Re: dump device

At Sequent, we used a 'stand alone dumper'. That is, the OSwould simply
halt, and return to the bios. The bios would check to see if a valid kernel
image existed before initializing memory. If it found a valid image, it
would launch a dumper utility. This allowed us to get good dumps no matter
how corrupt the operating system had become ( this is a bit of a
simplification of the acutual process, but it is essentially correct).

Some vendors have the OS do the dump. I agree that there is some risk here.
I believe the SGI kernel dump does indeed have the paniccing OS do the
dump.

We are unlikely to see a stand alone dumper for linux, as we are unlikely
to get bios support for such a function on all of the hw that supports
linux.

I cannot comment on what AIX does... I've only been with IBM a short time,
and have not looked into the crash dump functions.

 - jim

Nilesh Patel/India/IBM@[e-mail] on 06/29/2000 03:48:48 PM

Sent by:  [e-mail]

To:   Mark Hahn <[e-mail]>
cc:   [e-mail]
Subject:  Re: dump device

no, for the simple reason that if the system is crashed,
it's unsafe to attempt _any_ kind of work, including IO
to a dedicated device.

-- But here also we still try to write something to /var/log/messages
-- Some kind of traceback, and system is still unsafe !!

there's at least one patch, not in the mainstream kernel,
that does crashdumps.

how can you trust any IO when the system is corrupt?

-- But Aix have such a facility. Then how does Aix do this stuff ?