Re: Help with crashdump

[walt <wa1ter@xxxxxxxxxxxxx>]

On Wed, 2 Mar 2005, Matthew Dillon wrote:

>     Is your userland synchronized with your kernel?

Yes -- compiled at the same time.

>     Is your /var/crash large enough to hold a memory dump? (at least as much free
>     space as physical memory ?)

/var has tons of space.  Swap is the same size as memory:  half a gig.


> 	savecore /var/crash

I get 'savecore: no coredump', and nothing in /var/crash.

Some observations:

Yesterday's kernel reports this:
ad0: 76319MB <ST380021A> [155061/16/63] at ata0-master UDMA100
and yet when I force the dump from the kernel debugger I see a message
about ad0: DMA limited to UDMA33 because of an old-style cable.
(and DFly is installed on ad1 and the dumpdev is also on ad1).

When I start the dump from the debugger prompt I see this:
dev #ad/0x60009 offset 128

The old kernel reports this:
$ ls -l /dev/ad1s5b
crw-r-----  1 root  operator  116, 0x00060009 Dec 18 12:57 /dev/ad1s5b

# swapinfo
Device          1K-blocks     Used    Avail Capacity  Type
/dev/ad1s5b        524160        0   524160     0%    Interleaved

This all looks correct to me.

When the dump starts, I see an orderly series of numbers starting at 511
and working down to 0.  The last few lines look approximately like this:

4 3 2 1 0 error2 0
succeeded
rebooting in 15 seconds

I'm wondering about that 'error2' -- does this mean there was an error
during the dump?

Red herring(?):  I'm running DFly from ad1s5a (a logical partition).
Everything has been working perfectly until this problem getting a
crashdump.  Any chance this could be causing the confusion?