Patrick Quinn's Blog

A babbling brook of information and thoughts.

header photo

Report: Linux hardware support.

Posted April 28, 2012

Linux has come so far since its initial release in 1991. In fact it beat all the odds to become  the first commercially viable open source platform. The fair of hackers, computer enthusiasts and Enterprise alike there is a lot of love across the board for the little Unix clone that could.
However, Linux, specifically Linux on the desktop, has had its share of problems that has held it back from mainstream dominance.  
Ignoring the competition offered by the likes of Microsoft and Apple , which of these problems is the most serious and how could said problem be resolved?

image depicting Linux hardware

I'm referring to Linux hardware compatibility. This is a tricky and rather touchy subject amongst the diehard Linux community and one with no simple solution. 

Many new users to Linux face a huge wall that they must climb when they first install a Linux distribution and thats that a lot of the hardware on their machine either doesn't work properly or doesn't work at all. While this problem is nowhere near as bad as it was, even as recently as 2006 when I started using Linux as my full time OS,  it's still a serious issue.

An example was found when Ubuntu was trending on Google+ earlier this week, as mentioned by me in my Ubuntu 12.04 sneak peak. A Google+ member had posted about his new install of Linux, looking for some help with problems he was having. While he loved the desktop and so on (In this it was Ubuntu’s Unity, told you its not a bad entry point for new users) around 25-30 percent of his main hardware components which are essential for normal usage where MIA. Sound, hardware accelerated graphics, wireless support etc flat out didn't work.

After an hour of running the gauntlet with the aid of around 20 or so Linux vets he managed to get various problems sorted, thankfully he enjoyed the challenge of getting it working but he was more tech inclined than your average user. Should it really be like that?

A mainstream Linux distribution should be a ‘point 'n' shoot’ affair when it comes to install and set up and not like a Rube Goldberg machine. On windows hardware is set up using an auto installation wizard which detects and installs the appropriate driver for any outstanding hardware components. While this might be an option for Linux due to licensing restrictions such a system would be forced into using open source drivers as the those from the actual hardware vendors are proprietary.

Mac OS X doesn't need to worry about this issue as it uses a strict subset of hardware across its product range with the exception of printers which use a similar wizard system to Windows. The Mac OS X approach underlines a root issue and highlights one possible, but (extraordinarily) extreme solution.

 

That issue wether or not generic, all encompassing, operating systems are really viable any more or were they ever really viable to begin with? Would it not make more sense for the major distribution vendors which as Novell, Red Hat and Canonical to works closely with the hardware vendors. They could possibly contract them, use a third party contractors or even work in house to tailor a given distribution to specific products ensuring every aspect of their machine works and hence can be supported properly. 

This would remove the need for disto vendors to support generic versions of their product and the need for them to include support for every device under the sun. This approach would also remove the problems associated with licensing as the product vendors could deal with individual licensing themselves. Thus communities that have sprung up around specific products could then be charged with bringing hardware-specific, from source versions of distributions to that product (much like we have with the current smartphone paradigm, xda etc). Smaller distribution vendors cold then be left to continue filling the niche they represent. Hopefully this will also weed out the distributions that don't fill any niche whatsoever or offer any-more than a theme and a few preinstalled applications on top of the distros they are built on that clog up what little mind share linux has.

This will not be a popular idea amongst Linux-ites. I am aware of that. Its represents core change and they do not like core change especially change of this magnitude, but come on, trying to support everything you end up supporting nothing and everybody loses. 

Some feel that Linux can see mainstream success on the desktop by carrying on how it is and every year seems to be “the year of Linux desktop” but i have always been a big advocate of change and unless one as big as the one mentioned above is carried out the wonderful notion of Linux on the desktop as a mainstream phenomenon  will be forever relegated to the pages of history alongside BeOS and OS/2. Superior technology that just didn't see success. 

You may not agree with my point of view, hell you might even hate me for it, but one thing we can all agree on is that it would be a monumental shame for Linux to fail when its come this far. 

Comments

I agree in principle, but developers face a rapidly moving target in hardware support. Hardware vendors targeting Windows have an obvious financial incentive to write their own drivers for the platform. They don't have that incentive for Linux because of its small market penetration and the widespread perception that the typical Linux user isn't about to actually buy something.

