server by Jeff Shorey doesn't work alright, but it should work with HP Omnibook 800 166 MMX (MagicGraph 128ZV aka NeoMagic 2093).

VBE

The VBE driver by Ian Collier isn't working, because it is designed for VBE 2.0. The BIOS supports only VBE 1.3.

Framebuffer

The kernel 2.4.0 recognizes a VGA frame buffer device. Text mode works with 640x480 (vga=769) and 800x600 (vga=771), so you get a console screen of 100x37. XFree86 4.0.2 works in framebuffer mode 800x600!

Please check the latest release of DirectFB for a dedicated Framebuffer Driver for the NeoMagic chipsets, with support for accelleration.

AcceleratedX

The commercial X-Server LX31 by Xi Graphics for NeoMagic 2070 was installed as a testversion. It works fine "StaticColor" mode. Also the new version LX41 works with 800x600 and 1024x768.

RedHat (XBF)

At RedHat since july 1998 there is a X-Server (binary version) for NeoMagic chips available. This is designed by PrecisionInsight. The version 1.0.0 didn't work with my HP OmniBook 800 but the version 1.1.0-1 (XBF-neomagic-libc5-1.1.0-1) works well with 640x480 and 800x600 (virtual resolution 800x600). I couldn't get 1024x768 to work yet, but I didn't try much. Please note, the server is now named XFCom instead of XBF. I realised that, when updating the machine to glibc.

SciTech

The commercial SciTech Display Doctor 1.0 for Linux includes support for NeoMagic NM2070/90/93/97, NM2160 LCD. I didn't try the demo version yet, but I suppose it's worth to check it.

External Display

  • Display only at the external (CRT) monitor (BIOS-Option):
    • Textmode works.
    • Graphicmode with X11 causes the error (by the monitor I guess): "Out of scan range". I didn't have the time to work it out yet.
  • Display at the external (CRT) and the internal (LCD) monitor (BIOS-Option):
    • Textmode works.
    • Graphicmode works with X11.

Attention: Please use the right settings for your monitor, it could be damaged easily.

Tested with SONY 200sf monitor. The external display was tested with the XF86_VGA16 X server. It should be possible to use it with the XBF server too, see /usr/doc/XBF-version-/README for more info about the according settings in /etc/XF86Config.

I did a quick check with a XGA Hitachi CP-X960 beamer, worked in text and graphic mode. But with slightly distorted colours in graphic mode (this can probably be fixed). Note: I had to switch to the beamer display by using the <Fn><Print> function key combination (toggle). The beamer needs a few seconds to adjust to the new signals.

Text Mode 100x37

Please note that you may use the framebuffer setting vga=771 as described above to get the same result.

Cedric Adjih reported: "An apparently little known fact about Neomagic chipset is that you can run a text mode in 100x37 (i.e. 800x600)." His report is now separated into the appendix of the Linux-Mobile-Guide. It may work faster (untested!), than framebuffer mode, but seems more difficult to install.

Sound

Soundblaster and SoundblasterPro16-Bit compatible, ESS1888. HP specification recommends to disable the PnP support in the BIOS to get the soundcard working. I changed to PnP and back with no effect to sound in Window$95.

Quote from the support page of ESS:

Summary Parameters of AudioDrive Audio Controllers

Part NumberES1888
Plug and PlayBIOS
Integrated ESFM SynthYes
Dual Game PortYes
I2SPortNo
MPU-401 PortYes
DSP PortYes
Modem, CD-ROM, EEPROMNo
GPI/OGPO
Integrated 3-D Audio*No
Hardware Volume ControlYes
Full Duplex/ Half DuplexFD

Kernel Sound Driver

From the page Linux on the HP Omnibook 800 by Nathan Meyers (note this site will be no longer available) I got the following configuration details and modified them slightly. You have to check the parameters against your current BIOS setting. I tried them without MIDI in a non-modularized 2.0.35 kernel. They work fine with the packages sox and aumix. Kernel detects an ESS1688.

