127.0.0.1 deprecated??

[Mouse]

I just now had occasion to look at lo(4), for the first time in I don't
know how long.

All three versions I have on hand - 1.4T, 4.0.1, and 5.2 - have a BUGS
section saying

BUGS
Previous versions of the system enabled the loopback interface automati-
cally, using a nonstandard Internet address (127.1). Use of that address
is now discouraged; a reserved host address for the local network should
be used instead.

This is the first I've heard of 127.1, a very old syntax for 127.0.0.1,
being nonstandard for loopback use or of its being deprecated, and,
indeed, the startup scripts for each of those three versions are
hardwired to bring lo0 up as 127.0.0.1.

Anyone have any idea what's behind that paragraph?

/~\ The ASCII Mouse
\ / Ribbon Campaign
X Against HTML ***@rodents-montreal.org
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[matthew sporleder]

Post by Mouse
I just now had occasion to look at lo(4), for the first time in I don't
know how long.
All three versions I have on hand - 1.4T, 4.0.1, and 5.2 - have a BUGS
section saying
BUGS
Previous versions of the system enabled the loopback interface automati-
cally, using a nonstandard Internet address (127.1). Use of that address
is now discouraged; a reserved host address for the local network should
be used instead.
This is the first I've heard of 127.1, a very old syntax for 127.0.0.1,
being nonstandard for loopback use or of its being deprecated, and,
indeed, the startup scripts for each of those three versions are
hardwired to bring lo0 up as 127.0.0.1.
Anyone have any idea what's behind that paragraph?
/~\ The ASCII Mouse
\ / Ribbon Campaign
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B

http://www.freebsd.org/cgi/man.cgi?query=lo&apropos=0&sektion=0&manpath=2.10+BSD&arch=default&format=html

Probably lost to history? That's the man page for 2.10 BSD
3rd Berkeley Distribution August 1, 1987 LO(4)

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Dennis Ferguson]

Post by Mouse
BUGS
Previous versions of the system enabled the loopback interface automati-
cally, using a nonstandard Internet address (127.1). Use of that address
is now discouraged; a reserved host address for the local network should
be used instead.
This is the first I've heard of 127.1, a very old syntax for 127.0.0.1,
being nonstandard for loopback use or of its being deprecated, and,
indeed, the startup scripts for each of those three versions are
hardwired to bring lo0 up as 127.0.0.1.

In the first assigned numbers RFC that had the address space divided into
classes, here

http://tools.ietf.org/html/rfc790#page-4

network 127 was reserved for an unspecified purpose along with the last
class B and class C networks. 127.0.0.1 was just arbitrarily chosen for
the 4.2 BSD lo(4) interface, there was no standard for that. The 4.2
man page says as much:

http://modman.unixdev.net/?sektion=4&page=lo&manpath=4.2BSD

The first assigned numbers RFC where network 127 was assigned to
loopback was from the end of 1986:

http://tools.ietf.org/html/rfc990

Some of the other things they just chose to do in 4.2 BSD, for example
using the all-zeros host address on a subnet as the broadcast address,
ended up being changed later, and I remember the 127.0.0.1 loopback was
not a big hit early on because it was commonly seen on the wire a lot
then and it was hard to figure out who was doing it. I would guess the
comment on the current man page came from the time when the sentiment
was in favor of having all configured local addresses be real, unique
addresses, and before they officially standardized network 127 with
the (always repeated) restriction that it never appear on the wire.

Dennis Ferguson
--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Mouse]

Post by Mouse
BUGS
Previous versions of the system enabled the loopback interface automati-
cally, using a nonstandard Internet address (127.1). Use of that address
is now discouraged; a reserved host address for the local network should
be used instead.
This is the first I've heard of 127.1 [...] being nonstandard for
loopback use or of its being deprecated [...]

[probably current back about BSD4.1c-BSD4.2 timeframe]

Ah! Okay, that explains it - it's a very old historical artifact, no
longer accurate.

Thanks for explaining. (And I thought _I_ remembered the days way
back; I suppose 'twas a bit of hubris deserving of something like this
to remind me that I'm actually a newcomer in some respects....)

