Present Location: Projects >> linux-e7110

Linux on a Fujitsu E-7110

I actually sold this laptop back in 2004, when the screen broke and something bad happened to the video card.  Of course, I got it repaired first.


I came into possession of this beast a couple years ago, when my parents decided to buy me a laptop in exchange for my old IBM ThinkPad T20 (RPI issue).  I thought the specs looked pretty good : a P4 with 60GB of storage, and wireless.  At first, support was somewhat lacking, (for video and ACPI) but has been much improved lately.  Here's a small table of what works and what doesn't.

My lspci, proc/pci, dmesg, and glxinfo information can be found here.  Since I've moved to Gentoo Linux recently, I added another set of files.  If you have information that isn't on this page, please email me at or msg me on IRC at prox (on

Table of Hardware
Part Model Summary Prognosis
ACPI Controller Fujitsu Proprietory Chipset My previous prognosis on this component was a bit grim, but it seems to be getting better.

Kernel 2.4: Works nicely, except for any /proc/acpi/sleep functionality.  Battery info and performance/throttling options are available in /proc/acpi, as well as some miscellaneous info.  Patch your kernel with stuff from The ACPI ProjectExternal Content and enable everything but Toshiba/ASUS extras.  I've been enabling thermal zone, but I don't think there's any support for it.

Kernel 2.6: Works slight better than 2.4, with a slightly different interface.  The legacy /proc/acpi still exists (good for old dockapps) but the /sys interface is recommended.  Speed-stepping, battery info, and sleeping all work well.  Well, sleeping works, but it doesn't wake up!  Apparently this is a known issue for a number of systems, and a non-trivial one at that.  Enable ACPI and CPU Frequency Scaling in your kernel.  Select only the ACPI Processor P-States driver, and the Powersave and Userspace governor.

Just for reference, APM is nonexistant.  No information can be read from the apm device, and trying to suspend it results in an I/O error from the device node.  No legacy support here.
Network Controller RealTek 8139 (10/100MBit) Works perfectly.  Use the 8139too module in the kernel.  (Define CONFIG_8139TOO)

This chipset has been around for awhile, I have many of these PCI cards in other Linux boxes around the house.  They're not the best, but have great support.
Wireless Controller (802.11b) Harris Semiconductor (3873?) Use the orinoco_pci driver in 2.4 and 2.6.  Early versions used to spew TX errors to the syslog, but those have been cleared up, now.  Get new versions of the driver from Content, or just use the ones built into the kernel.  As of 2.6.1 and 2.4.24, the included versions work fine.
Installing the wireless-tools package will allow you to change the essid, channel, etc via iwconfig.
Graphics Card ATI Radeon Mobility LW (7500) / 32MB This didn't work out of the box for me due to some buggy Mesa software.  I spent too much time messing with this :-(

Installed XFree86 4.2.1 from Debian SID, and got 2D working fine.  The XVideo was buggy, and seemed to want to stay in the foreground the whole time I was playing videos with Xine (very annoying).

I headed over to Content and grabbed the latest Debian packages, and installed.  This fixed the XVideo problem, but didn't help 3D at all.  According to glxinfo, direct rendering was on, and the appropriate OpenGL extensions were present, but every OpenGL app crashes with "illegal instruction."  According to some docs, if "every opengl application crashes," libGL needs to be relinked or fixed.  I don't think the debian package is supposed to have this problem, but I download the whole XFree86 source, built it, and replaced the files.  No dice.

I left it like this for a few days, then started to bug people on about it.  Finally I learned that Mesa automatically thinks that if you have a iP4 chip, you have *all* the MMX-like instructions, including 3DNow.  This of course is wrong, that's the *only* instruction that the iP4's don't have.  Naturally, setting the environment variable MESA_NO_3DNOW to 1 before running any OpenGL apps made all the difference.

A Debian bug reportExternal Content seems to be filed for the xlibmesa3 package, addressing this issue.  However, I'm not sure what's going on since the post is 43 days old.  I'm hoping the XFree people will fix this.

So, yes, now this does work.  It's a little slow (~550fps with glxgears at 24/32bit color), so I'm hoping the DRI driver will mature...

Or ... you could use Gentoo, enable DRM in the kernel, omitting any specific drivers, then emerging xfree-drm.  Works perfectly.  In 2.6, just modularize the Radeon driver, and don't worry about xfree-drm.  I now get around 800fps with glxgears.
Sound Chipset Intel Corp. 82801CA/CAM AC'97 Audio Works fine with the OSS driver.  Define CONFIG_SOUND_ICH.  I think this chip has some 3d/spdif features, but I haven't messed with them. Excellent
USB Controller Intel Corp. 82801CA/CAM USB Works fine with the Alternate Driver support in v2.4.  Use CONFIG_USB_UHCI_ALT.  I didn't use it that much, but my mouse and external floppy drive seemed to work. Excellent
PCMCIA O2 Micro, Inc. OZ6933 Cardbus Controller I installed the latest pcmcia-cs package from Content and tried out a compactflash card reader.  It mounted and read correctly.  Guess this works without a hitch. Excellent
IEEE 1394 Controller Texas Instruments TSB43AB21 IEEE-1394 Controller Not sure about this one.  I think I selected the right options in the kernel, but I didn't see anything in dmesg.  Didn't play around too much (I have no firewire peripherals)... Unknown
CD-R/RW/DVD Drive Toshiba DVD-ROM SD-R2212 Works fine.  Just burned an audio CD with xcdroastExternal Content.  Not sure about the maximum speed though, the model says 16x in record mode, but it only reached around 12x when burning.

IDE-SCSI emulation is needed to burn CD's under Linux, I turned off IDE-CDROM support and enabled SCSI-CDROM support.