Subject: Re: NetBSD on Thinkpads
To: Brook Milligan <brook@biology.nmsu.edu>
From: Steven M. Bellovin <smb@cs.columbia.edu>
List: current-users
Date: 12/08/2007 22:51:28
On Thu, 29 Nov 2007 08:35:53 -0700
Brook Milligan <brook@biology.nmsu.edu> wrote:

> I am trying to determine if the new Thinkpads (e.g., R61, T61, X61)
> will work well with NetBSD.  From what I can tell, the various models
> offer permutations of the following devices:
> 
> - Intel 82566MM ethernet
> - Intel 82573L ethernet
> - Broadcom BCM5787M ethernet
> 
> - Intel 3945ABG wireless
> - Intel 4965AG wireless
> - Atheros AR5006EX wireless
> 
> I am unclear about how much of this is supported by NetBSD drivers.
> 
> Does anyone have direct experience with these laptops and NetBSD?
> Does anyone have insight about the driver support?
> 
I have a T61 with the 3945ABG wireless card.  It works fine, as does my
gigE card (sorry, I don't remember which it is).  However, when I last
tried it -- admittedly, several weeks ago -- there were enough things
that didn't work properly that I've been experimenting with Ubuntu
Linux instead.  Below is a summary of the NetBSD problems I encountered
as of 19 November.  I do not know which have been fixed.  I will note
that the USB interrupt storm problem affects Linux, too, and the cause
is *not* understood.  However, Linux eventually turns off the port.
While this kills the USB ports, it at least gives back the 75% or so of
the CPU that was going to interrupt servicing...  I also never got the
Cardbus slot to work; scanning this thread, I see that there is a
work-around -- I'd assumed it was an RBUS_MIN problem.  I'll add that
at the time, I was sufficiently unhappy with *all* choices that I was
wondering if I should have gotten a Mac instead.  Since then, I tried a
bleeding-edge Linux kernel; that plus some useful advice has made Linux
behave a lot better.

----------- (from 19 Nov)
	I tried out the power management branch of NetBSD-current.  I
	applaud the efforts of Jared and company, but there were still
	major gaps as of the last kernel I tried (this is as of about
	two weeks ago; I should build another LiveCD and try some of
	this again). Leaving out the *serious* bug in ahcisata support
	-- briefly, don't use it; use compatibility mode instead, if
	you value your disk -- suspend/resume has problems.  You *must*
	switch to a text-mode console before suspending if you're in X,
	though that can be automated via powerd. You lose the USB ports
	after a resume. There's a strange interaction between the
	touchpad and the audio driver.  If you have a CD mounted during
	suspend, you'll have to unmount/remount to use it on resume.  It
	doesn't resume when you open the lid; you have to hit Fn.
	There's no way to hot-swap devices in the Ultrabay, though this
	is a generic NetBSD problem.  Dual-core CPUs aren't supported,
	though that support is probably close. There are sometimes
	interrupt storms that appear to be associated with one of the
	USB controllers.


		--Steve Bellovin, http://www.cs.columbia.edu/~smb