>>
>> Thanks for the helpful responses. I was running netbsd.generic as my
>> kernel file
>
>[snip]
>
>> > > "savecore can't fine device 0/0" What does this mean? Is it something
>> > > Problem) netstat -r returns "netstat: kvm_read kvm_read: Bad address"
>> >
>> > They're related. The file /netbsd is not your current kernel. A number
>> > of programs, and libkvm, access /netbsd to learn what's going on
>> > in the kernel. If /netbsd isn't the active kernel, they look in the
>>
>> I renamed kernels appropriately and now the problem went away. Thanks.
>
>I've wondered about this problem for a while. I understand why
>programs like ps and route are happy to look in a kernel for
>info (for debugging of crashes), but I'm always frustrated by
>their default of looking at /netbsd.
>
>My thought is that they look at /netbsd when they want the real kernel
>(when we're looking at the live kernel). But isn't the real kernel
>in RAM at that time? Isn't all the info that was in the disk-version of
>the running kernel still in RAM?
>
>So my thought is to add another file to kernfs, which would look at
>the live kernel & make it look like it did on disk.
>
>I'd volinteer to do this, but I'm a ways down on the fs & vm learning
>curves & I'm working on tty stuff for the Mac68k port. But if it's
>not too hard & someone can tell me what to do, I might try.

I was thinking of a simpler solution.  How aboiut we never have a kernal
called /netbsd.  Instead call it something else, I have have 5 or 6
different /netbsd.generic's, what ever, it really doesn't matter.  Any way
call it something else, use the booters name of kernal feature, and modify
the rc scripts to remove /netbsd and replace it with a hardlink to which
ever of the several particular kernals at the root level of the drive is
running.

---------
Christopher R. Bowman
crb@eng.umd.edu
My home page