Please note - the size of the DMA buffer is a crucial point, when I choosed the default value 65536 the soundcard didn't work! I got a hint from Xavier Redon, who uses a size of 32768. I suppose you can increase this value with newer models. Probably you have to use make config, because make xconfig and make menuconfig don't seem to show all sound parameters:

CONFIG_SOUND=y
CONFIG_AUDIO=y
CONFIG_MIDI=y
CONFIG_YM3812=y
CONFIG_SB=y
CONFIG_ADLIB=y
SBC_BASE=220
SBC_IRQ=5
SBC_DMA=1
SB_DMA2=5
SB_MPU_BASE=330
SB_MPU_IRQ=11
DSP_BUFFSIZE=32768

For my 2.2.7 kernel I couldn't find the DSB_BUFFSIZE option in make config . I have ignored the warning in /usr/src/linux/drivers/sound/sound_config.h and have edited the file manually:

...
/*
 * Use always 64k buffer size. There is no reason to use shorter.
 * changed (64*1024) to(32*1024) -wh-
 */
...

Tony Chesser-Evans wrote: "I get a horrific amount of feedback through the built-in microphone in the case, because of the speaker in the bottom of the case. The only way I've found around it is to fire up a mixer program (e.g. aumixer) and cut the input level on the microphone WAY down, as well as cutting the output level down a bit. Maybe there's a way to put that in the kernel; do you know where the kernel initializes the mixer settings?"

Ron Yorston <rmy_AT_tigress.co.uk> wrote: "I've looked at the kernel source and have found that the initial values of the mixer settings are in /usr/src/linux/drivers/sound/sb_mixer.c. There's an array of initial values called sb_default_levels. Changing the value on the line marked /\* Mic \*/ from 0x1010 to 0x0000 has the effect of turning the microphone input level right down on initialisation.

Andi Gohr the new owner of my former machine provides latest information on sound with 2.4.x.

OpenSoundSystem - OSS

I tried a demo version of the commercial OSS-Linux sound driver by 4Front Technologies too, but with no success yet. But I am still working. The ESS1688 soundcard is detected but doesn't work tough. I suppose you have to change the DMA buffer size as mentioned above .

Advanced Linux Sound Architecture - ALSA

I couldn't check Advanced Linux Sound Architecture - ALSA yet, but I have got this report from Dan Levy:

After a hard nights testing I finally got the ALSA to work with my omnibook 800 ct 133 Mhz. I must say that (for me) it works alot better than the kernel stuff which never made with mp3:s.

I use the following: RH 7.2 2.4.16 kernel lilo boot-loader (make sure you compile OSS-module support in to the kernel)

Quick cook book recipe: download and extract the alsa-driver, alsa-lib and alsa-utils tarballs. in the alsa-driver dir configure with following options:

./configure --with-isapnp=yes --with-oss=yes --with-cards=es1688
make
make install

Configure and make the lib and the utils, it's straight forward. Put the following lines in your modules file, i.e. /etc/modules.conf:

# Alsa portion
options snd snd_major=116 snd_cards_limit=1
alias snd-card-0 snd-card-es1688
options snd-card-es1688 snd_port=0x220 snd_dma8_size=32 snd_dma8=1 snd_irq=5
# OSS portion
alias snd-slot-0 snd-card-0
alias char-major-14 soundcore
alias sound-service-0-0 snd-mixer-oss
alias sound-service-0-1 snd-seq-oss
alias sound-service-0-3 snd-pcm-oss
alias sound-service-0-8 snd-seq-oss
alias sound-service-0-12 snd-pcm-oss

I have absolutely no idea what the OSS portions does or if some lines are obsololete but it works.

Next make sure you don't have any pcmcia modules loaded (lsmod), you have to insert the sound modules before the pcmcia. "/etc/init.d/pcmcia stop" works sometimes:) "modprobe snd-card-es1688" loads the alsa modules, et voila! IMPORTANT the alsa-driver is muted by default, start the alsamixer (in the utils pakage) to set the volume. This can easily be scripted away.

alsa-driver-0.5.12a alsa-lib-0.5.10b alsa-utils-0.5.10

Infrared Port - IrDA(TM)

Standard IrDA - SIR

These laptop doesn't use the common routine for the initialisation of the SIR port as a serial port emulation. Therefore you need a dedicated IrDA driver (SIR/FIR) , which is provided in Kernels >= 2.4.9

Fast IrDA - FIR

The PCI - IrDA(TM) controller VL82C147 manufactured by VLSI is FIR capable. Since Kernel 2.4.9 there is a dedicated Linux driver available. Andi Gohr the new owner of my former machine provides latest information on FIR and SIR.

You may read more about Linux and Infrared in my Linux-Infrared-HOWTO .

IR Remote

I had no success yet to get the machine sending or receiving IR remote control commands. For Linux information see the LIRC project.

Parallel Port

Parallel port works fine with parport module.

