Re: mbuf from scratch
Re: mbuf from scratch
- Subject: Re: mbuf from scratch
- From: Josh Graessley <email@hidden>
- Date: Tue, 28 Jun 2005 15:07:36 -0700
On Jun 28, 2005, at 2:37 PM, Justin C. Walker wrote:
Thanks for enlarging on that idea.
It's been a while since I looked deeply at the output path, but
there are a few things to consider here. Perhaps Josh can comment.
I believe that, unless the destination is the loopback adapter
("lo0"), rcvif on outbound packets is NULL; it is mostly used on
the input path.
If it is used in a serious way on the outbound path, you should be
really careful about doing what you are doing. If there is an
"ifp" associated to the mbuf, and it is somehow pointing to the
outbound interface, a change in routing information that occurs
between the time you snag the original and the time that you inject
the munged mbuf might affect the proper value for that ifp. I
don't think this is the case, but you should try to verify it
(i.e., try to follow the ip_output code, as well as that for the
transport protocols :-}).
This is just a note of caution; as I said, I don't think it's an
issue. Josh?
As Justin noted, the rcvif is ignored for the outbound path, hence
the name receive if. This will usually be NULL on the outbound path.
If you go to inject a packet inbound, you will need to set the rcvif
to something as well as fiddle with the packet to get it in the right
state.
-josh
Attachment:
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-kernel mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden