On Sat, Aug 20, 2005 at 02:35:35PM +0200, Vincent Blondel
wrote:>
> Hi,
>
> System : FreeBSD 4.10-RELEASE-p2
> Host : Tyan Thunder K7X (S2468GN)
> Cpu : 2 * Athlon MP 1800+
> Memory : 2 * 512Mb DDR-266 Memory
> Adaptec 5400s SCSI RAID Controller
>
> I got for the second time, this week, a system crash with my FreeBSD
server. You
> can find below what I get on the screen but I don't understand all what
it means.
> I just see there was a problem with the process imapd ( previous time I got
a
> segmentation fault with apache 1.3 )
>
> fault code = supervisor read, page not present
> instruction pointer = 0x8:0xc02a3e90
> stack pointer = 0x10:0xe3b22a10
> frame pointer = 0x10:0xe3b22a14
> code segment = base 0x10, limit 0xfffff, type 0x1b
> = DPL 0, press 1, def32 1, gran 1
> processor eflags = interrupt enabled, resume, IOPL = 0
> current process = 76856 (imapd)
> interrupt mask = none <- SMP: xxx
> kernel : type 12 trap, code=0
> db>
>
> I also remarked the keyboard focused on a process with 'db>'.
What is this and what can
> I do with it. I tried 'exit' 'quit' 'bye' but all
these statements don't do anything so I
> just pressed on button 'Reset' and rebooted the machine.
>
> Before starting the machine I checked the hardware by running
'memtest' during 6 hours and
> no errors were found. This seems to point the problem with my FreeBSD
installation better
> than a hardware ( example memory ) error.
>
> I also don't pretend my system is compiled with the best method. I
remember ( this is
> now a log time ago ... ) I compiled lot of things with '-O2 -pipe'
( except the
> kernel with '-O -pipe' ) but is this safe enough for an SMP system
with FreeBSD 4.x ???
>
> So can somebody explain me what does this mean and also if I have to
configure
> something to make a dump file when I get an error. I suppose this will be
> easier to investigate the problem if I still have other system crash.
>
> Regards
> Vincent.
Here's how to get a crash dump.
http://www.freebsd.org/doc/en_US.ISO8859-1/books/developers-handbook/kerneldebug.html
In the mean time, I suggest recompiling world and kernel without -O2 and
-O.
Cheers,
Jason