Another update, using this morning's kernel.

The network interfaces are fine for me, though I haven't tried WPA -- my
access points aren't set up for it.

USB is totally dead after a resume.

Sound works; here's the relevant dmesg:

auich0 at pci0 dev 31 function 5: i82801DB/DBM (ICH4/ICH4M) AC-97 Audio
auich0: interrupting at irq 11
auich0: ac97: Analog Devices AD1981B codec; headphone, 20 bit DAC, no 3D
stereo auich0: ac97: ext id 601<AC97_22,AMAP,VRA>
auich0: measured ac97 link rate at 47997 Hz, will use 48000 Hz
audio0 at auich0: full duplex, mmap, independent

(non-working auich1 modem elided.)

When I'm in X and suspend, on resume I see a blank screen with a blinking
VGA-type cursor in the upper left.  I can restore my X by switching
consoles away from screen 5 and then back.  That works programatically,
using switchtoscreen, so if there was a 'resume' script I could invoke
it.  However, when I get back to X there's some interference -- many thin,
multicolor lines -- going horizontally across the picture.  My assumption
is that the frame buffer isn't being refreshed properly.  The density of
the line pattern seems to depend on the background picture -- I tend to
have random train pictures of various sizes being displayed as wallpaper...
(The chip is a ATI Radeon Mobility 9600/9700 (M10/M11) NP (AGP), according
to Xorg.0.log.)

I wrote simple lid_switch and sleep_button scripts; both work properly for
suspending if I have them set sleep_state via the sysctl.  However, the
machine does not resume when I open the lid, even though it does under APM
or Windows.  If I comment out the sysctls, I see 'pressed' and 'released'
events for lid_switch; for sleep_button, I only see the 'pressed' event.

A brief test of cardbus slots after resume succeeded -- my EVDO card was
able to dial out.

Speedstep, via estd, is *not* working; the speed stays at 1800 Mhz.