Date: Fri, 13 Jun 2003 12:20:09 -0700 (PDT) From: Bruce M Simpson <bms@spc.org> To: freebsd-bugs@FreeBSD.org Subject: Re: kern/53187 Message-ID: <200306131920.h5DJK9ji095175@freefall.freebsd.org>
next in thread | raw e-mail | index | archive | help
The following reply was made to PR kern/53187; it has been noted by GNATS.
From: Bruce M Simpson <bms@spc.org>
To: freebsd-gnats-submit@FreeBSD.org
Cc:
Subject: Re: kern/53187
Date: Wed, 11 Jun 2003 09:24:29 +0100
Here are more of my notes from working through the problem:-
- firmly established that old-style monitor-mode/procframe handling is
gone from the driver. this is what breaks bsd-airtools.
- tcpdump is able to sniff 802.11 headers though; but it does the
following: pcap_set_datalink(p, DLT_IEEE802_11).
- looks like enabling monitor mode does actually go through to hw still.
- dstumbler is heavily dependent on the old monitor-mode behaviour
from 5.0-RELEASE. but it could be patched to do things in a way
which is compatible with the new api changes.
- looks like struct wi_rx_frame {} no longer used, but still defined.
- the wi_frame is broadly equivalent to the wi_rx_frame struct
abused by bsd-airtools.
- one would need to define a different link layer, IEEE802_11_PRISM2,
dummy mbuf with procfame, and bpf_mtap() that.
- Conditionalise it for people who are worried about this support
bloating up their packet input path.
- Moving wi_rx_intr() processing into taskqs is probably a far distant
future exercise right now anyway.
I could potentially write the code to do the above. I would like review
from sam@ and imp@ first before deciding how to proceed.
BMS
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200306131920.h5DJK9ji095175>