Subject: Re: -current broken?
To: None <>
From: maximum entropy <>
List: port-pmax
Date: 05/01/2000 01:23:24
>From: Simon Burge <>
>Date: Mon, 01 May 2000 13:38:16 +1000
>maximum entropy wrote:
>> A GENERIC kernel compiled from today's sources hangs part way through
>> the boot process on my 5000/133.  The last thing displayed before the
>> freeze is "root file system type: ffs".
>> If more information is needed to debug this, let me know.
>Can you try deleting the following two lines
>	jal     _C_LABEL(uvm_pageidlezero)
>	nop
>from sys/arch/mips/mips/locore.S to see if it makes things better?

That made a significant difference, but there's still a problem.  The
system now makes it well into the multiuser boot, before it hangs.  By
carefully editing rc.conf and rc.local to reduce the number of
programs started during boot, I was able to make it all the way to the
point where it displays the date, but didn't get a login prompt.

After about 10 minutes sitting like that, it panicked and dropped into
DDB.  The following is transcribed by hand.

trap:  TLB miss (load or instr. fetch) in kernel mode
status=0xff00, cause=0x80002008, epc=0x80001004, vaddr=0x7fff9204
pid=1 cmd=init usp=0x7fffeca0 ksp=0xc9158f18
Stopped in init at      0x80001004:     b       <_DYNAMIC_LINK+7fff9254>
                bdslot: 0x80001008:     b       <_DYNAMIC_LINK+7fff927c>
                bdslot: 0x8000100c:     nop

My userland is fairly old, could that be causing this?  The old kernel
(which works fine) and the userland are both from December 1999.

entropy -- it's not just a good idea, it's the second law.