I think the weak penetration of Linux into the desktop market, especially if we ignore corporate and institutional environments that opt for Linux, is due as much to lack of polish and efficiency in Gnome and KDE as it is to lack of hardware support. People judge things by how they look. If what they see on screen doesn't impress them, they will have a negative impression of Linux regardless of hardware compatibility or lack thereof.

Frankly, I think the answer rests in a vendor forking off a for-profit Linux distro. That might generate revenue they could use to pay for driver support. Otherwise, I don't see those vendors taking their developers off writing marketable Windows code to write unmarketable Linux code.

I read articles like this everyday of the week, It's surprising how many people think they know this answers Yet they don't know who supports and develops Linux Kernel, Every hardware manufacturer supports and develops Linux, including Microsoft. take a look at the members of Linux foundation. and the top 500 supper computers, at the same time check how many are running windows. How many web companies run windows servers, None do, they all run on Linux, If these high profile companies advertised the fact that they use, develop and support Linux it would have taken off years ago. Again look at how many countries, education departments have and are moving to Linux, check the latest hardware drivers that's be placed in the Linux kernel over the passed 12 month's, it's a fallacy that there are no hardware drivers for Linux Do your research first then you'll know what's what

What corporations and institutions run on their servers isn't really relevant for desktop users. Two very different set of requirements.

More newer drivers in the latest kernel is obviously a good thing. But, unless those kernels are provided, painlessly, in a distribution, they will only be of use to the very few who can build their own kernels.

It's also very difficult (impossible?) for someone to easily determine how their specific hardware plays with a specific version of a Linux distribution. It's not an issue on Macs and considerably easier on Windows because anyone who sells hardware into that market ensures drivers are available. No driver on the vendor's site? Probably can't use it. It's way more complex and irritating in Linux.

Linux users and developers do agree that it would be great if hardware vendors - from components to systems - supported Linux. Perhaps you need to look a little farther back than 2006 to see the answer. Microsoft had exclusive contracts with systems vendors until its antitrust trial. It was then banned from that exclusivity. We ssuspect, though, that it is using other methods to keep vendors tethered. There are lots of dirty Microsoft tricks floating around to keep Linux and Free Software in particular, from being used by governments and large customers.

Take a look at the Microsoft antitrust trial, the Comes v. Microsoft trial in Iowa, and the antitrust actions against Microsoft in the US and the EU.

Linux will remain a niche OS until systems vendors offer it as part of their desktop market. HP and IBM both offered Linux installion and support contracts on their servers and workstations as far back as 1998, but never made the leap to the desktop. I ordered an HP business desktop just to support that effort. But I don't spend enough to change markets.

What does Microsoft interference mean for Linux hardware support? It means that hardware is developed that ONLY runs on Windows. The driver is dissected, and part of the driver support is moved from the device to the PC. There is no earthly reason to do so, other than to trip up alternative OS selection. It makes the Linux developers write extra code for such devices.

Other vendors - wireless and graphics cards are notable here - just don't want to share information on how to write a driver for their devices. They fear that their trade secrets will be revealed if the user can actually use the hardware that they purchased. This requires volunteers to write drivers for proprietary hardware using reverse engineering and disassembling.

Many of those volunteers do this as a labor of love, wanting to contribute back to the community. That is always an option, and it is the foundation on which Linux has come as far as it has. Some of us can't code drivers, but we could proof documentation, or test out new releases on the hardware that we have, or even write documentation. There are many ways that the community can make things happen.

Finally, while I have heard of problems with specific hardware, usually cheap hardware, I have not experienced that problem to the degree that you describe it. There was one release of Mandrake back in 2002 or so that gave me problems because it dropped support for an older piece of hardware that I had. Linux in general supports MORE hardware than Windows. You could look it up. The hardware that I have had to quit using under Linux was because the PC didn't have the interface slots available in the chassis any more.

