Subject: Failure to crash-dump: help?
To: NetBSD Current Users <>
From: Chris Ross <>
List: current-users
Date: 09/20/2007 18:53:51
  Hello all.  I have a sparc64 machine, currently running 4.0_RC1.   
My root, and everything else for that matter, are on a single  
RAIDframe device.  However, I want to be able to get/retrieve crash  
dumps, so I've configured my kernel to crash-dump to sd0b, one of the  
components of the raid0 device, and calculated to be "the swap part"  
of raid0.

   The problem I'm seeing is that dumps just aren't working, and I  
don't know why.  I'm seeing:

login: [halt sent]
cpu0: kdb breakpoint at 1297e20
Stopped in pid 534.1 (raid_parity) at   netbsd:cpu_Debugger 
+0x4:        nop
db> reboot 0x104
Frame pointer is at 0xe0016671
Call traceback:
128def0(1, d, 0, 1848400, e00171f0, 0, e0016741) fp = e0016741
10a17c0(104, 0, 185149c, 1851400, 1851490, 185146c, e0016801) fp =  
10a1240(e00172a8, 0, 1, e0017188, 72, 72000000, e00168d1) fp = e00168d1
10a16b4(180c128, 1320530, 0, 0, e0017398, 1f, e0016a21) fp = e0016a21
10a4cdc(1297e28, 0, 1, 10f47e71, 0, 0, e0016b11) fp = e0016b11
12991a0(0, 0, 0, 0, 713e800, 1000000, e0016be1) fp = e0016be1
1296c98(101, e0017b60, ffffffffffffffff, 0, 0, 188cc00, e0017141) fp  
= e0017141
1008b34(e0017b60, 101, 1297e20, 1d0006, 0, ffffffffffffffff,  
e00172b1) fp = e00172b1
127042c(140e400, 0, 80000000fd36f436, 80000000fd36f436, 1, 1d,  
e0017491) fp = e0017491
1270a88(61c2800, e0017e0c, 188cc00, 1, 0, 40, e0017551) fp = e0017551
1008f94(0, 5, e0017ed0, 1, 1270a40, 1000000, e0017621) fp = e0017621
10bffd0(74eb0a0, 0, 6f58, 1, 74d2000, 1000000, 10f231a1) fp = 10f231a1

dump to dev 7,1 not possible

   So, what does "dump to dev 7,1 not possible" actually mean?  While  
the machine is up, I can confirm that 7,1 is /dev/sd0b, and I can  
read and write that device just fine.  Anyone know of any better way  
to get information out of the kernel crash-dump code on *why* it  
thinks that's not possible?


                             - Chris