From: Jouni Malinen (jkmaline_at_cc.hut.fi)
Date: 2002-09-10 15:36:28 UTC
On Sun, Sep 08, 2002 at 03:26:10PM -0400, Gerald Britton wrote:
> Three patches here all relative to recent CVS snapshots of the driver.
> The first one (01-pci_suspend) is a possible fix for my pci suspend crash. It
> seems to do the right thing for suspend, but it's hard to tell if solves the
> crash as i'm still unable to reliably reproduce it, but this seems to be a
> better way of suspending/resuming than what we were doing before anyway.
I've applied this to CVS. I have never tested PCI suspend code, so I'm trusting on your tests for this.
> The second patch (02-promisc) enables IFF_PROMISC and IFF_ALLMULTI as the
> drivers in linux-wlan-ng do. Seems to mostly work. When promiscuous mode is
> enabled, a lot of FCS errors get through and packets received on other ports
> (mostly port 5 it seems). I've silenced those errors when in promisc mode.
I would like to do some testing with this before applying it and return to this after getting the next release of the driver out.
> The final patch (03-new_monitor) is a little bit of a rework of the monitoring
> code. It moves monitoring to wlan#ap if hostapd is enabled (this allows the
> interfaces to remain with the same arp type most of the time). Also, I've
> added support for simply flipping between the IEEE80211 and IEEE80211_PRISM arp
> types for that interface with an iwpriv command (arphrd_prism). I figured
> the information in the rx header might potentially be useful to hostapd in
> some situations. Finally, it moves enabling/disabling of the monitor mode to
> the IW_MODE_MONITOR mode listed in the latest wireless extensions.
Same for this patch; I'll take a closer look later.
I really do hate the format of IEEE80211_PRISM, so using it in more places is not exactly what I would like to see. Unfortunately, it has been included in libpcap and Ethereal so it might be a bit difficult to change this. I would prefer to keep the header stuff simple and remove all the reduntant info.
This arphrd_prism configuration would require some minimal changes for hostapd since it assumes that the interface is using IEEE80211 without Prism2 header data.
> I tried to leave compatibility where I could in the patches, the old monitor
> mode iwpriv command has been removed though. Monitor mode acts as it did if
> hostapd is not enabled in the config though (changing the arp type on wlan0).
iwpriv monitor could be used to provide a backward compatible way of changing mode, but it might be OK to remove it (and keep the driver cleaner). I haven't tested IW_MODE_MONITOR, but hopefully it can be used also with older kernels and wireless tools.
-- Jouni Malinen PGP id EFC895FA