Ok, more details:

I can reproduce the problem by booting (sh cores when runing rc.d/network,
but running "sh /etc/rc.d/network start" manually works) and by doing
cd /usr/src; make cleandir; make includes (sh cores in /usr/src/gnu/lib/lbbfd,
but again a cd there and make includes works).

Examining the core shows the stack completely borked at the end of
jobs.c:forkschell:

Program terminated with signal 11, Segmentation fault.
#0  0x10bafc in forkshell (jp=0x0, n=0xffffffffffffffff, mode=0) at jobs.c:672
672     }
(gdb) print pid
$1 = 0
(gdb) bt
#0  0x10bafc in forkshell (jp=0x0, n=0xffffffffffffffff, mode=0) at jobs.c:672
#1  0xc in ?? ()
#2  0xc in ?? ()
Cannot access memory at address 0x3100000000000070

This is from the gnu/lib/libbfd fault, I don't know how far I can trust gdb
here. From the booting fault I get this:

Program terminated with signal 11, Segmentation fault.
#0  0x10bafc in forkshell (jp=0x0, n=0xffffffffffffffff, mode=2648144)
    at jobs.c:672
672     }
(gdb) print pid
$1 = 0
(gdb) bt
#0  0x10bafc in forkshell (jp=0x0, n=0xffffffffffffffff, mode=2648144)
    at jobs.c:672
#1  0x286818 in handler ()
#2  0xc in ?? ()
Cannot access memory at address 0x3100000000000070

Maybe related, maybe not: on boot I get the following messags:

No counter-timer -- using %tick at 269MHz as system clock.
cpu_fork: nsaved 32
root on wd0a dumps on wd0b
root file system type: ffs
cpu_fork: nsaved 32
cpu_fork: nsaved 32
cpu_fork: nsaved 32
cpu_fork: nsaved 32
Alignment error: dsfsr=00000000:00800001 dsfar=0:13 isfsr=00000000:00000000 pc=10baf8

The cpu_fork debugging messages are new, and the alignment error too (I don't
get that for the gnu/lib... case).


Martin