number of eth0 device

number of eth0 device

[Karel Kulhavy]

Hello

I am looking into Documentation/devices.txt in 2.4.25 and eth0 is not listed
there. If I grep "eth", I get only

38 char        Myricom PCI Myrinet board
[...]
"This device is used for status query, board control and "user level
packet I/O."  This board is also accessible as a standard networking
"eth" device.  "

and then

/dev/pethr0

Is eth0 some kind of special device that doesn't have any number
assigned?

CL<

Re: number of eth0 device

[Erik Mouw]

On Wed, Oct 19, 2005 at 12:31:35PM +0200, Karel Kulhavy wrote:
> I am looking into Documentation/devices.txt in 2.4.25 and eth0 is not listed
> there. If I grep "eth", I get only
> 
> 38 char        Myricom PCI Myrinet board
> [...]
> "This device is used for status query, board control and "user level
> packet I/O."  This board is also accessible as a standard networking
> "eth" device.  "
> 
> and then
> 
> /dev/pethr0
> 
> Is eth0 some kind of special device that doesn't have any number
> assigned?

Yes, there's no such thing as /dev/eth0, network interfaces have their
own namespace. Linux uses the defacto standard BSD socket interface for
networking, so blame the BSD people for violating the "everything is a
file" rule.


Erik

-- 
+-- Erik Mouw -- www.harddisk-recovery.com -- +31 70 370 12 90 --
| Lab address: Delftechpark 26, 2628 XH, Delft, The Netherlands

Re: number of eth0 device

[Mathieu Segaud]

Erik Mouw <erik@harddisk-recovery.com> disait dernièrement que :

> On Wed, Oct 19, 2005 at 12:31:35PM +0200, Karel Kulhavy wrote:
>> I am looking into Documentation/devices.txt in 2.4.25 and eth0 is not listed
>> there. If I grep "eth", I get only
>> 
>> 38 char        Myricom PCI Myrinet board
>> [...]
>> "This device is used for status query, board control and "user level
>> packet I/O."  This board is also accessible as a standard networking
>> "eth" device.  "
>> 
>> and then
>> 
>> /dev/pethr0
>> 
>> Is eth0 some kind of special device that doesn't have any number
>> assigned?
>
> Yes, there's no such thing as /dev/eth0, network interfaces have their
> own namespace. Linux uses the defacto standard BSD socket interface for
> networking, so blame the BSD people for violating the "everything is a
> file" rule.

well, the way NIC's behave kind of forbids this
taken from Linux Device Drivers, 3rd Edition, page 497
"The normal file operations (read, write, and so on) do not make sense
when applied to network interfaces, so it is not possible to apply the
Unix ''everything is a file'' approach to them"   

-- 
Mathieu

Re: number of eth0 device

[Coywolf Qi Hunt]

On Wed, Oct 19, 2005 at 01:23:48PM +0200, Mathieu Segaud wrote:
> Erik Mouw <erik@harddisk-recovery.com> disait dernièrement que :
> 
> > On Wed, Oct 19, 2005 at 12:31:35PM +0200, Karel Kulhavy wrote:
> >> I am looking into Documentation/devices.txt in 2.4.25 and eth0 is not listed
> >> there. If I grep "eth", I get only
> >> 
> >> 38 char        Myricom PCI Myrinet board
> >> [...]
> >> "This device is used for status query, board control and "user level
> >> packet I/O."  This board is also accessible as a standard networking
> >> "eth" device.  "
> >> 
> >> and then
> >> 
> >> /dev/pethr0
> >> 
> >> Is eth0 some kind of special device that doesn't have any number
> >> assigned?
> >
> > Yes, there's no such thing as /dev/eth0, network interfaces have their
> > own namespace. Linux uses the defacto standard BSD socket interface for
> > networking, so blame the BSD people for violating the "everything is a
> > file" rule.
> 
> well, the way NIC's behave kind of forbids this
> taken from Linux Device Drivers, 3rd Edition, page 497
> "The normal file operations (read, write, and so on) do not make sense
> when applied to network interfaces, so it is not possible to apply the
> Unix ''everything is a file'' approach to them"   

I think there're other nodes in /dev on which normal file
operations do not make sense either.
-- 
Coywolf
-

Re: number of eth0 device

[Lee Revell]

On Wed, 2005-10-19 at 13:23 +0200, Mathieu Segaud wrote:
> well, the way NIC's behave kind of forbids this
> taken from Linux Device Drivers, 3rd Edition, page 497
> "The normal file operations (read, write, and so on) do not make sense
> when applied to network interfaces, so it is not possible to apply the
> Unix ''everything is a file'' approach to them"   
> 

Ditto sound cards, which is why ALSA abandoned the OSS "everything is a
file" paradigm for sound cards.  Read/write etc. don't take into account
the inherent realtime constraints in streaming audio to/from a device.

Lee