Rask Ingemann Lambertsen wrote:
On Mon, Nov 24, 2003 at 03:24:11PM -0800, Ben Greear wrote:
Thanks to those who pointed me in the right direction, here is a patch
to the e100 (2.4.23-pre9) that allows it to capture all frames, bogons included.
It also coppies the FCS to the skb so ethereal et al can read it.
It utilizes ethtool commands to get/set the rx-all feature, and uses
a new flag in the skbuff (and socket struct) structure to determine when to
disable generating
the FCS on transmit. I have the entire patch that adds the management
bits and flags, but as usual, it's mixed in with various other things...
Reading the tulip manual (see below) triggered a question: When transmitting
a custom CRC, who is responsible for padding the frame to the minimum length?
If frame padding is left to the driver, what should be used for padding?
Hrm, I have not tested this, as my sending app itself enforces the minimum
size. I am guessing that padding is up to the driver/OS in this case,
but I'd have to re-read the e100 docs to be sure...
I (or rather, Google) found the tulip documentation at
<URL:http://www.intel.com/design/network/manuals/278074.htm>. The tulip
chips are capable of all these tricks too.
Cool, I'll check out this doc soon. If you happen to write up a tulip driver
patch for this feature, please let me know. I have lots of 4-port tulip nics to
test on here...
DaveM: Are you interested in getting these patches into 2.4.24-preX? (It would
be helpful to get the infrastructure and ethtool portions in, even if the driver
parts remain outside the tree for a bit longer.)
Thanks,
Ben
--
Ben Greear <greearb@xxxxxxxxxxxxxxx>
Candela Technologies Inc http://www.candelatech.com
|