Mouse

The unique and funny Pop-Up-Mouse (two buttons) uses the PS/2 protocol (/dev/psmouse). It is located at the right side of the laptop, maybe a problem for the lefthanded. The gpm works well together with XFree86 or the XBF server. 3ButtonEmulation works in X11, but it seems hairy to me to get the right trigger point. Three button functionality shows up in mev (the mouse-event-manager is part of the gpm ), too. An external serial mouse works, too.

Recently with 2.4.x kernels and Debian WOODY the mouse no longer works in GPM mode, but in X. Very strange! As a remedy you may use an older version of gpm and set the package on hold in dselect.

CD

I use an external SCSI-CD-Rom. But not the original one by HP, this seems to expensive and to slow.

Peter Sprenger reported to me: I got the genuine CD-ROM and so far I was not able to run it ... the device keeps on resetting until it starts to rattle so loud that it scares you, I guess it is a missing kernel parameter.

Michael Wiedmann reported the genuine CD drive working in the docking station without any problems.

Floppydrive

Reliability

The floppy drive broke after 18 months, I got always CRC errors, when booting from a floppy. And writing a normal floppy became unreliable. But the drive was changed by HP with no problems.

superformat

I couldn't get the floppydrive to superformat (fdutils 5.3) more than 1,44 MB. Though I tried different parameters with floppycontroland setfdprm, I get the error messages mentioned below. But preformatted floppies (/dev/fd0H1722 and /dev/fd0H1920) can be used (mount, mcopy, boot). FDC 0 is a post-1991 82077.

#superformat /dev/fd0 sect=21 cyl=83
floppy0: sector not found: track 0, head 0, sector 1, size 2
floppy0: sector not found: track 0, head 0, sector 1, size 2
end_request: I/O error, dev 02:00, sector 0
floppy0: sector not found: track 0, head 0, sector 3, size 2
floppy0: sector not found: track 0, head 0, sector 3, size 2
end_request: I/O error, dev 02:00, sector 2

The floppydrive used with the Toshiba Portege seems to be the same model. I tried to get further info but didn't succeed yet, because most of the Toshiba sites and newsgroups are in Japanese. Maybe someone can help out with this information?

Brandon Davis <brdavis9_AT_pacbell.net> wrote: "I do know that the external floppy for the OB600/800 is the same floppy (pretty much, anyway: the eject button is different) that is used in the IBM ThinkPad 560 ...I hooked up my HP floppy to my wife's TP560 and it worked just fine."

When building a rescue floppy with muLinux by Michele Andreoli I found a hint about WinImage . This Window$95 shareware is made by Gilles Vollant. I tried his demo version and was able to format 1722MB bootable floppies, without any error. Writing this floppies with Linux dd was no problem then.

Finally I found a Linux solution, too. I have set up this /etc/driveprm:

drive0: deviation=656500

I found out that floppymeter /dev/fd0 reported this recommendation 'drive0: deviation=1250720' and superformat /dev/fd0 reported 'drive0: deviation=6565'. Therefore I guessed a new value. I'm now able to format the drive with 1.440MB and 1.772MB without errors. Please note: you probably don't have to fiddle with these parameters if you only work with 1.440MB floppies.

It worked for me but there are no guarantees. You should read the according information of the fdutils package carefully.

CRC Errors

Sometimes but not always I get CRC errors when booting from floppy-images, which I prepared with dd or cat. This doesn't happen when I use the DOS programm RAWRITE2.EXE instead. For example I had to do this when preparing the DEBIAN installation floppies.

Peter Sprenger recommends to use the kernel parameter 'floppy=omnibook', he never noticed the crc error; without he had it once. Instead of 'omnibook' you may also use 'floppy=nodma' which has the same effect. But for me this doesn't help. From a DOS tool I know the floppy controller is a PNP0700, which is the generic (standard) one. The proprietary HP OmniBook floppy controller is PNP0605. Also I have got reports about disabling the cache might help, I will try this later.

ID = 0x0007D041 = 'PNP0700' -- Std. FDC (765 type)
Types: Base = 0x01, Sub = 0x02, Interface = 0x00
Mass Storage, FDC (765 type), Generic Floppy

Keyboard

After half a year of usage the keyboard controller broke. When typing a letter another letter appeared at random on the screen, especially lower and upper case were mixed unpredictecaly. Can you imagine to type in the correct password :) HP repaired it in a short time (3 weeks) without complaint. But the error seems to return sometimes, but to seldom to complain about.

