Re: NKEs on Intel-based Macs
Re: NKEs on Intel-based Macs
- Subject: Re: NKEs on Intel-based Macs
- From: "Peter Sichel" <email@hidden>
- Date: Mon, 21 Aug 2006 17:39:51 -0400
Josh,
Thanks for taking the time to investigate.
On 8/21/06, Josh Graessley wrote:
>Unless there are bugs, the calls to mbuf_inbound_modified and
>mbuf_clear_csum_requested should not be required. In addition, if you
>don't want the stack to calculate the checksums, it should be possible
>to to just call mbuf_clear_csum_request.
In my testing, there were bugs that forced me to add the call to
mbuf_inbound_modified() in order to get it to work reliably. The bugs
were subtle relating to what happens if you bridge or forward a packet
arriving on one data-link back out the same data-link or another one
(who is supposed to reset these flags when an mbuf gets re-used and
possibly turned around because IP forwarding is enabled?). Testing the
simple case of sending and receiving TCP/IP over a single data-link
might not reveal the problem.
>mbuf_get_csum_requested: Intended for use by drivers. The stack will set
>flags indicating what checksums were postponed to be done in hardware.
>This is the function for the driver to pull this meta data out of the
>mbuf and convey that to the hardware as appropriate.
When I tested this (around 10.4.2), the flags returned correspond to the
Panther defined BSD constants, not what is documented in the KPI
reference or header files.
I realize this stuff is pretty obscure, but not fixing it creates bugs
that other developers end up stumbling over the hard way.
Kind Regards,
- Peter
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Macnetworkprog mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden