(Hi Miles!)
I've been doing that for over a year with the games I help
develop at Activision. Every time a game crashes, the end-user
is presented with a dialog box that shows the eight-digit hex
crc of the normalized call stack (which is usually the same
for all crashes caused by the same bug), and a field where
he can type in his email address or a comment; when he
presses "OK", the normalized call stack and comment are
sent to a central site for analysis. The central site
has a web database of crashes organized by build number,
module, and crc. Crashes with the same crc are grouped together
into a single web page. It's been invaluable for development,
and is turning out to be handy for customer support
(any time a user reports crc E2FE-F93D, our support
people know exactly what the problem is).
I have been wanting to do something similar for the
Linux kernel. Ideally, it should be able to work
at least on some systems without any extra hardware,
so people don't have to lift a finger to use it.
(That video memory log patch sounds like just the thing!)
It shouldn't be hard to make it configurable to
gather log data from floppy disk or from another
system that was logging via a serial or parallel port
as well.
On reboot, it would grab the log data from whatever
local or device device, analyze it with ksymoops,
toss in info about system configuration,
add in the sysadmin's email address if desired,
and send it off to (say) port 5000 of ksymoops.org.
The tabulated oops results would be available via
the web at www.ksymoops.org, and via a mailing list.
A way for developers to contact sysops without the
sysops's email address being made public might be handy
(nobody likes spam).
Too bad my job doesn't pay me to write tools to help stabilize
the Linux kernel :-( or I'd jump up and do it right now...
- Dan
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/