Dell Inspiron 2600

This came with a nasty force-fed license screen on the initial start-up, ie. Electronic Break The Seal.

Configuration:

ATAPI Slimline DVD-ROM DRIVE 8X/24X SD-C2502
DOS driver
Synaptics Touchpad
tpconfig
i82830M graphics
linux instructions
PIIX4 IDE controller
3Com 3C920 (3c59x driver, 3c905C-TX-M compatible)
82802CAM AC'97 Audio (rev 02)
1.2GHz Tualatin, 256KB cache
512MB SDRAM-133
28GB 4200rpm IBM Travelstar IC25N030ATCS04-0 (failed July 2005)
80GB 5400rpm Samsung MP0804H (July 2005)
Actiontec MD56ORD V92 MDC modem (driver)

enabling video:

With the exception of the Intel i830 video and the WinModem (neglected by choice), the Inspiron 2600 works perfectly with RedHat Linux 7.3.
The video can be made to work, bypassing the default 1MB video memory BIOS limitation, through the following:

patching a kernel (agpgart stolen memory support) and compiling AGP support (for both GLX and Xserver) without direct rendering (DRM/DRI) support, to be provided externally
patching XFree86 (more stolen memory support), compiling and copying the modules from (BUILD)/exports/lib/modules to /usr/X11R6/lib/modules
compiling the direct rendering kernel module (i830.o) from (xf86 build root)/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/Makefile.linux and installing it

These steps worked for me with the RedHat-provided 2.4.18-10 kernel and XFree86 4.2.0-8 (S)RPMS, although I had to merge versions of i810_dma.c, i810_drv.h, i830_dma.c and i830_drv.h forked between the kernel ((BUILD)/drivers/char/drm) and XFree86 sources first.

Note that as of no later than kernel 2.4.20, the first patch has been incorporated into the main kernel tree. As of 2.4.21-pre4, the last work-around is still required for DRI, however.

enabling sound:

With OSS drivers, the i830 audio seems fixed at 48000 KHz. The ALSA drivers fix this; however, past (at least) kernel 2.4.20, there are occasional static noise problems.

enabling usb:

This problem isn't restricted to this laptop, but it does show up and can be quite time-consuming to trace down. The RedHat-provided kernel 2.4.18-x.x sources have a small glitch preventing HID input from functioning properly. For USB input, the following options must be enabled:

USB Human Interface Device (full HID) support
HID input layer support
/dev/hiddev raw HID device support

As it is, however, the configuration scripts result in the following lines within .config:

CONFIG_USB_HID=m
CONFIG_USB_HIDINPUT=m
CONFIG_USB_HIDDEV=m

Of course, this needs to be corrected to:

CONFIG_USB_HID=m
CONFIG_USB_HIDINPUT=y
CONFIG_USB_HIDDEV=y

enabling acpi:

With the ACPI patches, this is almost too easy. Not quite, however; just like as for video, the BIOS seems to have broken AML code, as the kernel indicates at start-up:

    ACPI-1101: *** Error: Method execution failed [\_SB_.PCI0.LPCB.ACAD._PSR] (Node dfce26c0), AE_AML_REGION_LIMIT

Among other nice things, this has the lovely result of breaking /proc/acpi/event, rendering the event daemon acpid useless. Fortunately, the rest of the ACPI system works, such as the battery, lid and button indicators, which need to be polled. To some extent, suspend works, and I imagine the fan and thermal features do as well.
The work-around for the BIOS is supposedly to disassemble, fix, recompile and hardcode into the kernel source code the DSDT table (ie., /proc/acpi/dsdt) -- I haven't done this, but it would be nice to see it done.

After uptimes past a few hours, ACPI battery support fails (kernel 2.4.21-pre4-ac4, acpi patch acpi-20030122). Attempting to access /proc/acpi/battery/BAT1/info gives the kernel message

exresnte-0110 [38023] ex_resolve_node_to_val: No object attached to node c15932c8
acpi_battery-0134 [38018] acpi_battery_get_info : Error evaluating _BIF

while accesses to /proc/acpi/battery/BAT1/state lock the computer for about 10 minutes.

Compatibility Issues:

Unfortunately, the kernels included with tomsrtbt-2.0.103, superrescue 1.3.1a and superrescue 2.1.0 hang during start-up on this Inspiron, the former two freezing at identically the same point. RedHat 7.3's 2.4.18-3 kernel works flawlessly, however.

The stock 2.4.20 kernel in conjunction with XFree86 4.2.0-8 spontaneously crashes; this is related to the DRM issues, as given by this XFree86 log:

pgetbl_ctl: 0x1ff60001 pgetbl_err: 0x0
ipeir: 0 iphdr: 0
LP ring tail: 8 head: 0 len: 0 start 0
eir: 0 esr: 0 emr: ffff
instdone: ffc0 instpm: 0
memmode: 108 instps: 0
hwstam: 6000 ier: 3 imr: 6000 iir: a0
space: 65520 wanted 65528
(II) I810(0): [drm] removed 1 reserved context for kernel
(II) I810(0): [drm] unmapping 8192 bytes of SAREA 0xe09c7000 at 0x40014000

Fatal server error:
lockup

as well as this kernel log:

Jan 27 21:55:58 localhost kernel: [drm:i830_wait_ring] *ERROR* space: 65512 wanted 65528
Jan 27 21:55:58 localhost kernel: [drm:i830_wait_ring] *ERROR* lockup

Fortunately, there are no such problems with the ac kernels -- for instance, 2.4.21-pre4-ac4 is stable.

As for Windows, the Windows2000 CD refuses to boot. On the other hand, it's possible to access the CD-ROM drive with at least a Windows98 emergency boot disk.

Jonathan Lee

August 2002, January 2003