/~\ The ASCII Mouse
\ / Ribbon Campaign
X Against HTML ***@rodents-montreal.org
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Robert Elz]

Date: Mon, 20 Oct 2014 15:32:47 -0500
From: Dennis Ferguson <***@gmail.com>
Message-ID: <E8CFDAF4-5E62-4EAD-9C99-***@gmail.com>

| 127.0.0.1 was just arbitrarily chosen for
| the 4.2 BSD lo(4) interface, there was no standard for that.

Dennis, do you know/remember whether Berkeley actually invented that
use, or did they "borrow" it from elsewhere? I never knew (and don't
think I ever bothered to ask), it just seemed to be there from the very
first time I saw the networking code.

kre


--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Robert Elz]

Date: Mon, 20 Oct 2014 14:15:14 -0400 (EDT)
From: Mouse <***@Rodents-Montreal.ORG>
Message-ID: <***@Stone.Rodents-Montreal.ORG>

| I just now had occasion to look at lo(4), for the first time in I don't
| know how long.

Your message inspired me to as well, after all, why would any bother to
look there, we all know ....

I know the specific issue your raised has been answered, but did you
also ready the rest of the man page ...

The
loopback interface should be the last interface configured, as protocols
may use the order of configuration as an indication of priority. The
loopback should never be configured first unless no hardware interfaces
exist.

The whole thing (that, and more I did not quote) kind of implies that
people would probably be better off not configuring lo0 at all, unless they
have some specific need for it.

These days of course, all that has changed - lo0 is always (on NetBSD anyway)
configured first, the network stack simply won't work unless it is there, and
"operating" correctly - it is fundamental, not just a quirky add on.

The point of this message, is that if anyone is inspired to go remove the
BUGS section from lo(4) - don't just do that, rewrite the whole thing.

kre


--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Mindaugas Rasiukevicius]

Post by Dennis Ferguson

Post by Mouse
BUGS
Previous versions of the system enabled the loopback interface
automati- cally, using a nonstandard Internet address (127.1). Use of
that address is now discouraged; a reserved host address for the local
network should be used instead.
This is the first I've heard of 127.1, a very old syntax for 127.0.0.1,
being nonstandard for loopback use or of its being deprecated, and,
indeed, the startup scripts for each of those three versions are
hardwired to bring lo0 up as 127.0.0.1.

<...>
The first assigned numbers RFC where network 127 was assigned to
http://tools.ietf.org/html/rfc990
<...>

Was it a warning before the standardisation attempt? According to
CSRG archive, Mike Karels adjusted the manpage section in May 1986.

http://nxr.netbsd.org/history/src-csrg/share/man/man4/lo.4

[Dennis Ferguson]

Post by Robert Elz
Date: Mon, 20 Oct 2014 15:32:47 -0500
| 127.0.0.1 was just arbitrarily chosen for
| the 4.2 BSD lo(4) interface, there was no standard for that.
Dennis, do you know/remember whether Berkeley actually invented that
use, or did they "borrow" it from elsewhere? I never knew (and don't
think I ever bothered to ask), it just seemed to be there from the very
first time I saw the networking code.

I'm not sure. The only other networking I saw that early was the BBN
stack that was used with 4.1 BSD sometimes, and I wasn't familiar with
it since the kernel sources weren't on the machine I had access to,
but I recall that machine having a loopback interface that was left
unconfigured. I think that wasn't a possibility for 4.2 BSD, in fact
I think the 127.0.0.1 was hardwired so that if networking was configured
in the kernel it booted with lo0 already configured with the address. You
could change the address but I'm not sure you could delete it. There
was something about the 4.2 code that really, really wanted at least
one fully configured interface at all times, so I think the hardwired
loopback address might have been a hack-around for that (this also might
be why the machines were so prone to getting attached to that address
early and then sending packets sourced from it out the real network
interface later).

I hence wouldn't be surprised if Berkeley actually invented it, since
they were the ones with a need for a hardwirable address like that.
And the timing is right; you would never pick network 127 for this prior
to network classes, and classed networks officially arrived not too
long before the first Berkeley networking did.

Note that I only remember (or maybe misremember) any of this because
4.2 BSD was the first operating system I ever saw with a network stack,
and I was fascinated enough by it that I spent quite some time figuring
out how that code worked. In retrospect it probably wasn't the most
wonderful example of that art, but at the time I thought it was
absolutely excellent.

Dennis Ferguson

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de