Some constructive criticism: I didn't see anything overly antagonistic about Carling's note. It was critical, but stuck mostly to the facts. You have about 15 years of Linux history to fill in, and decades of Microsoft dirty tricks to fill in. Keep at it! We need more informed Linux advocates.

I suspect that English is not your first language. Thanks for writing this in a language that I can read. My great-grandfather came to the US speaking four languages, and I am stuck with one, and enough to embarrass myself in a couple others. Thanks for your efforts.

Thanks for the comment, there is, of course, a lot i still have to learn in regards the struggle between Linux and Microsoft in the early days. I am working on that :)

Also a little hurt by the "English is not your first language" remark. I'm Irish and english is, in fact my primary language.

Otherwise thanks for your reply, i appreciate anything that can improve my work and any feedback from my readers.

Two things. First, Linux most definitely is *not* a Minix clone. That fact is extremely well-documented--look it up.

Second, IMHO the main obstacle to adoption of Linux on desktops is that Microsoft has been able to maintain its monopoly so effectively. Walk into any retail store and try to find a system that has Linux pre-installed. All you will find is Windows, and perhaps Apple. Online, Dell, though it has offered a few models with Linux installed, has configured their Web site such that even someone who goes there explicitly looking for Linux boxes is lucky to find them, while the average visitor would never even know they were there. The mantra is "Dell recommends Windows [fill in the version]."

All those off-the-shelf systems have Windows already installed, with device drivers installed and configured. All the work is done for the purchaser. Anyone who wants to run Linux on one of them must first pay for Windows, then throw it away and do their own Linux install. (Yes, dual-boot is often, but not always, possible.) Some manufacturers consider Linux installation cause to void the warranty. If users had to install Windows themselves, I believe most would be incapable of doing so--but they don't have to, which is the point.

It's true that the lack of drivers for newer hardware can make Linux installation impossible. A friend of mine recently bought a Lenovo laptop and asked me to install Linux on it. I couldn't get the wireless networking to work, so she's stuck with Windows. In the process of searching for a usable driver, I discovered that the version of the Windows driver that came with the machine was 0.01--thoroughly tested and debugged, I don't doubt! Those who buy a box from Zareason or one of the other online producers of machines that have Linux pre-installed don't have such problems. But those operations are small and unknown to most computer buyers and have no presence on retail store shelves. There, Windows rules absolutely.

A second reason for low adoption of desktop Linux is proprietary software that runs only on Windows. My standard response to people who ask me to fix their broken Windows systems is to offer to install Linux for them. Many have taken me up on it and are happy they did, but just as many have not because they rely on some software that won't run on Linux. Photoshop has been a frequent stumbling block, along with iTunes and online collaboration systems that require .NET or custom Windows DLLs to run. And, of course, games.

My prediction is that Microsoft will continue to control the desktop/laptop market as long as it exists, with perhaps small but steady inroads by Linux. The thing is, that market's share of the total computer pie is rapidly shrinking as people move to tablets and smart phones for their computing needs. MS has very little presence on those other sectors and certainly will not be able to extend their monopoly to them, no matter how many dirty tricks they come up with. Instead, Android, running on top of a Linux kernel, is on its way to dominance.

Did you know that Linux devs even got together and officially offered to write Linux drivers on their behalf if the hardware manufacturers thought it was too hard or too costly... (I believe the offer is an ongoing one)

It's not that Linux doesn't support a lot of hardware -- it's that some hardware doesn't support Linux.

- - - - -

And I'm old enough to recall when even semi-savvy "consumers" got the point that "Win-hardware" ("win-printers", "win-modems", etc.) were something that only the most ignorant would buy, because producing such "hardware" was at best a sign of cutting corners and/or incompetence, and at worst cynical exploitation of the customer. It got to the point that even most sales-people at "Big Box" chains were likely to warn customers away from that junk.

Seriously, I think that might well be a point worth emphasizing even today. No matter how slick or apparently cool a product appears to be:

if the label says:

"Made Especially for Windows", (or worse, "Made Especially for Windows version X"),

and doesn't mention other operating systems, the manufacturer is probably being lazy, sloppy, incompetent and/or ripping you off. And if the hardware actually won't work with anything else but Windows -- make that "definitely".