Besides this the keyboard works fine. There is only one LED to show the power status. The status of NUM, CAPS and SCRL may be displayed in a small auxiliary display in one of the four display corners, by toggling <Fn><ESC>. With XFree86 this status display doesn't work, but with the XBF server it works fine. Also in text mode setleds works with the status display, under X11 xkbvledscauses an error. I didn't try an external keyboard yet.

Advanced Power Management (APM)

The BIOS supports APM version 1.2 .The kernel function Power-Off-On-Shutdown (CONFIG_APM_POWER_OFF) works alright. The other functions I couldn't test yet. The APM of the BIOS works with Linux too and goes into power save mode when the choosen time is elapsed. I haven't installed the 'apm' package anymore, but I guess apm -s worked.

Peter Sprenger reported: APM does work fine -- no problems at all!

One minor problem: When removing the power plug at the far side from the machine I get this error and usually can't recover my work:

hda: lost interrupt
hda: read_intr: status=0x50 { DriveReady SeekComplete }

If I pull the power plug at the machine side everthing goes well.

Battery

Battery (Lithium Ion) works 2 hours for each charge cycle, charge time 1 hour.

Real Time Clock (RTC)

RTC is detected by the kernel and shows up in /proc/rtc.

Survey PCI Devices

busdevfunvendor
id
device
id
vendordevicedescription
0001004104VLSIVL82C535host bridge
0101004102VLSIVL82C534PCI to PCI bridge
0201004101VLSIVL82C532peripheral controller
030  NeoMagicNM2070VGA controller
040  TIPCI1130CardBus controller
041  TIPCI1130CardBus controller
050  SymbiosLogicncr53c810 *)PCI-SCSI bridge
0601004105VLSIVL82C147PCI-IrDA controller

*) aka Symbios Logic 8100S (info from the device manager in Window$95)

Sources:

  • C'T PCI-Optimierer Version 3.03.00 (c) Georg Schnurer, Redaktion CT, 1993 ... 1998
  • PCI.EXE Intel Corporation Version 2.2
  • cat /proc/pci
  • Window$95 INF-Files

Docking Station

I don't use a docking station.They seem really expensive and I can't see any usefulness. I would like to buy a PC instead and connect it via network to the laptop. Or use an external display, which works well as described above, and an external keyboard and mouse.

From Martin J. Evans martin_AT_mjedev.demon.co.uk "The main problem with docking stations is getting the operating system to detect you are docked. Fortunately, if you configure your kernel with the /proc file system (does anyone not do this?) you can examine the devices available and thus detect a docked state. With this in mind a few simple scripts is all you need to get your machine configured correctly in a docked state.

You may want to build support for you docking station hardware as modules instead of putting it directly into the kernel. This will save space in your kernel but your choice probably largely depends on how often you are docked."

Peter Sprenger reported to me: I got a docking station and I was able to make a kernel virtually out of the box, the only problem you have is the un-docking for the following reason: the BIOS is programmed that, if your detach a SCSI-device, automatically reboots the computer. It is not a disaster, but it then takes some time to re-sync the hard disk. The question is whether we could convince HP to produce a BIOS patch, for us 'LINUXER', that allows detaching without a reboot.

Michael White wrote: "If you use the docking station along with the PCI card slot in the docking station, Linux (at least Redhat 7.2) will only load one PCMCIA card, complaining of "RequestIRQ: Resource in use". I've tried this with several card combinations, all of which produce the same result.

If the upper and lower (or only the upper) card is plugged in, only the upper card that will be initialized (but both will be recognized). If only the lower card is plugged in, the lower card will be recognized. If you remove the upper card and insert the lower card, the lower card will be initialized.

Note that I've tried the work-arounds on the PCMCIA How-to page, and the options that are recommended are not recognized by the Yenta driver."

I got a docking station lent recently. Its features are:

  • network card (unknown model probably 10/100BaseT, manufacturer might be KALEX K668), not working with Linux yet, since lspci reports the ethernet controller Intel Corporation 82557, it might be an Intel EtherExpress Pro100, driver see Donald Becker
  • seems to have an additional SCSI controller SYMBIOS LOGIC 53C810A, the chip is labelled SYMBIOS LOGIC 8100S HOST ADAPTER
  • magnetic lock, to avoid unintended unlocking, not working with Linux yet
  • additional PS/2 keyboard port
  • additional PS/2 mouse port
  • standard SCSI port
  • floppy drive port
  • parallel port
  • serial port
  • VGA port
  • audio port

Output from lspci, the last three entries are from devices of the the docking station:

