[oi-dev] Xorg hang bug, advice needed

Marion Hakanson hakansom at ohsu.edu
Fri Dec 9 02:37:32 UTC 2011


Back in October, I reported OI bug #1625, which is about Xorg/nVidia hang
behavior, and have been attempting to collect debug info.  Here's the URL
of the bug report:


Anyway, the kernel crash dump that I mention having collected, seems
to not be very useful.  It's not clear why, but it doesn't seem to have
saved a useful stack backtrace.  Since then I've used "dumpadm" to ask
it to save all pages, instead of the default kernel pages only, but I've
also failed to get it to dump at all, even though the hang has occurred
a few more times.

Enabling "mdb" doesn't work, because when the hang happens, the system
does not respond to the keyboard, so the magic keystroke doesn't do
anything useful (or if it does, you can't see it on the display due
to the Xorg/nVidia issue having made it unresponsive).

A "reboot -d" from a remote ssh login simply reports that it was unable
to stop the gdm service (probably because Xorg is hosed), and that it's
reverting to a "normal reboot", whatever that is.  I've also tried with
"reboot -l -n -q -d", but that has the same result as above.

I thought I at least half-way knew my way around this type of situation,
since I've forced crash dumps via mdb, etc. a number of times on servers,
but those all have serial consoles, not actual video cards and USB keyboards.

So, I could use recommendations on how to get a useful crash dump when
this hang occurs.  Suggestions?

Thanks and regards,


More information about the oi-dev mailing list