MTU discovery

MTU discovery

[Robert PipCA]

  Hi,
  I'm working on a project that require knowing the
max MTU size supported by the 3Com PCI 3c905C
(Boomerang).
  The datasheet provided by 3Com does not mention it,
and I already did the usual google search, but didn't
find it neither.
  Does anyone knows a "generic way" of knowing this
(or chip-specific)?
  Thanks in advance.
--Robert

__________________________________________________
Do You Yahoo!?
Yahoo! - Official partner of 2002 FIFA World Cup
http://fifaworldcup.yahoo.com

Re: MTU discovery

[Matti Aarnio]

On Mon, Jun 10, 2002 at 12:45:07AM -0700, Robert PipCA wrote:
>   Hi,
>   I'm working on a project that require knowing the max MTU size
> supported by the 3Com PCI 3c905C (Boomerang).
> The datasheet provided by 3Com does not mention it, and I already
> did the usual google search, but didn't find it neither.
> Does anyone knows a "generic way" of knowing this (or chip-specific)?

  Oh, it is mentioned there, although not with that name.

  The Ethernet Standard (IEEE 802.3) specifies that the frame size
  shall be 1500 octets.   That is the NORMAL CASE max MTU value for
  all ethernet devices.

  Some devices do, however, support reception (and transmit) of what
  is called "jumbograms".  With boomerang you can set a register
  to contain the limit value.  Alternatively with boomerang, and
  its predecessors, you can set a bit to accept extra-large frames.

  I recall the ultimate limit is in order of 4kB.

>   Thanks in advance.
> --Robert

/Matti Aarnio

Re: MTU discovery

[Robert PipCA]

  Hi,
--- Matti Aarnio <matti.aarnio@zmailer.org> wrote:
>   Oh, it is mentioned there, although not with that
> name.
>   Some devices do, however, support reception (and
> transmit) of what
>   is called "jumbograms".  With boomerang you can
> set a register
>   to contain the limit value.  Alternatively with
> boomerang, and
>   its predecessors, you can set a bit to accept
> extra-large frames.
>   I recall the ultimate limit is in order of 4kB.

   How did you came up with "4Kb"?
   I've looked through the datasheets of the board and
didn't find any mention to it.
--Robert

__________________________________________________
Do You Yahoo!?
Yahoo! - Official partner of 2002 FIFA World Cup
http://fifaworldcup.yahoo.com

Re: MTU discovery

[Olivier Galibert]

On Mon, Jun 10, 2002 at 11:05:13AM +0300, Matti Aarnio wrote:
>   Some devices do, however, support reception (and transmit) of what
>   is called "jumbograms".  With boomerang you can set a register
>   to contain the limit value.  Alternatively with boomerang, and
>   its predecessors, you can set a bit to accept extra-large frames.
> 
>   I recall the ultimate limit is in order of 4kB.

Actually, in my experience jumbograms are usually 9000 bytes.

  OG.

Re: MTU discovery

[Bill Davidsen]

On Mon, 10 Jun 2002, Olivier Galibert wrote:

> On Mon, Jun 10, 2002 at 11:05:13AM +0300, Matti Aarnio wrote:
> >   Some devices do, however, support reception (and transmit) of what
> >   is called "jumbograms".  With boomerang you can set a register
> >   to contain the limit value.  Alternatively with boomerang, and
> >   its predecessors, you can set a bit to accept extra-large frames.
> > 
> >   I recall the ultimate limit is in order of 4kB.
> 
> Actually, in my experience jumbograms are usually 9000 bytes.

To assist in searching for info, I've also seen the terms "jumbo packets"
and "jumbo frames." 

-- 
bill davidsen <davidsen@tmr.com>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.

Re: MTU discovery

[Joel Jaeggli]

9000 is the limit on gigabit ethernet, other media type have different 
maximum frame sizes (ie 4470 on fddi, 9216 on pos oc12 interfaces). 

On Mon, 10 Jun 2002, Bill Davidsen wrote:

> On Mon, 10 Jun 2002, Olivier Galibert wrote:
> 
> > On Mon, Jun 10, 2002 at 11:05:13AM +0300, Matti Aarnio wrote:
> > >   Some devices do, however, support reception (and transmit) of what
> > >   is called "jumbograms".  With boomerang you can set a register
> > >   to contain the limit value.  Alternatively with boomerang, and
> > >   its predecessors, you can set a bit to accept extra-large frames.
> > > 
> > >   I recall the ultimate limit is in order of 4kB.
> > 
> > Actually, in my experience jumbograms are usually 9000 bytes.
> 
> To assist in searching for info, I've also seen the terms "jumbo packets"
> and "jumbo frames." 
> 
> 

-- 
-------------------------------------------------------------------------- 
Joel Jaeggli	      Academic User Services   joelja@darkwing.uoregon.edu    
--    PGP Key Fingerprint: 1DE9 8FCA 51FB 4195 B42A 9C32 A30D 121E      --
  In Dr. Johnson's famous dictionary patriotism is defined as the last
  resort of the scoundrel.  With all due respect to an enlightened but
  inferior lexicographer I beg to submit that it is the first.
	   	            -- Ambrose Bierce, "The Devil's Dictionary"

RE: MTU discovery

[BALBIR SINGH]

[-- Attachment #1: Type: text/plain, Size: 672 bytes --]

Actually not, the Intel driver/card supports a size of

look into e1000_mac.h

#define MAX_JUMBO_FRAME_SIZE         0x3F00 (which is  16128).

