I’ve been meaning to clean up the net-wireless/orinoco* ebuilds for quite some time, now I’ve finally gotten around to it.
I’ve added an ebuild for the new 0.15rc3 snapshot (which still doesn’t contain USB support) and merged net-wireless/orinoco-cvs with net-wireless/orinoco. The new net-wireless/orinoco-9999 ebuild is a live CVS HEAD ebuild (which includes prism_usb support).
The net-wireless/orinoco-usb-firmware ebuild is now masked due to questionable licenses and will be removed soon. It is replaced by the official upstream package called net-wireless/orinoco-fwutils. This package contains various scripts for fetching the firmware (only needed by some cards – and some only in certain modes).
On a side note, the 0.15rc3 version of the orinoco driver was included in linux-2.6.15-rc1, hopefully prism_usb will follow soon.
We currently have two supplicants for IEEE 802.11/wired authentication available in portage, namely net-wireless/wpa_supplicant and net-misc/xsupplicant.
The latter is currently not up to date in portage nor integrated into our new baselayout, while wpa_supplicant is both up to date and well integrated in basesystem (thanks to Uberlord). We currently have a few open bugs for xsupplicant in our bugzilla, including a a version bump (bug #100680) and the design and development of a new rcscript for integrating it with our new baselayout (bug #98253). Also, wpa_supplicant seems to be much better supported from upstream than xsupplicant; more frequent releases and much more wide spread usage.
To justify having to maintain two packages (along with rcscripts) with the exact same purpose, I’d like to hear from xsupplicant users why they prefer xsupplicant over wpa_supplicant – and what prevents them from moving to using wpa_supplicant instead. Please post a comment, write me an email – or contact me on IRC.
Update: Seems the GWN authors have misunderstood this entry a bit.
Using the Atheros card sent to me by latexer, I was finally able to put some work into getting proper madwifi support in Gentoo Linux. I have created ebuilds for a new SVN snapshot (r1328-20051120) of both madwifi-driver and madwifi-tools and cleaned up the ebuilds somewhat.
Madwifi-tools now installs an udev rules file for automatically running wlanconfig when a new device is added. The prior preup()/postup() solution didn’t work properly with wpa_supplicant, which can take the interface down/up in the background.
Speaking of wpa_supplicant; I have also made a new revision of the wpa_supplicant-0.4.7 ebuild to reenable madwifi support. I have tested this against an access point using WPA-PSK TKIP, and it Works For Me ™. Now I just need to figure out why my Atheros card wont connect to an AP on channel 13 – `iwlist ath0 channel` shows channel 13 just fine…
Anyways, I’d really like some testers for the new madwifi-driver, madwifi-tools and wpa_supplicant ebuilds before I stick them into portage. They can be found in my portage overlay – both positive and negative feedback will be appreciated, just post a comment or contact me on IRC.
Yay! Pete was kind enough to send me two IEEE 802.11 CardBus cards all the way from the U.S.: an Atheros AR5212 based Proxim card and a PrismGT ISL3890 based Netgear card. These will be of great help when testing wpa_supplicant and hostapd, both packages which should have support for either of these cards.
Of course, I will also look into providing more up-to-date (and hopefully better, which shouldn’t be too hard taken the last couple of versions into consideration ;) ) ebuilds for both madwifi-driver and prism54 – and try to solve all the fuzz we recently had with madwifi support under Gentoo.
A big thank you to latexer for making this possible! :)
Update: also a big thanks to SuperLag for sponsoring the Prism54 card to latexer in the first place ;)