Power Management is (Still) Hard
I tried, for the first time, to put the computer to sleep. I wanted to start doing that at night, since my computer really isn't doing anything vital at night, and each computer that's on 24/7 is probably costing me $25 or so per month in power.
And this led to my first Vista BSOD (Blue Screen of Death): DRIVER_POWER_STATE_FAILURE.
The BSOD screen didn't have any information on what driver caused the problem, which is unusual since Microsoft is usually so proactive about pointing the finger away from the OS when something goes wrong. I figured it was getting logged somewhere and I'd get more information when I rebooted.
When Vista booted back up, it gave me this friendly message that basically said "There was a problem, we'll let you know when we have a solution". No more information.
Looking in the Event Log the only signs I got that something unusual had happened were:
The previous system shutdown at 11:46:49 PM on 26/02/2007 was unexpected.
and
The computer has rebooted from a bugcheck. The bugcheck was: 0x0000009f (0x00000003, 0x85e13b98, 0x870e8030, 0x88b58ac0). A dump was saved in: C:\Windows\MEMORY.DMP.
The dumpchk tool from Debugging Tools for Windows pointed the finger at the real culprit:
BugCheck 9F, {3, 85e13b98, 870e8030, 88b58ac0}
Probably caused by : athr.sys
I looked up athr.sys and turns out it's driver for the 802.11 chipset in my computer. I have a wired connection and don't need it, so I'm just going to shut it off and see if that helps. Note that dumpchk needed the symbol server set up to provide the right information - it said the problem was in another module when it didn't have symbols, though it's hard to miss - it fills the dump output with warnings that the symbol server isn't set up in that case.
(It works - after 23 seconds of disk grinding, my computer went to sleep and successfully worke back up!)