Dump already in progress, bailing... (after halt on ServeRAID-6M via ips.ko)

[Tomaž Borštnar]

YONETANI Tomokazu wrote:
On Wed, Oct 13, 2004 at 01:16:00PM +0200, Tomaž Borštnar wrote:

Matthew Dillon wrote:

:Se
  Assuming you have DDB compiled into your kernel, set this sysctl:
  sysctl -w debug.trace_on_panic=1

  Then hopefully it will print a stack backtrace when it panics, before
  it locks up.
would DDB_UNATTENDED go against this setting? ips driver can write 
crashdumps to swap?


The current ips driver can't crashdump to a ServeRAID partition.
i thought so

I failed to reproduce this today with DragonFly_Stable on a 2-CPU SMP
machine, so unless DragonFly_Stable tag moved within a day, it must be
something else that you have and I don't(you use USB keyboard, and I use
PS/2 for instance; not to mention that your 325 is many times faster
than ours), rather than the driver. You didn't at least see this panic
when you tried my devstat patch, how current was your source tree at that
time?
This happened for the first time, but what worried me that it did not 
stop and that keyboard was dead (of course, since usbd was dead!). I 
used sources as of yesterday.

Your devstat patch seems to run just fine.

What I can do is hook up a null modem cable to machine to see any last 
output via another machine. Do I need anything to configure this or is 
it automatically? Machine has one serial port for such purposes.

Tomaž
Tomaž