00:00.0 Host bridge: VLSI Technology Inc 82C535 (rev 03)
00:01.0 PCI bridge: VLSI Technology Inc 82C534 (rev 03)
00:02.0 Unknown class [ff00]: VLSI Technology Inc 82C532 (rev 02)
00:03.0 VGA [..]: Neomagic Corporation NM2070 [MagicGraph NM2070] [..]
00:04.0 CardBus bridge: Texas Instruments PCI1130 (rev 04)
00:04.1 CardBus bridge: Texas Instruments PCI1130 (rev 04)
00:06.0 Unknown class [0d00]: VLSI Technology Inc 82C147 (rev 02)
01:00.0 SCSI [..] : Symbios Logic Inc. (formerly NCR) 53c810 [..]
01:05.0 ISA bridge: VLSI Technology Inc 82C538
01:06.0 Ethernet controller: Intel Corporation 82557 (rev 01)

AFAIK the docking station can only be released from the notebook, if the power is on, by pressing the release button.

Models of the HP OmniBook 800

 ModelCPUHard DiskRAMDisplayGraphic ChipL2APMSCSI
IrDA
PCMCIA
Product Number
800 CT2100Intel Pentium 166 MMX2,00 GB16 - 80MB10,4" TFT SVGANM 2093 1MB512 KB  F1360A
800 CT1440Intel Pentium 133-MHz1,44 GB16 - 48MB10,4" TFT SVGANM2070 896KB256 KB  F1175A
800 CT1440Intel Pentium 100-MHz1,44 GB16 - 48MB10,4" TFT SVGANM2070 896KB256 KB  F1174A
800 CT810Intel Pentium 100-MHz810 MB16 - 48MB10,4" TFT SVGANM2070 896KB256 KB1.2NCR53c8xx
VL82C147
TI113x
F1173A
800 CS1440Intel Pentium 100-MHz1,44 GB16 - 48MB10" DSTN SVGANM2070 896KBNo  F1172A
800 CS810Intel Pentium 100-MHz810 MB16 - 48MB10" DSTN SVGANM2070 896KBNo  F1171A

HP OmniBook Survey (All Models)

Linux with the HP OmniBook Series (Links)

Support des machines HP sous Linux

A report by Bruno Cornec (in French)

HP OmniBook 600 (Out-of-production)

HP OmniBook 800

HP OmniBook 900

  • N.N.

HP OmniBook 2000

  • N.N.

HP OmniBook 2100

  • N.N.

HP OmniBook 3000

HP OmniBook 3100

HP OmniBook 4000 (Out-of-production)

  • Graphic-Chip WD90C24, see David Fox
  • Alta Ergo alta_AT_zebra.net HP Omnibook 4000c *)
  • N.N. **)
  • N.N. **)

HP OmniBook 4100

  • Mark Pitman map_AT_primenet.com HP Omnibook 4100 *)
  • Carsten Heyl SuSE 6.2
  • XBF Server from RedHat (PrecisionInsight) is reported working with the NeoMagic chip

HP OmniBook 4150

HP OmniBook 5000 (Out-of-production)

HP OmniBook 5500 (Out-of-production)

  • by Stephan Ryan *)
  • Peter Libassi libassi_AT_swipnet.se HP OmniBook 5500CT, Slackware based 2.0.30 *)
  • David Puryear dayear_AT_codenet.net HP OmniBook 5500CT 10.4TFT w/800x600 in 8&16 bit, Debian 1.3.1, kernel 2.0.30 *)
  • Matthew Chappee matthew_AT_mattshouse.com HP Omnibook 5700

HP OmniBook 5700

  • Mark Crosbie by Mark Crosbie **)
  • N.N. *)
  • Kris Van Hees kvanhees_AT_fnx.com HP Omnibook 5700CT *)
  • Geoffrey Myers geof_AT_abraxis.com HP OmniBook 5700CTX *)
  • Matthew Chappee matthew_AT_mattshouse.com HP Omnibook 5700 *)
  • Matthias Danckwerts N.N.