Cellphones run Linux. Super-computers run Linux, The big stock exchanges all run Linux. Google and eBay run Linux. GPS locators and ebook readers and household entertainment devices run Linux; So if your hardware won't run Linux, well -- that says a lot more about your hardware than it says about Linux.

Seriously.

Where do I even begin on this mess of an article? The history is so flawed. Linus Torvalds wasn't creating a Minix clone. He was creating a free UNIX-like (not Minix-like) alternative because of the deficiencies in Minix. Calling Linux "the little Minix that could" is ridiculous in the extreme.

Then there is the whole point of your article: hardware compatibility. Linux is compatible with more hardware than any other OS bar none. That certainly includes Windows. Try installing Windows 7 on some random laptop from scratch and see how much is missing or unsupported without third party drivers. I'm not saying the anecdotal stories you relay aren't true. I am saying that the idea that 35-40% of the drivers needed are routinely missing is arrant nonsense when you try to make it a general rule. My experience is that a lot of off the shelf hardware "just works" and the rest needs proprietary drivers downloaded to make it work, just like Windows.

Your solution? Have Linux vendors like "Novel", Red Hat and Canonical work with hardware manufacturers. Sorry to tell you this Novell was broken into pieces and sold off in 2010 and no longer exists. SUSE is owned by Attachmate. Second, and more important, what makes you think it hasn't been tried? Canonical worked with Dell and HP and that didn't pan out. Neither did Novell's attempts to work with HP and MSI. The most successful attempts, which I wrote about in my last O'Reilly article, were the partnership between ASUS and Xandros and the similar pairing of Acer and Linpus Technologies on netbooks. As I related in my article last week, the retail marketplace is structured in a way to make Linux an unprofitable choice compared to Windows.

Let's also take the cases of HP and ASUS as examples. HP withdrew their Linux offerings the day Windows 7 was released. Does anyone think that was a coincidence? Then ASUS showed an Android based EeePC at Computex. The next day they were publicly apologizing and saying there would be no such offering, with a Microsoft VP on the stage next to the CEO of ASUS. Microsoft has the clout to undercut any manufacturer that doesn't tow the line.

You suggest an Apple like solution where only a limited hardware set is supported. That's worked so well for Apple that their market share for Mac is perhaps 1% higher than Linux. Gee, that's a great example. Clue: Apple makes their money on iPhones, iPods and iPads. Android is competing well in the phone and tablet markets. You go on to say that "Linux-ites" won't like your solution. Of course not. It couldn't possibly work. There is plenty of history to show that it can't. A little research goes a long way.

Your closing thoughts: if the Linux world doesn't follow your suggestions it will go the way of BeOS and OS/2. Funny, neither of those operating systems enjoyed the success Linux has on the desktop now, let alone the server room and in embedded devices where it really is a major success. Please reread the comment by BernardSwiss above as it really is excellent. The comparison is completely fallacious.

Don't worry Patrick, we won't hate you for your writing. We may mock you and make fun of you, but there is no reason to hate.

Oh, and it's Rube Goldberg, not Rude. Geez.

Thank you for the corrections you made. They do make for a much better article.

I did get the context on the Mac comments. I'll point out that HP and Dell are two of the largest vendors out there. In the case of Dell they offered desktops and full sized notebooks as well as netbooks preloaded with Linux. For about a year they were claiming the Linux netbooks made up one third of their netbook sales. It still all died on the vine, not due to lack of consumer acceptance and certainly not due to technical problems -- the resulting Ubuntu builds worked very well indeed. Microsoft still had the power to quash the whole thing and the retail market still is going to go with what lets them sell add on software.

You missed the point of my "arrant nonsense" comment. Let me try to rephrase: I have no doubt that the new user had the problems you describe. I don't doubt your integrity or the accuracy of what you describe. My point is that you are taking one anecdotal experience and claiming that is a typical experience. It's anything but.

Finally, I'm from New York. We New Yorkers tend to say what we mean and mean what we say. We're to the point and we don't mince words. Being direct or calling things as I see them may be blunt but it certainly isn't trolling nor is it a personal attack on you.

Displaying all 11 comments

Add comment