I said I'd never blog

First, I submitted a new version of my Enhanced Speedstep patch for NetBSD/amd64 today. It certainly has a much better chance to hit the 4.0_BETA2 tree this time.

Second, here is a quick script to tell estd to update its policy when you play with the plug:

#!/bin/sh -

case "${2}" in
"pressed")
	pkill -USR2 estd estd
	;;
"released")
	pkill -USR1 estd estd
	;;
*)
	logger -p warning "${0}: unsupported event ${2} on device ${1}" >&1
	exit 1
	;;
esac

Copy this into /etc/powerd/scripts/acadapter.

Yes, insomnia again... But this time is maybe because I'm getting a bit sick, as it's getting colder here. Anyway:

• port-amd64/36591 [non-critical/low]: wpi(4) does not compile on amd64 with a fix (easy)
• kern/36592 [serious/medium]: cdce does not work: arpresolve: can't allocate llinfo on cdce0 with a fix (not from me)
• pkg/36597 [serious/medium]: sysutils/wmSMPmon does not compile with X11_TYPE=modular fixed already
• pkg/36602 [serious/medium]: qemu does not compile on NetBSD before 4.0 with a fix (quite ugly but apparently necessary)

On the amd64 front

I received my new Lenovo T60. It's an Intel Core 2 Duo, the one with the controversial errata. For example, Linus Torvalds does not think it's so bad, while Matthew Dillon had more comments on the issue.

So far so good, it's behaving almost identically like my previous T60. The main differences:

• it has EM64T
• Fn+F5 (bluetooth disable) now works without the need for a special ACPI driver (BIOS setting?)
• the wireless switch on the front also controls the presence of the bluetooth device (BIOS setting again? true hardware control?)
• the power supply not only whistles but also overheats

Surprisingly the default amd64 kernel configuration is not as laptop-friendly as its i386 equivalent. This is probably going to be one of my next PRs. Meanwhile, I am successfully using these additional settings:

#	$NetBSD: GENERIC.local,v 1.1 2003/04/26 18:39:34 fvdl Exp $
#
#	GENERIC.local -- local additions to the GENERIC configuration
#

acpiacad* 	at acpi?
acpibat* 	at acpi?
acpibut* 	at acpi?
acpiec* 	at acpi?
acpilid* 	at acpi?
acpitz* 	at acpi?

cdce*	at uhub? port ?
wpi*	at pci? dev ? function ?

options		ENHANCED_SPEEDSTEP

pseudo-device	cgd		4	# cryptographic disk devices

I am monitoring the battery status with my hacked wmapm.

Some issues remain though.

Graphics

There is no driver for the Radeon X1400 besides VESA yet, but check this out, some guys are apparently busy at making the driver (X1300 for sure). I was crazy enough to make it compile with the modular xorg 7.2 headers instead, but I'm not completely done with the glue code. I can submit it on demand. I think it would be cool to be able to test it with a "stock" 7.2 Xorg server, and it's definitely great to be able to compile it separately from the Xorg tree.

Oh, and you will need the libpciaccess as well. It compiled, but I don't know if if actually works on NetBSD.

Wireless

It's still crashing with a "fatal firmware error" everytime I issue a command while the card is up. I could not find what was wrong a few months ago, I'll probably try again when I have the chance. It was otherwise working with WEP enabled, I suppose this is still the case.

PCMCIA

It was crashing my old machine very easily, and now there is simply no driver attaching. Maybe it's just a matter of enabling it in the kernel. I was told the crashes were fixed though.

Sound

I think some volume controls are new or changed, but it no longer requires the initialization hack (something like outputs.direction = lineout). The driver is still lacking DMA support IIRC, and I still have to test it in full-duplex mode.

ACPI and Suspend / Resume

I did not try it yet, but I suppose it is still not resuming. If it does not work I will have a deeper look as well. Regarding the ibm-acpi driver, I think it would make more sense to propose sysctl controls instead of directly hooking the function keys. Plus, I was having issues with the responsiveness and with reading hotkey code values lately.

The DSDT is still broken. I'll have a deeper look later.

Conclusion

The NetBSD/amd64 port definitely needs more work but is in a good shape. A few packages are not available (misc/openoffice2) and others crashing (everything embedding firefox, complaining about Undefined PLT symbol "__clzdi2" in libmozjs.so), but so far my machine is stable and fast. That's all I need for now.