Re: LOCKDEBUG panic on x68k

[Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost>]

isaki%pastel-flower.jp@localhost wrote:

> x68k GENERIC kernel + LOCKDEBUG cause panic in uvm_init().
> GENERIC without LOCKDEBUG passes here.
> 
> Here is the message:
> 
> panic: lockdebug_lookup: uninitialized lock (lock=0x254784)
> Stopped in pid 0.1 (system) at  netbsd:cpu_Debugger+0x6:    unlk a6
> db> bt
> cpu_Debugger(2704,21c8a0,0,2edd78,2edd8c) + 6
> panic(20623e,254784,8,1,fffffff4) + 172
> lockdebug_wantlock(254784,d6fa8,0) + 124
> rw_vector_tryenter(254784,1) + ee
> vm_map_lock_try(254780) + 34
> uvm_map_prepare(254780,0,1000,0,ffffffff,ffffffff,0,41727,2ede50) + 3e
> uvm_kmapent_alloc(254780,0) + c4
> uvm_mapent_alloc(254780,0,1,0,0) + 2a
> uvm_map(254780,2edf18,f000,0,ffffffff,ffffffff,0,801727) + 32
> uvm_km_alloc(254780,f000,0,11,c) + 56
> pmap_init(75100,ffc00000,2edf7c,2edf78,2632d8) + 8e
> uvm_init(c,0,0,ffffcffc,0) + 5c
> main() + 1e
> db>

I see the similar panic on news68k.

20080427 kernel (updated around 3:00am UTC) with pmap_motorola.c
fixes works with LOCKDEBUG, so some preemption changes after that
might cause this problem.
---
Izumi Tsutsui