I compiled a -current kernel today just to see if everything still worked
on -current (I have been running ALPHA2 for quite a while now). The sources
were obtained by rsyncing to the repository this morning (5/10/2000).

Unfortunately, the kernel paniced pretty soon...

This is the dmesg output:
NetBSD 1.5G (HADES++) #3: Thu Oct  5 13:26:48 MEST 2000
    root@lwp_hades:/usr/local/syssrc/sys/arch/atari/compile/HADES++
Atari Hades (68060 rev.5 CPU/MMU/FPU)
total memory = 65536 KB
avail memory = 59312 KB
using 410 buffers containing 3280 KB of memory
mainbus0 (root)
clock0 at mainbus0: system hz 64 timer-A divisor 200/192
grfbus0 at mainbus0
   [ bla bla bla ]
scsibus0: waiting 2 seconds for devices to settle...
   [ after pretty much exactly 2 seconds: ]
panic: cpu_switch - Queue empty

I added the 'Queue empty' part myself (hence kernel #3 ;-) and it tracks to
to following part of cpu_switch:

        movl    %d0,%d1
        lslb    #3,%d1                  | convert queue number to index
        addl    #_C_LABEL(sched_qs),%d1 | locate queue (q)
        movl    %d1,%a1
        movl    %a1@(P_FORW),%a0        | p = q->p_forw
        cmpal   %d1,%a0                 | anyone on queue?
        jeq     Lbadsw0                 | no, panic

The following was output from a small ddb session:
panic: cpu_switch - Queue empty
Stopped at      _cpu_Debugger+0x6:      unlk    a6
db>t
_cpu_Debugger(0,0,0,152ff4,199ee4) + 6
_panic(287c,24edc,151690,0,2004) + 60
Idle(?)
_mi_switch(151690) + 18
_ltsleep(d024c,10,d0243,80,0,d0212,2bfdd4) + 176
_scsibus_config_interrupts(2bfdc0,134a000,2b736,14c524,0) + 34
_config_process_deferred(14c524,0,1a2ac,199fa8,18fa6) + 56
_configure(40,80,ffffffff,0,0) + 54
_main(199fb4) + 386

db>ps
PID             PPID       PGRP        UID S   FLAGS          COMMAND    WAIT
0                 -1          0          0 2 0x20604          swapper

Does this ring a bell for someone? Is more info needed?

Leo.