On 01/09/2003 10:24:10 PM ZE9 Izumi Tsutsui wrote:
>
>In article <87lm1v5sfz.fsf@power.cnet.aladdin.de>
>cpg@aladdin.de wrote:
>
>> panic: tlb_handler: invalid user-space access from kernel mode va=0x00000000, ssr=0x400000f0, spc=0x8c0c7ab4  proc=0x8c46d63c 
>onfault=0x8c0ae606
>> Stopped in pid 2094 (conftest) at       cpu_Debugger+0x6:       mov     r14, r15
>>
>> db> t
>> cpu_Debugger() at panic+0x90
>> panic() at tlb_exception+0x238
>> tlb_exception() at 0x8c000484
>> sh3/cache_sh4.h() at          0
>> db>
>>
>> Doesn't it happen for you?
>
>Hmm, I also had spontaneous reboots, and I wondered whether
>all TLB operations should be protected by splvm() or not.
>
>I'm using a patch which sprinkles splvm() in sh3/pmap.c:
>http://www.ceres.dti.ne.jp/~tsutsui/netbsd/sh3-pmap-20030109.diff
>but I'm not sure it's really worth.

It's not really worth :-(
It's worse than before. Previously it reliably dropped into ddb at the
same point while configuring. Now it simply reboots, but before
reaching this point. It reboots after configure prints "checking for
dlsym..."

regards,
chris