The bottom line is 9000 is not the limit, it is
just the beginning of jumbo frame sizes.

Balbir

|-----Original Message-----
|From: linux-kernel-owner@vger.kernel.org 
|[mailto:linux-kernel-owner@vger.kernel.org] On Behalf Of Joel Jaeggli
|Sent: Tuesday, June 11, 2002 3:32 AM
|To: Bill Davidsen
|Cc: Olivier Galibert; linux-kernel@vger.kernel.org
|Subject: Re: MTU discovery
|
|
|
|9000 is the limit on gigabit ethernet, other media type have different 
|maximum frame sizes (ie 4470 on fddi, 9216 on pos oc12 interfaces). 
|


[-- Attachment #2: Wipro_Disclaimer.txt --]
[-- Type: text/plain, Size: 490 bytes --]

**************************Disclaimer************************************

Information contained in this E-MAIL being proprietary to Wipro Limited is 
'privileged' and 'confidential' and intended for use only by the individual
 or entity to which it is addressed. You are notified that any use, copying 
or dissemination of the information contained in the E-MAIL in any manner 
whatsoever is strictly prohibited.

***************************************************************************

RE: MTU discovery

[Joel Jaeggli]

does ieee 802.3z specify a standard? I don;t have a copy handy.

joelja

On Tue, 11 Jun 2002, BALBIR SINGH wrote:

> Actually not, the Intel driver/card supports a size of
> 
> look into e1000_mac.h
> 
> #define MAX_JUMBO_FRAME_SIZE         0x3F00 (which is  16128).
> 
> The bottom line is 9000 is not the limit, it is
> just the beginning of jumbo frame sizes.
> 
> Balbir
> 
> |-----Original Message-----
> |From: linux-kernel-owner@vger.kernel.org 
> |[mailto:linux-kernel-owner@vger.kernel.org] On Behalf Of Joel Jaeggli
> |Sent: Tuesday, June 11, 2002 3:32 AM
> |To: Bill Davidsen
> |Cc: Olivier Galibert; linux-kernel@vger.kernel.org
> |Subject: Re: MTU discovery
> |
> |
> |
> |9000 is the limit on gigabit ethernet, other media type have different 
> |maximum frame sizes (ie 4470 on fddi, 9216 on pos oc12 interfaces). 
> |
> 
> 

-- 
-------------------------------------------------------------------------- 
Joel Jaeggli	      Academic User Services   joelja@darkwing.uoregon.edu    
--    PGP Key Fingerprint: 1DE9 8FCA 51FB 4195 B42A 9C32 A30D 121E      --
  In Dr. Johnson's famous dictionary patriotism is defined as the last
  resort of the scoundrel.  With all due respect to an enlightened but
  inferior lexicographer I beg to submit that it is the first.
	   	            -- Ambrose Bierce, "The Devil's Dictionary"

Re: [vortex] Re: MTU discovery

[Donald Becker]

On Mon, 10 Jun 2002, Matti Aarnio wrote:

> Date: Mon, 10 Jun 2002 11:05:13 +0300
> From: Matti Aarnio <matti.aarnio@zmailer.org>
> To: Robert PipCA <robertpipca@yahoo.com>
> Cc: vortex@scyld.com, linux-kernel@vger.kernel.org
> Subject: [vortex] Re: MTU discovery
>
> On Mon, Jun 10, 2002 at 12:45:07AM -0700, Robert PipCA wrote:
> >   Hi,
> >   I'm working on a project that require knowing the max MTU size
> > supported by the 3Com PCI 3c905C (Boomerang).
> > The datasheet provided by 3Com does not mention it, and I already
> > did the usual google search, but didn't find it neither.
> > Does anyone knows a "generic way" of knowing this (or chip-specific)?
..
>   Some devices do, however, support reception (and transmit) of what
>   is called "jumbograms".  With boomerang you can set a register
>   to contain the limit value.

A 16 bit register.. 64KB packets.  There are various issues with using
large packet sizes.  There is no driver that has been verified with
jumbo frames.  I have been throwing driver versions at Rishi Srivatsavai
<rishis at CLEMSON.EDU> trying to sort out the issues.  You might notice
the changes in 0.99W, although they don't handle the FIFO limit issues.

>  Alternatively with boomerang, and
>   its predecessors, you can set a bit to accept extra-large frames.
>
>   I recall the ultimate limit is in order of 4kB.

More precisely, FDDI frame size minus the FDDI-specific bits, about
4.5KB.

-- 
Donald Becker				becker@scyld.com
Scyld Computing Corporation		http://www.scyld.com
410 Severn Ave. Suite 210		Second Generation Beowulf Clusters
Annapolis MD 21403			410-990-9993

Re: [vortex] Re: MTU discovery

[Jeff Garzik]

Donald Becker wrote:
> A 16 bit register.. 64KB packets.  There are various issues with using
> large packet sizes.  There is no driver that has been verified with
> jumbo frames.  I have been throwing driver versions at Rishi Srivatsavai
> <rishis at CLEMSON.EDU> trying to sort out the issues.  You might notice
> the changes in 0.99W, although they don't handle the FIFO limit issues.


With the VLAN stuff that suddenly appeared, I did some large packet 
testing...  I would typically cap the MTU limit at just under the FIFO 
size, since it often requires special handling doing frames > FIFO size.

I do not look forward to re-verifying the 2.4 tulip driver across all 
those chipsets, to make sure my large packet code [which is not yet in 
the kernel] works. :)

	Jeff