HP OmniBook 6000

HP OmniBook 7100

  • N.N. **)
  • 266MMX Intel Mobile Pentium II, 32MB RAM, 32KB L1 Cache, 512KB L2 Cache
  • 2.5" EIDE Harddisk 6.4GB or 8.1GB
  • Graphic-Chip "ATI Rage LT Pro 64-Bit" seems to be supported by the commercial SciTech "Display Doctor 1.0 for Linux" and probably by others too.
  • Infrared port seems to work in SIR mode (serial emulation).
  • Output from /proc/pci (modified):
    Bus  0, device   7, function  3:
    Bridge: Intel 82371AB PIIX4 ACPI (rev 2).
     Bus  0, device   7, function  2:
    USB Controller: Intel 82371AB PIIX4 USB (rev 1).
     Bus  0, device   7, function  1:
    IDE interface: Intel 82371AB PIIX4 IDE (rev 1).
     Bus  0, device   7, function  0:
    Bridge: Intel 82371AB PIIX4 ISA (rev 2).
     Bus  0, device   4, function  1:
    CardBus bridge: Texas Instruments PCI1250 (rev 2).
     Bus  0, device   4, function  0:
    CardBus bridge: Texas Instruments PCI1250 (rev 2).
     Bus  0, device   2, function  0:
    VGA compatible controller: ATI Rage LT Pro 64-Bit (rev 220).
     Vendor id=1002. Device id=4c50.
     Bus  0, device   0, function  0:
    Host bridge: Intel 440BX - 82443BX Host (no AGP) (rev 2).
    

HP OmniBook XE

HP OmniBook Sojourn

  • XBF Server from RedHat (PrecisionInsight) is reported working with the NeoMagic chip.

Linux/Unix/BSD on other Mini-Notebooks (Links)

Credits

  • David A. Hinds maintainer of the PCMCIA-CS package
  • Linux IrDA Project
  • **) Linux on Laptops by Kenneth E. Harker -kharker_AT_cs.utexas.edu-
  • *) Linux Laptop Volunteer Support Database by Kenneth E. Harker
  • The members of the linux-laptop@vger.rutgers.edu mailing list
  • Jeff Shorey for his NeoMagic X-Server -js_AT_mnsinc.com-
  • Overview Linux on Notebooks by David Fox
  • Linux Online - Linux on a Laptop
  • adorable's Toshiba Libretto
  • Mark Crosbie -mcrosbie_AT_best.com- for his help with IrDA.
  • Torsten Schuster -torsten.schuster_AT_usa.net- for his advice about sound.
  • Ralf Zabka -ralfz_AT_ibm.net- for his advice about IrDA and sound.
  • Friedhelm Kueck -friedhelm.kueck_AT_impress.de- for his advice about the HP OmniBook 3100.
  • Ioannis Tambouras -ioannis_AT_flinet.com- for his suggestions about sound.
  • Charlie Stross -charlie_AT_antipope.org- for his recommendations about X11 and PCMCIA.
  • Xavier Redon -Xavier.Redon_AT_eudil.fr- for his help with soundcard configuration.
  • Sadik Hafizovic for his hints about the page layout.
  • Norman Peterson -ncphd_AT_bellatlantic.net- for infos about memory modules.
  • Peter Sprenger <spre_AT_lugs.ch>
  • Cedric Adjih <cedric.adjih_AT_inria.fr>
  • Thorsten Vahlsing <vahlsing_AT_weh.rwth-aachen.de>
  • Michael Wiedmann <mw_AT_miwie.in-berlin.de>
  • Tony Chesser-Evans <tony_AT_james-river.net>
  • David Spalding <david_AT_korova.com>
  • Brandon Davis <brdavis9_AT_pacbell.net>
  • Tom Kennedy <tomk_AT_magnum.fc.hp.com>
  • Michael Ashley <mcba_AT_ugrad.phys.unsw.edu.au>
  • Ron Yorston <rmy_AT_tigress.co.uk>
  • Tony Godshall <tg_AT_of.net>
  • Andreas Gohr <a.gohr_AT_web.de
  • Juergen Hausmann <Juergen_Hausmann_AT_agilent.com>
  • Cory Bell <cory.bell_AT_usa.net>
  • Dan Levy <danlevy_AT_island.liu.se>
  • Michael White <michael12_AT_mindspring.com>
  • Jason Theis <Jason_Theis_AT_non.cocreate.com>
  • Others whose names I forgot to record.

MobiliX - Mobile Unix

Please feel free to mail me if you have recommendations or criticisms. I'm very receptive to additions, suggestions and changes from the readers of this document. Please note: If you don't object I will include your name and e-mail address in the credits section above.

Werner Heuser <wehe@mobilix.org>

 
Site Map · Search
Print · i-mode · WAP
About · Contact · Disclaimer · Sponsors

© Werner Heuser 1997-2002 · http://mobilix.org/hp800e.html · last change Sat Aug 17 2002