Debian Bug report logs - #101728
Automatic detection of hardware interface

Reply or subscribe to this bug.

Toggle useless messages

Message #5 received at submit@bugs.debian.org (full text, mbox, reply):

From: Anthony Towns <aj@azure.humbug.org.au>

To: submit@bugs.debian.org

Cc: Tom Rini <trini@kernel.crashing.org>

Subject: Automatic detection of hardware interface

Date: Thu, 21 Jun 2001 14:55:15 +1000

Package: ifupdown
Severity: wishlist

(Tom, FYI only)

ifupdown should support automatically determining the physical interface
given the logical interface (or some hint anyway), as well as vice-versa.

Using the message below as an example, you should be able to say something
like:

	ifup =airport

and have a script detect that the "airport" card is actually "eth1";
and then continue normal processing, so you eventually end up bring
"eth1" up in the "home" configuration (eg).

That generalises to servers where eth0/eth1 could get mixed up (eg), and
might make things easier for proper ppp support later. Maybe.

----- Forwarded message from Tom Rini <trini@kernel.crashing.org> -----
Date: Wed, 20 Jun 2001 13:02:48 -0700
From: Tom Rini <trini@kernel.crashing.org>
To: debian-powerpc@lists.debian.org
Subject: Slightly newer airport sleep bits for pmud
Message-ID: <20010620130248.A26006@opus.bloom.county>

Hey all, after talking to someone else, I found out not everyone has their
airport modules loaded with autoclean, So I've appended my current version
of the script I use which should work around this.  And I moved it into
/etc/power/pwrctl-local since no official kernels support sleep yet on these
machines.  Here it is:

logger=/usr/bin/logger

case "$1" in
sleep)
        # Currently the airport driver isn't happy with sleep, so we
        # must remove it
        if [ "`lsmod | grep airport 2>/dev/null`" ]; then
                AIRPORT=`cat /proc/net/wireless | grep eth | sed 's/^  //' | cut -d : -f 1`
                ifdown $AIRPORT
                tempfile -n /var/run/airport || {
                        $logger -p daemon.error -t pwrctl "$0: cannot create tempfile"
                        exit 1
                }
                echo $AIRPORT > /var/run/airport
                # Make sure we get airport, even if it's not autoprobed.
                rmmod airport && rmmod orinoco && rmmod hermes
        fi
;;
wakeup)
        [ -f /var/run/airport ] && {
                modprobe -k airport
                ifup `cat /var/run/airport`
                rm /var/run/airport
        }
;;
*)
;;
esac
----- End forwarded message -----

Message #10 received at 101728@bugs.debian.org (full text, mbox, reply):

From: Thomas Hood <jdthood@yahoo.co.uk>

To: 101728@bugs.debian.org

Cc: Tom Rini <trini@kernel.crashing.org>

Subject: Flawed idea?

Date: 01 Feb 2003 09:54:08 +0100

If my understanding of how ifupdown works is correct
then this (#101728) idea is flawed.  The proposal is
that ifup|down be enhanced so that logical interfaces
can be named on the command line instead of physical
interfaces.  However, if it is the case that more than
one physical interface can be mapped to a single logical
interface then such a command can be ambiguous.  Unless
of course we want to introduce "reverse-mapping" 
declarations.  What is the motivation for this proposal?

Cheers
-- 
Thomas Hood <jdthood@yahoo.co.uk>

Message #15 received at 101728@bugs.debian.org (full text, mbox, reply):

From: Tom Rini <trini@kernel.crashing.org>

To: Thomas Hood <jdthood@yahoo.co.uk>, Anthony Towns <aj@azure.humbug.org.au>

Cc: 101728@bugs.debian.org

Subject: Re: Flawed idea?

Date: Sat, 1 Feb 2003 10:50:46 -0700

On Sat, Feb 01, 2003 at 09:54:08AM +0100, Thomas Hood wrote:

> If my understanding of how ifupdown works is correct
> then this (#101728) idea is flawed.  The proposal is
> that ifup|down be enhanced so that logical interfaces
> can be named on the command line instead of physical
> interfaces.  However, if it is the case that more than
> one physical interface can be mapped to a single logical
> interface then such a command can be ambiguous.  Unless
> of course we want to introduce "reverse-mapping" 
> declarations.  What is the motivation for this proposal?

To be honest, it's been a long time since I've had to think about this.
But, the problem was that at one point if you put an Apple TiBook to
sleep, with the 'airport' card active, upon wakeup the driver would be
horribly unhappy (requiring a reboot).  I don't _think_ this is a
problem anymore.  Anthony, do you have any desire still to see the
feature in the bug exist?  I wouldn't object to this bug being closed.

