Subject: Re: Symbol lookups.
To: Greywolf <>
From: Andrew Brown <>
List: tech-userlevel
Date: 05/09/2003 23:36:17
>AB> >Ok, nobody seems to have any special objections, so I will change the
>AB> >kvm/nlist routines to use the /dev/ksyms interface instead, and retire
>AB> >the old kvm.db stuff.
>AB> even thought it had the nice side effect of providing a simple single
>AB> place to look for kernel symbols?  perhaps kvm_mkdb should be changed
>AB> to something that copies the kernel to a known place?  /dev/ksyms is
>AB> still an option...
>If I understand Ragge correctly, the kvm.db was rather slower than the
>improved direct lookup code.  /dev/ksyms (or, failing that, kernel-nlist)
>would be a preferable interface since we would then not _need_ to maintain
>it in two separate places.  This is well evidenced in /etc/mtab vs. the
>kernel mount table, and /etc/rmtab vs. the kernel export list.  I'm sure
>if it were pertinent to kernel-specific information which needed to be
>shared with userland, we'd have utmp in the kernel as well.  Part of me
>still kind of wonders why utmp *isn't* there, but leave me alone -- I'll
>figure it out :).

i'm just concerned that /var/db/kvm.db and /dev/ksyms are standard
locations, whereas the actual location of the booted kernel is not.

and that lots of programs, when given a namelist file, will drop all
privs, so most things won't work for the average user if the standard
locations don't work.

come to think of it, lots of things seem not to be working anyway, but
i'm sure that's not related to this specifically...  :)

|-----< "CODE WARRIOR" >-----|             * "ah!  i see you have the internet (Andrew Brown)                that goes *ping*!"       * "information is power -- share the wealth."