Bug#167009: Additional details

[John Goerzen <jgoerzen@complete.org>]

Hello,

I tried downgrading to 4.2.1-1 (from 4.2.1-3) to see if that would improve
matters.  It apparently did not, although after one sleep my mouse alone
remained responsive.

However, I discovered that the machine was still listening on its airport
interface and was able to log in and do some more debugging.

It turns out that when X throws one of these fits, it's spinlocking. 
Running strace on it yields this sort of result, repeated over and over:

ioctl(6, 0x20006444, 0)                 = -1 EBUSY (Device or resource busy)
ioctl(6, 0x20006444, 0)                 = -1 EBUSY (Device or resource busy)
--- SIGALRM (Alarm clock) ---
ioctl(6, 0x20006444, 0)                 = -1 EBUSY (Device or resource busy)
ioctl(6, 0x20006444, 0)                 = -1 EBUSY (Device or resource busy)
--- SIGALRM (Alarm clock) ---
ioctl(6, 0x20006444, 0)                 = -1 EBUSY (Device or resource busy)
ioctl(6, 0x20006444, 0)                 = -1 EBUSY (Device or resource busy)
ioctl(6, 0x20006444, 0)                 = -1 EBUSY (Device or resource busy)

Now, looking at the X server process, I see that fd 6 is attached to
/dev/dri/card0.  Maybe a clue.

I checked my kdm.log, XFree86 log, and syslog, and noticed nothing out of
the ordinary in any of them.  The first two had not logged any data since
the sleep/wakeup.

Hope this can shed some light on the matter.

-- John