-- 
Tom Rini (TR1265)
http://gate.crashing.org/~trini/

Message #20 received at 101728@bugs.debian.org (full text, mbox, reply):

From: Anthony Towns <aj@azure.humbug.org.au>

To: Thomas Hood <jdthood@yahoo.co.uk>, 101728@bugs.debian.org

Cc: Tom Rini <trini@kernel.crashing.org>

Subject: Re: Bug#101728: Flawed idea?

Date: Sun, 2 Feb 2003 17:15:31 +1000

On Sat, Feb 01, 2003 at 09:54:08AM +0100, Thomas Hood wrote:
> If my understanding of how ifupdown works is correct
> then this (#101728) idea is flawed.  

I'm not sure why you think you've got a better understanding of how
ifupdown works than I do, Thomas. Your assistance is appreciated, but
please try to realise you ought to just be offering what assistance you
can, not making decisions you don't have the experience to make.

The original report is still entirely correct.

Cheers,
aj

-- 
Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/>
I don't speak for anyone save myself. GPG signed mail preferred.

    ``Australian Linux Lovefest Heads West''
                   -- linux.conf.au, Perth W.A., 22nd-25th January 2003

Message #25 received at 101728@bugs.debian.org (full text, mbox, reply):

From: Thomas Hood <jdthood@yahoo.co.uk>

To: 101728@bugs.debian.org

Cc: Tom Rini <trini@kernel.crashing.org>

Subject: Re: Bug#101728: Flawed idea?

Date: 02 Feb 2003 11:59:09 +0100

On Sun, 2003-02-02 at 08:15, Anthony Towns wrote:
> On Sat, Feb 01, 2003 at 09:54:08AM +0100, Thomas Hood wrote:
> > If my understanding of how ifupdown works is correct
> > then this (#101728) idea is flawed.  
> 
> I'm not sure why you think you've got a better understanding of how
> ifupdown works than I do, Thomas. Your assistance is appreciated, but
> please try to realise you ought to just be offering what assistance you
> can, not making decisions you don't have the experience to make.
> 
> The original report is still entirely correct.

I never said that I thought I had a better understanding than
you have.  I just said, roughly, "IIUC this is a flawed idea"
and explained my reasoning.  I took no decision; I did not
change the status of the report in any way, apart from
appending my comment.

Your reaction appears to be to take offence, to criticize me
unfairly, and to insist that you were right without offering
any explanation of where my reasoning went wrong.  I don't
think that that is conducive to a constructive conversation.
But I'll drop the matter.

-- 
Thomas Hood <jdthood@yahoo.co.uk>

Message #34 received at 101728@bugs.debian.org (full text, mbox, reply):

From: Craig Shelley <craig@microtron.org.uk>

To: Debian Bug Tracking System <101728@bugs.debian.org>

Subject: ifupdown: Using logical mappings only

Date: Sun, 05 Sep 2004 10:57:24 +0100

Package: ifupdown
Version: 0.6.4-4.10
Followup-For: Bug #101728

I think a better way would be to allow complete abstraction from the
ethx names without requiring an external mapping script.
This would involve a section within the interfaces file directly mapping the interface names to mac addresses.

These can then be assigned to the physical devices for example:
ip link set eth1 name broadband

so ifconfig would then report:
broadband Link encap:Ethernet  HWaddr 00:50:8D:49:88:E6
          inet addr:192.168.0.65  Bcast:192.168.0.255  Mask:255.255.255.192
          UP BROADCAST MULTICAST  MTU:1500  Metric:1
....



The overall effect is that the end user never gets to see the ethx
names. Instead they are presented with the names they assigned in the
config file. It would also remove the need for real names and
logical names.

By using a command such as:
ifdown broadband
Since the physical interface name is given on the command line (as they
are currently), I would guess that it shouldn't be too hard to implement.

There may be a flaw in this idea, please let me know what you think.

Craig Shelley

-- System Information:
Debian Release: 3.1
  APT prefers unstable
  APT policy: (500, 'unstable'), (1, 'experimental')
Architecture: i386 (i686)
Kernel: Linux 2.6.8.1
Locale: LANG=en_GB.UTF-8, LC_CTYPE=en_GB.UTF-8

Versions of packages ifupdown depends on:
ii  debconf [debconf-2.0]       1.4.32       Debian configuration management sy
ii  libc6                       2.3.2.ds1-16 GNU C Library: Shared libraries an
ii  net-tools                   1.60-10      The NET-3 networking toolkit

-- debconf information:
  ifupdown/convert-interfaces: true

Message #39 received at 101728@bugs.debian.org (full text, mbox, reply):

From: Marc Haber <mh+debian-packages@zugschlus.de>

To: Craig Shelley <craig@microtron.org.uk>, 101728@bugs.debian.org

Subject: Re: Bug#101728: ifupdown: Using logical mappings only

Date: Sun, 5 Sep 2004 12:32:23 +0200

On Sun, Sep 05, 2004 at 10:57:24AM +0100, Craig Shelley wrote:
> I think a better way would be to allow complete abstraction from the
> ethx names without requiring an external mapping script.
> This would involve a section within the interfaces file directly mapping the interface names to mac addresses.
> 
> These can then be assigned to the physical devices for example:
> ip link set eth1 name broadband
> 
> so ifconfig would then report:
> broadband Link encap:Ethernet  HWaddr 00:50:8D:49:88:E6
>           inet addr:192.168.0.65  Bcast:192.168.0.255  Mask:255.255.255.192
>           UP BROADCAST MULTICAST  MTU:1500  Metric:1
> ....

