One of the most innovative features to find its way into the Linux kernel recently is support for master mode on several wireless chipsets. Though there’s been little fanfare surrounding this development, it could soon be giving home users, in particular, another reason to celebrate Ubuntu.
Until a few years ago, Linux’s poor wireless support was one of the free-software community’s greatest embarrassments. Many devices lacked native support, meaning that the fickle process of loading Windows drivers through ndiswrapper was often the only way to get cards working–and even then, they were limited to managed and ad-hoc modes.
This situation has been completely reversed recently, with the Linux kernel now boasting at least basic support for most wireless cards, thanks to the Linux wireless project. Although Ubuntu’s rather conservative approach to the new wireless stack leaves many users still having to install drivers manually in Intrepid, the era is at least on the horizon when wireless networking will work out-of-the-box for everyone.
Master mode
Even better than a wireless card that “just works” in Ubuntu when connecting to a router in managed mode is one that can also replace the router, by operating in master mode. As of April 2009, the latest drivers from the Linux wireless project support this functionality on the following chipsets:
- Atheros (newer chips only)
- Broadcom
- Intel (2200 chips only)
- Prism
- Ralink (except the newest devices)
Granted, only experienced Linux geeks are likely to be able to enjoy master mode for the time being, since putting it into action requires intimate familiarity with the ‘iwconfig’ command, the ‘hostap’ utility and most likely dhcp servers, not to mention bridging network interfaces together.
Nonetheless, the day has at least become conceivable when setting up an Ubuntu computer as a wireless router is as easy to clicking a button in NetworkManager, allowing even the most technologically inept Linux users to create local wireless networks effortlessly–and not just ad-hoc networks like those supported by ‘Internet Connection Sharing’ in Windows, but true access point-based networks.
Why it matters
Beyond being cool, the ability to turn an Ubuntu computer easily into a wireless router has practical benefits. For starters, it saves consumers money, since they can avoid purchasing access points from commercial vendors.
A home computer turned into a router also offers much greater flexibility than most moderately priced commercial products. Rather than being stuck with the limited number of settings made available by hardware vendors, users can route traffic, customize firewalls, log activity and so on in whichever way they see fit.
As driver developers continue to enhance the capabilities of wireless on Linux and surpass the offerings of proprietary operating systems, Ubuntu and other distributions should concentrate on making those features easily accessible to end users, and on promoting them as one more thing that Linux can do and closed-source software can’t.
Don't miss Charlene O'Hanlon's weekly columns...
There is a Google Summer of Code Project associated with equipping networkmanager to make use of master mode
http://linuxwireless.org/en/developers/GSoC/2009/Add_AP_Support_to_Network_Manager
Notice who the mentor is? Isn’t it great that Red Hat does so much development work at all levels of the software stack for ubuntu users to take advantage of. From low level kernel work..like maintaining wireless drivers..as well as desktop components like NetworkManager, they are doing the integrated development to make it possible for things to just work. It’s almost like Red Hat is building the best and most interesting bits for future of Ubuntu. Imagine how much faster Ubuntu users could get their hands on important technologies like this if Canonical made the manpower investments to work on the kernel drivers or the end user wireless tools.
It’s really too bad Canonical decided they didn’t have time to mentor Google Summer of Code students this year after they were accepted as a mentoring organization. Google Summer of Code projects can produce some very useful things when there are mentors who are deeply involved in project development who can guide students. Then again, they’d need to have developers who were deeply involved as project leads in upstream projects to be good mentors..so in the end maybe its for the best they stepped back, even though they were already accepted to participate.
-jef
Jef: thanks for pointing out that link. I wasn’t aware of it, but it highlights the ease with which master-mode support can be harnessed to the benefit of general end users.
As for Red Hat, although I’ve been critical on this site of its professed disengagement from the desktop-Linux market, I do agree that it makes plenty of useful contributions to the free-software community. I think Canonical also does more than you give it credit for, but there’s always room for improvement.
“Imagine how much faster Ubuntu users could get their hands on important technologies like this if Canonical made the manpower investments to work on the kernel drivers or the end user wireless tools.”
Well, of course. Canonical has somewhere in the region of 5% the revenue of RedHat and it isn’t profitable yet. What do you expect them to do? Invest money that they don’t have or has to come out of Shuttleworth’s pocket? They can only do so much of this.
