Introduction

18 Aug 2006

Here is a brief summary of hardware:

The notebook was shipped with A05 BIOS revision. As of this writing, current version is A17.

Memory was upgraded from 1 GB to 2 GB by using Corsair 1 GB 533 MHz Value Select memory (200-pin SO-DIMM). It passed memtest86 tests.

I also own another Inspiron 6400 notebook, but with Pentium Dual-Core 1.6 GHz processor (1 MB of L2 cache), 5400 rpm disk, different DVD burner, 1280 x 800 screen and 1 GB of RAM. All of this applies to it as well.

Fedora 13

The notebook was upgraded to Fedora 13 using yum upgrade without issues.

Suspend/resume and hibernate/thaw now work on this notebook.

This will be the last upgrade of this notebook. After this, it will be retired. A new Dell Studio 1558 Lenovo ThinkPad T510 is on its way...

Fedora 12

I upgraded both notebooks to Rawhide, which currently holds F-12 (15 Nov 2009). The upgrade went smoothly, but there are some regressions users of compiz need to be aware of. A bug has been filed for this (https://bugzilla.redhat.com/show_bug.cgi?id=537494). So, I'm forced to use metacity once again, because I just need to be able to hibernate/thaw.

Workaround for the above bug is: gconftool-2 -s /apps/compiz/general/screen0/options/sync_to_vblank -t bool false. I only tested with nomodeset for now. New kernels that contain the fix are being built in Koji: http://koji.fedoraproject.org/koji/packageinfo?packageID=8

Suspend and hibernate (i.e. suspend to disk) still only require: SUSPEND_MODULES="b44" in /etc/pm/config.d/unload_modules.

Network Manager is still giving me some trouble. System wide interfaces require root password for non-root users (even when they are controllable by all users) and eventually don't actually work. Per-user configurations work fine this time. Anyway, I'm still using my own little tool to help me switch between LAN and WLAN. You can get it on the Fedora page.

There are also problems with b44 driver (wired network) disconnecting. Various bugs have been filed in Red Hat Bugzilla for this, as issue appears to have been present in the earlier version of Fedora too. I don't know why it's hitting my machine only now.

Be careful of this bug too: https://bugzilla.redhat.com/show_bug.cgi?id=534047. It allows all local users to install any software they like, from a signed repository.

As of today (17 Nov 2009), my second Inspiron 6400 notebook is not running Fedora any more. So, any reports going further will be for the original machine only.

Fedora 11

I upgraded both notebooks to Rawhide, which currently holds F-11 (18 May 2009). The upgrade went smoothly and this is a very nice release. The only problem I found so far is related to kernel graphics mode setting and hibernate/thaw. A bug has been filed for this (https://bugzilla.redhat.com/show_bug.cgi?id=500983).

Suspend and hibernate (i.e. suspend to disk) now only require: SUSPEND_MODULES="b44" in /etc/pm/config.d/unload_modules.

If had some trouble getting Network Manager to work, so I created a little tool to help me switch between LAN and WLAN. You can get it on the Fedora page.

Fedora 10

I upgraded both notebooks to Rawhide, which currently holds F-10 (23 Nov 2008). The upgrade went smoothly, but I cannot say that this release doesn't have issues:

Fedora 9

Nothing much worth reporting here. Most stuff just works, including suspend and hibernate.

Fedora 8

Upgrade from F7 went smoothly on one of the notebooks. The other one (with more packages installed) got hit by https://bugzilla.redhat.com/show_bug.cgi?id=371111. After removing several packages from the system identified by yum update, the upgrade enventually succeded.

On the machine that was upgraded smoothly, hibernate worked just fine. On the other system, repeated attempts to hibernate resulted in file system errors. At times this was inability to write certain disk blocks, at other times there were ext3 file system errors, which resulted in the file systems being remounted read only. Subsequent file system checks and scans for bad block did not reveal any problems. So, this is very much still in the "cause unknown" phase. Temporary workaround for this is to downgrade the kernel to the latest F7 kernel, version 2.6.23.1-21.fc7. You can track that problem here: https://bugzilla.redhat.com/show_bug.cgi?id=373821.

