Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 4 Jun 1998 22:02:50 +0000 (GMT)
From:      Terry Lambert <tlambert@primenet.com>
To:        mns@pobox.com (Mark Shepard)
Cc:        dyson@FreeBSD.ORG, mike@smith.net.au, jonny@jonny.eng.br, mike@dingo.cdrom.com, hackers@FreeBSD.ORG
Subject:   Re: kernfs/procfs questions...
Message-ID:  <199806042202.PAA14257@usr08.primenet.com>
In-Reply-To: <199806041951.MAB17978@proxy4.ba.best.com> from "Mark Shepard" at Jun 4, 98 12:49:14 pm

next in thread | previous in thread | raw e-mail | index | archive | help
> procfs/kernfs is essentially a kernel API in disguise.  Considered as an
> API, sysctl seems "cleaner" and lighter-weight than procfs/kernfs --
> there's no need to format/parse ASCII, no need to generate column headers,
> and FreeBSD's sysctl provides data in a more uniform format than Linux's
> procfs.

The main issue I take with this is that "ps, w, who, and netstat break
when kernel structures change".

It may be correct to point at sysctl and say "yea, verily, this is
light weight", but to paraphrase Eddie Murphy, what has it done for
me lately?

I have a *real* problem with the externalization of kernel structures
as API.

One camp wants the features left alone because they don't want to roll
the code into kgdb ("crash").  Another wants the /dev/kmem dependence
taken out, come hell or high water, and screw the ability to ps system
dumps.

I don't think either camp is right (of course 8-)), and would prefer
binding an absracted interface to the kernel object file, and requiring
the use of ELF section tags, but I am a minority of one.


					Terry Lambert
					terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199806042202.PAA14257>