Panic, and problems with crash dump

[John Klos <john%ziaspace.com@localhost>]

Hi,

Since switching my SunFire V245 to NetBSD-9, it has paniced several times 
while compiling pkgsrc packages. Note that SSIZE still needs to be 
increased to boot.

The last time it paniced, I tried to get a crash dump using sync:

1 tt=30 tstate=4411001503 tpc=0x1001488 tnpc=0x100148c
2 tt=30 tstate=9982000601 tpc=0x14b63c0 tnpc=0x14b63c4
Stopped in pid 25722.1 (sh) at  1010438:        nop
db{1}> sync
Frame pointer is at 0xe004edf1
[ 433883.4489836] Call traceback:
[ 433883.4489836]  netbsd:cpu_reboot+0x244(a, 1c888b0, f, ffffffffffffffff, f, 1c67400) fp = e004eed1
[ 433883.4489836]  netbsd:db_sync_cmd+0x20(100, 0, 1c38400, 1cad800, f, 1138fe180) fp = e004ef81
[ 433883.4489836]  netbsd:db_command+0xac(0, 0, 2, e004f8f0, 1c383d8, 1c9cf28) fp = e004f031
[ 433883.4489836]  netbsd:db_command_loop+0xb8(1c383b8, 1c9cc00, 1c38000, 0, 173fec0, 1c5c8c0) fp = e004f171
[ 433883.4489836]  netbsd:db_trap+0x104(101043c, 1000000, 17e39b0, 1c38400, 1c9cc00, 1c383b8) fp = e004f251
[ 433883.4489836]  netbsd:kdb_trap+0xdc(101, 0, 2, e0048000, 1c9dc00, 0) fp = e004f311
[ 433883.4489836]  netbsd:trap+0x39c(101, e004fe20, f, 1c02800, 1c00000, 1cb5800) fp = e004f3c1
[ 433883.4489836]  netbsd:101155c+0(e004fe20, 101, 1010434, 140414, 1c5c8c0, e0048000) fp = e004f571
[ 433883.4489836]  netbsd:10d118+0(1c010b0, 6, 7, 23c304, 8, 402774d8) fp = e004f751
[ 433883.4489836]  netbsd:0+0(0, 0, 0, 0, 0, 0) fp = ffffffffffff9521

[ 433883.4489836] dumping to dev 7,9 offset 95
 16382 M

The problem is that I was watching the number of megabytes count down, and 
it appears to have been going at about 17 MB per minute, so it would've 
taken sixteen hours or so to finish.

Swap is on a Samsung SSD on the internal LSI Fusion controller, which 
normally writes at about 120 MB/sec. Any thoughts about why a crash dump 
is writing at less than 1/3 a megabyte a second?

Also, any ideas about why the panics are occurring in the first place? If 
waiting the sixteen hours for a crash dump would help someone, please let 
me know.

Thanks,
John