Fedora 7

Upgrade from FC6 to Fedora 7 RC2 (this same code will be in the final release of Fedora 7) went smoothly. The update was performed using the HTTP method, directly off Red Hat servers (against the development tree, which was frozen at that point).

However, once the system was installed, it would not boot without maxcpus=1 options passed to the kernel. Later I found out that passing in just clocksource=acpi_pm was sufficient (i.e. you get to keep the second core after all :-). Some recent updates to stable 2.6.21 kernel have broken things on this notebook, but some kernel developers believe it may also be a BIOS bug (Dell have been notified). See this bug report to track the situation: https://bugzilla.redhat.com/show_bug.cgi?id=241249.

Fedora 7 Live RC2 CD had problems booting up on my second Dell Inspiron 6400 (the kernel would hang consistently). The above boot option (clocksource=acpi_pm) does work around the problem.

Unfortunately, in terms of suspend/resume, there have been some regressions in Fedora 7. Suspend and hibernate (i.e. suspend to disk) can be made to work by creating a file /etc/pm/config.d/unload_modules which contains these blacklisted modules: SUSPEND_MODULES="b44 iwl3945". With the early F7 kernels, upon resume from suspend to RAM, the system was painfully slow. So, I had DISABLE_SUSPEND='yes' in the above file as well. With more recent kernels (e.g. 2.6.22.1-41.fc7), this does not happen any more. In fact, I've even been able to suspend/resume without having any modules listed in SUSPEND_MODULES, but this would need more extensive testing in order to be completely confirmed (I only did a few cycles).

Fedora Core 6

Upgrade from FC5 to FC6T1 went smoothly. The system was then upgraded to final version of Fedora Core 6 using yum. You will still need 915resolution utility to get the screen going. Unfortunately, the i810 mode setting driver does not work correctly on this notebook.

Suspend to disk (i.e. hibernate) works a lot better with FC6. No need to compile anything - Fedora supplied kernels will do just fine, although suspending and resuming is a bit slow (i.e. this is the old vanilla swsusp code - see below for Suspend2 support). Also, Fedora supplied pm-utils package does the job well.

To get Suspend2 on this notebook, use the following files:

The first one is a patch against rpmbuild directory in which the relevant kernel source has been installed via the RPM package. Just run bzcat kernel-VERSION.patch.bz2 | patch -p1 from rpmbuild directory and you'll get the correct files in place. Then run this from the SPECS subdirectory:

rpmbuild --sign -ba --target=i686 --define '_without_smp 1' --define '_without_pae 1' --define '_without_xen 1' --define '_without_kdump 1' --define '_without_debug 1' --define '_without_headers 1' --define '_without_doc 1' --define '_enable_debug_packages 0' --define '_unpackaged_files_terminate_build 0' --define 'dist .fc6' --define 'buildid .1rxv' kernel-2.6.spec

The other two RPMS build conventionally.

Suspend to RAM works fine as well, save occasional blank screen on resume.

AIGLX works fine on this notebook.

I don't use wireless network with this notebook, but there there are drivers out there for this type of wireless network card.

If you want decent performance out of the DVD-RW drive, add combined_mode=libata to the kernel line of /boot/grub/grub.conf file.

Fedora Core 5

Initial install of FC5 on this notebook went relatively smoothly, but the graphics chip was only recognised as VESA. In order to get the i810 driver to recognise the chip properly, you need to have at least version 1.6 of the driver. Luckily, this is available form FC5 AIGLX repository, so you may get the screen going that way.

Something that doesn't exist in FC5 and is required in order to get 1680 x 1050 resolution is 915resulution utility. It is easy enough to find source and binaries in the Internet, though.

After a few attempts to get suspend to disk going properly with both built in swsusp code and Suspend2, I could not get this to work reliably. Problems were ranging from ata_piix driver inablility to be used as a module to random crashes on suspend or resume. So, I decided to give FC6 development branch a go. Turns out, it was a good call :-)

I did try suspend to RAM briefly and that didn't go all that well, but I haven't played with this much as I'm more interested in suspend to disk anyway.

Copyright © 2006 - 2010 Bojan Smojver.
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation; with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts. A copy of the licence is here.