But your post points to something interesting, i.e. the superiority of open source in bringing benefit to all its users.
“Until a few years ago, Linux’s poor wireless support was one of the free-software community’s greatest embarrassments.”
It still is, have you looked at the Ubuntuforums message boards at all recently. My current laptop still won’t do wireless or audio properly in 8.10 or 9.04 after 60-80 hours spent on the problems. This article made me laugh. It illustrates Linux/Ubuntu’s biggest issue, tunnel visioned developers spending a lot of man hours adding obscure unnecessary capabilities when the common required ones still don’t work properly or at all. Why would I replace my $60.00 wireless router that works just fine with a $600 laptop that still doesn’t? Absurd!
Although Linux-box-as-a-router is not really a new concept, I think dedicating a full fledged operating system to work as a Gateway could be an overkill.
Ubuntu is designed as a full-fledged operating system and it’s full benefits can be realised if used for an array of tasks as opposed to only one.
Having said this, embedded linux is undoubtedly the best choice for a gateway software and Ubuntu Mobile Edition can something to be looked at for porting on small devices.
Chris,
I think I have a very accurate picture as to what Canonical pays employees do. The majority of Canonical developers are paid to write private web services application code behind closed doors. But if you disagree with that feel free to prove me wrong. See if you can get people talking about UbuntuOne project. The UbuntuOne hacker team is larger than the Desktop Experience team according to launchpad…both internal Canonical teams to my understanding…hmmm.
Please start writing articles about the open technical advancements that Canonical is driving as the technical lead for Ubuntu. Jockey, checkbox, bzr, upstart.. all reasonably good fodder for articles as Canonical employees are part of the core development team for those projects. Upstart is a very interesting story because if you look closely you’ll see that a significant amount of work to integrate it with D-Bus was done by a Red Hat employee in the 0.5 upstart timeframe. Upstart is a good coop-etition story. Too bad its not sexy enough to get the attention from the blogosphere. Ubuntu is still using it in backwards compatible SysVinit mode so even in Jaunty its not actually taking advantage of the event based approach:
https://bugs.launchpad.net/ubuntu/+source/upstart/+bug/338648
Isn’t that worth talking about? Multiple releases of Ubuntu later and Upstart features arent being used beyond basic SysVinit compatibility.
Or if you want something a little more sensational, go ahead and write about the future of notify-osd and libindicate which speaks to whether Canonical has done enough to discuss their place as a potential part of the Gnome Desktop of the future.
http://bugzilla.gnome.org/show_bug.cgi?id=557647#c16
Canonical does not have the manpower to build and maintain a highly customized GNOME desktop for any significant length of time. That’s pretty clear from looking at how they handle existing GNOME desktop bug reports.
https://bugs.launchpad.net/ubuntu/+source/gnome-session/+bug/249373
The more customization patching Canonical does downstream of GNOME the heavier the burden is on Canonical to actually fix things on their own. The fact that they are experimenting with a change in how notifications work inside an Ubuntu release which deeply impacts functionality across a wide number of applications without being on the roadmap for upstream GNOME should be a concern for you as an Ubuntu supporter. Even if the feature is great, and its wildly popular, does Canonical have the manpower to keep the new notification concept maintained as a set of downstream patches against applications?
Or on the other hand maybe the notification work Canonical is investing developer time in really isn’t really compelling to you. If the most compelling and interesting things on the horizon for linux users that you can think to write about are technical advancements that Canonical isn’t playing a significant role in you need to do a much better job of point out who is actually leading that effort.
I also find it ironic that Shuttleworth says that kernel work isn’t that important to Ubuntu users.. like his mom (see the recent video interview). And yet here you are talking about kernel level driver features as one of the most important things coming up for Ubuntu. That is a significant mismatch in world views. Do you consider yourself part of the target audience for Ubuntu? I’m not sure Shuttleworth does.
Articles like the one you have written above put the emphasis on the wrong part of the equation. The hard work..the work that need to be lifted up..is not the consumption of the functionality by integrators like Canonical. The hard work is the actual development of the functionality. If you are going to write articles about feature development then give credit where credit is due. Make the effort to actually talk to a developer..any developer..who is involved in doing the work and give them a chance to be quoted in the article with their perspective. Fair’s fair.
-jef