Subject: Re: Nasty user trap problem on Alpha...
To: None <mellon@hoffman.vix.com, cgd@pa.dec.com>
From: Matthew Jacob <mjacob@feral.com>
List: port-alpha
Date: 06/03/1997 13:05:05
>
>> fatal user trap:
>>
>> trap entry = 0x2 (memory management fault)
>> a0 = 0x0
>> a1 = 0x1
>> a2 = 0xffffffffffffffff
>> pc = 0x0
>> ra = 0x0
>> curproc = 0xfffffe004a647000
>> pid = 3692, comm = getty
>
>are you wondering about the trap itself, or the spew?
>
>Unfortunately, the information present is just about useless for
>debugging the trap, except that it looks like it's dereferenceing
>NULL, perhaps by jumping to zero. I suggest you rlogin to the machine
>and attach to the process with gdb. (it won't dump core since it's
>setuid.)
>
>As for the spew, i'm going to remove that in most situations. kernels
>_should not_ spew random crap to the console if a program happens to
>crash. (*chuckle* it actually opens the door wider for an obscure
>denial of service attack.) I'll leave it in #ifdef DEBUG, but will be
>removing DEBUG from the GENERIC kernel.
>
I'll 'fess up to adding this 'spew'. Sorry- *I* find it useful- but
I'm happy to have it under DEBUG.
So shoot me, awready......