Re: tried to isntall NetBSD-6.1_RC2 on a VS3100 M38

[Martin Husemann <martin%duskware.de@localhost>]

On Sun, Mar 31, 2013 at 11:26:19AM +0200, Holm Tiffe wrote:
> http://www.tiffe.de/other/newfs.core
> 
> This should be the last core from yesterday eavening...

Looks like this:

Core was generated by `newfs'.
Program terminated with signal 4, Illegal instruction.
#0  0x7f7aa682 in getrawpartition () from /lib/libutil.so.7
(gdb) bt
#0  0x7f7aa682 in getrawpartition () from /lib/libutil.so.7
#1  0x7f7a887d in ?? (2147483406, 0, 174724, 1024, 0, 72476)
   from /lib/libutil.so.7
#2  0x7f7a89e1 in opendisk (2147483406, 0, 174724, 1024, 0)
   from /lib/libutil.so.7
#3  0x00012cb3 in main (2, 2147479124, 2147479136)

and objdump -d says:

0000a680 <getrawpartition>:
    a680:       00 00           .word 0x0000 # Entry mask: < >
    a682:       c2 14 5e        subl2 $0x14,sp
    a685:       d0 01 ad f0     movl $0x1,0xfffffff0(fp)
    a689:       d0 18 ad f4     movl $0x18,0xfffffff4(fp)
    a68d:       d0 04 ad ec     movl $0x4,0xffffffec(fp)
    a691:       d4 7e           clrf -(sp)
    a693:       d4 7e           clrf -(sp)
    a695:       9f ad ec        pushab 0xffffffec(fp)
    a698:       9f ad f8        pushab 0xfffffff8(fp)
    a69b:       dd 02           pushl $0x2
    a69d:       9f ad f0        pushab 0xfffffff0(fp)
    a6a0:       fb 06 ff 21     calls $0x6,*1fec8 <_GLOBAL_OFFSET_TABLE_+0x254>
    a6a4:       58 01 00 
    a6a7:       d5 50           tstl r0
    a6a9:       18 04           bgeq a6af <getrawpartition+0x2f>
    a6ab:       d2 00 50        mcoml $0x0,r0
    a6ae:       04              ret
    a6af:       d0 ad f8 50     movl 0xfffffff8(fp),r0
    a6b3:       04              ret


I see nothing wrong with that, and the only other case where I see the
kernel sending SIGILL would be a broken stack pointer:

(gdb) info reg
r0             0x5      5
r1             0x56650276       1449460342
r2             0x0      0
r3             0x7fffedac       2147478956
r4             0x0      0
r5             0x0      0
r6             0x1      1
r7             0x2aa84  174724
r8             0x2      2
r9             0x0      0
r10            0x7fffff0e       2147483406
r11            0x400    1024
ap             0x7fffd540       2147472704
fp             0x7fffd504       2147472644
sp             0x7fffd504       2147472644
pc             0x7f7aa682       2138744450
ps             0x3c00000        62914560

Frame #1 seems to be midway in parsedate (in libutil):

    886d:       d0 16 60        movl $0x16,(r0)
    8870:       d2 00 58        mcoml $0x0,r8
    8873:       31 2a 01        brw 89a0 <__parsedate50+0x5c2>
    8876:       fb 00 ff 17     calls $0x0,*1fd94 <_GLOBAL_OFFSET_TABLE_+0x120>
    887a:       75 01 00 
    887d:       d0 50 56        movl r0,r6

but this does not make any sense, unless the stack has been smashed.

Any ideas? Can anyone reproduce it?

Martin