Mostly successful dv9000z Debian Etch AMD64 install

[James Zuelow]

I have a new dv9013cl, and had some trouble over the weekend getting Linux installed. The initial install went well enough. I resized the ntfs partition and installed Debian and a swap partition onto a logical partition (sda5 and sda6). Grub detected the XP partition, the system restore partition, and the XP Embedded media player partition. I can boot to any of them. I am particularly pleased with the remote control. This control will boot the machine, and the arrow/ok keys work with Grub (and Linux in general -- see below). So I can select a partition and boot using the remote.

There are two main subjects -- kernel boot arguments and hardware that still fails to fully function.

KERNEL BOOT ARGUMENTS:

The BIOS and chipset are not quite Linux friendly. In fact Google shows some sites explicitly listing the dv9000z as not compatible with Linux.

It is compatible, however you must use boot arguments to get a stable system. Some distribution installers simply will not boot without kernel arguments (i.e. Ubuntu 6.06LTS). Others will boot and perhaps even get through the install using the wired ethernet port (i.e. Debian Etch RC-1 installer). However the system is subject to random lockups.

This happens with the F.19 BIOS that shipped with my laptop, and the newer F.1A BIOS. Tech support helpfully pointed me at the BIOS with the suggestion it would solve my problem -- it doesn't appear to have any Linux related fixes.

The kernel boot argument that gave me my most stable configuration, including reliable shutdown and reboot, all devices apparently working well (I haven't exhaustively tested some things such as ieee1394) is:

iommu=off

Other people have reported success with:

pci=usepirqmask

However it is important to note that you cannot use that kernel argument alone. The kernel documentation states that the argument does not have any effect if acpi is routing IRQs. Therefore if you try it, you should pair it with an acpi argument like this:

pci=userpirqmask acpi=noirq

Do NOT use

acpi=off

you break USB, sound, and most importantly for battery life, powernow/cpufreqd. My results with the pci/acpi boot arguments were not suitable for me. Shutdown and reboot were not reliable and appear to have been similar to just cutting the power. I've seen references to this "controlled crash" behavior on the net as well. Your mileage may vary.

If you have a dv9000z and you are using different kernel arguments successfully, please post the arguments.

REMAINING HARDWARE ISSUES:

One thing that DOES NOT work for me, regardless of boot argument (or lack thereof) is the headphone jack on the laptop. Plugging a set of headphones into the jack simply has no effect. The speakers are still active, and there is no output to the headphones. The microphone and s/pdif jacks are the same. These jacks work fine in XP and the embedded XP install, so it's not hardware.

The onboard broadcom bcm43xx wireless NIC does not work with the open source bcm43xx driver. The device is just not detected, so firmware is not installed. You must use ndiswrapper and the driver/firmware from HP's site (or your XP install if you didn't trash it).

RANDOM COMPLAINTS (Not Linux related):

HP could probably have rethought that SATA drive caddy thing. It is over-engineered, and just plain NOT AVAILABLE. Thanks. I have a second hard drive and no way of mounting anything into it. 3rd party wants almost $100 for just the caddy, and HP offers a 100GB drive with the cady (but not a 120 like what I have installed) for $355. Since there's no way the drive costs that much, I blame the expensive caddy again. Did I say thanks?

The engineer who designed the AC adapter (I hope you read the forums!) has obviously never flown on an aircraft with a laptop, and does not have cats. This thing has some play in it, which makes me concerned for the security of the plug mounted on the motherboard. It sticks out at a straight angle from the laptop, with lots of room to wiggle up, down and sideways. There is a very good chance that the plug will become either a weight bearing support for the laptop itself if it sits on the arm of a chair, or a weight bearing support for said cat when it walks by. I am concerned that either result will pop the soldering on the motherboard, and then I will lose AC power. Probably just after the warranty expires. I have a 12 year old Fujitsu laptop that has a nice, secure AC plug that is bent at a 90 degree angle, hugs the side of the laptop to avoid being snagged, and has a nice solid fit that doesn't wiggle inside components. So the technology to do it right has been around "a while." All you'd need to do to fix it is redesign the end of the cable to fit closer to the side of the laptop and have a firmer fit. Maybe with some sort of lip or flange to grip the inside of the hole, so that the machine's case is used to hold the adapter plug, not the electrical components.

The new finish is slippery, and shows dirt well. Yuck.

Other than those three small complaints, the laptop looks like it will serve well. Hopefully it's durable enough to last a few years.

[George Liu_4]

very informational.