Something like this?

1: lo: <LOOPBACK,UP> mtu 16436 qdisc noqueue 
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 brd 127.255.255.255 scope host lo
2: unt0: <BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast qlen 1000
    link/ether 00:00:e8:df:07:78 brd ff:ff:ff:ff:ff:ff
    inet 212.126.222.70/29 brd 212.126.222.71 scope global unt0:de46d
    inet 192.168.131.1/29 brd 192.168.131.7 scope global unt0:83018
3: int0: <BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast qlen 1000
    link/ether 00:00:e8:6c:d0:9b brd ff:ff:ff:ff:ff:ff
    inet 192.168.130.1/24 brd 192.168.130.255 scope global int0:82018
4: per0: <BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast qlen 1000
    link/ether 00:00:e8:6b:3d:9c brd ff:ff:ff:ff:ff:ff
    inet 212.126.222.73/29 brd 212.126.222.79 scope global per0:de49d

Unt0 is the untrusted interface, int0 the internal one, and per0 a
perimeter network.

If that's what you seek, take a look at the ifupdown-scripts-zg2
package in unstable which uses ifupdown's manual method to implement
this. It is, however, quite far from ifupdown's original design goals.

Greetings
Marc

-- 
-----------------------------------------------------------------------------
Marc Haber         | "I don't trust Computers. They | Mailadresse im Header
Karlsruhe, Germany |  lose things."    Winona Ryder | Fon: *49 721 966 32 15
Nordisch by Nature |  How to make an American Quilt | Fax: *49 721 966 31 29

Message #44 received at 101728@bugs.debian.org (full text, mbox, reply):

From: Thomas Hood <jdthood@yahoo.co.uk>

To: Craig Shelley <craig@microtron.org.uk>, 101728@bugs.debian.org

Cc: debian-user@lists.debian.org

Subject: Re: Bug#101728: ifupdown: Using logical mappings only

Date: Sun, 05 Sep 2004 14:02:55 +0200

First some terminology.  There are real network adapters and
there are, assigned to the latter:

* MAC addresses
* "physical" interface names (assigned by the kernel)
* "logical" interface names (i.e., names of ifupdown profiles)

AIUI you want to be able to define logical interfaces in
/etc/network/interfaces as this is currently done but you want
a simpler way of getting these assigned to real interfaces on
the basis of their MAC addresses and you want the "physical"
interface names to be changed so that they are the same as the
logical interface names.

The behavior you want is already obtainable by installing the
ifrename package.  The ifrename command changes the physical
interface name to another name on the basis of the real
interface's MAC address.  Once interfaces have been ifrenamed,
all you have to do is define logical interfaces in /e/n/i 
using the same names.

E.g. Suppose you have an interface adapter card with MAC address
11:22:33:44:55:66.  You set up /etc/iftab with this line:

    george     mac 11:22:33:44:55:66

and /etc/network/interfaces with:

    iface george inet static
        address 11.22.33.44
        netmask ...

Suppose the interface has been named 'eth0' by the kernel.  Then you
do:

    ifrename -i eth0

to rename the interface and then do:

    ifup george

to bring it up.

If the interface is hot plugged and you have the hotplug package
installed then the ifrename command gets executed automatically.
An ifup command also gets executed automatically, but in this form:

    ifup george=hotplug

In order to allow this to to have the effect of bring up interface
"george" as logical interface "george" you need to add the following
stanza to /etc/network/interfaces:

    mapping hotplug
        script echo

--
Thomas Hood

Message #49 received at 101728@bugs.debian.org (full text, mbox, reply):

From: Bernd Eckenfels <be-mail2005@lina.inka.de>

To: control@bugs.debian.org, 101728@bugs.debian.org, 108974@bugs.debian.org

Subject: wontfix in nettools

Date: Mon, 24 Oct 2005 21:22:34 +0200

tags 108974 + wontfix
thanks
this can be fixed in ifup, see Bug #101728

Message #54 received at 101728@bugs.debian.org (full text, mbox, reply):

From: Jacques Rodary <jrodary@free.fr>

To: 101728@bugs.debian.org

Subject: confirmation de souscription

Date: Thu, 14 Apr 2016 01:49:58 +0200

[Message part 1 (text/plain, inline)]

	cordialement
		Jacques Rodary
		jrodary@free.fr

[Message part 2 (text/html, inline)]

Send a report that this bug log contains spam.