Weekly Edition
Return to the Announcements pageSponsored link
Vyatta –
Linux & Open Source
Alternative to Cisco –
Advanced Routing,
Firewall, VPN, QoS..
Free Download ->
| |||||||||||||
Linuxant releases DriverLoader
LINUXANT RELEASES REVOLUTIONARY DRIVERLOADER FOR BROADCOM WIRELESS LAN DEVICES MONTREAL, QC Oct. 10, 2003 - Linuxant inc., a world-class supplier of consulting, software development and professional support services is announcing the immediate availability of DriverLoader, a revolutionary compatibility-wrapper allowing standard Windows NDIS drivers shipped by hardware vendors to be used as-is on Linux x86 systems. DriverLoader technology is the ideal Linux solution to support devices for which no adequate native open-source drivers are available. It also allows vendors to drastically reduce time to market or eliminate the need to support multiple drivers for Windows and Linux. By using the same driver on both platforms, significant resources can be saved. Thanks to DriverLoader, owners of 802.11g 54mbps Wireless LAN devices (CardBus and PCI) based on Broadcom chipsets can now use their devices under Linux, enjoying the full speed of the latest Wireless LAN technology with the freedom of the renowned open-source operating-system. Broadcom chipsets are the foundation of many Notebook or Desktop Wireless LAN adapter products by numerous vendors, such as Belkin, Buffalo, Cisco/Linksys, Dell, eMachines, Fujitsu, Gateway, HP/Compaq, Microsoft, Motorola, etc.. DriverLoader packages can be downloaded from Linuxant's web site at no cost*. The software is easy to install on any supported Linux distribution (RedHat, SuSE, Mandrake, Debian) with any recent 2.4 or 2.6 kernel, and includes a user-friendly Web-based configuration system. Vendors interested in using DriverLoader technology to enable their products under Linux should contact services-AT-linuxant.com. For more information or to download your copy of DriverLoader, please go to http://www.linuxant.com About Linuxant Linuxant is a world-class supplier of consulting, software development and professional support services. Linuxant works closely with leading vendors and OEMs of semiconductor, PC, embedded and communication/wireless products, as well as with companies in other industries, providing technological expertise and solutions to maximize the potential of Linux and open-source. Additionally, Linuxant develops and distributes specialized system software, such as device drivers for specific applications. * Linuxant is happy to provide free trial DriverLoader licenses, while discussions are under way with hardware vendors to finance development costs. Linuxant hopes that DriverLoader will remain free for end-users. (Log in to post comments)
exclusion of reverse-engineering Posted Oct 12, 2003 10:14 UTC (Sun) by mk270 (guest, #4485) [Link] Unless I'm mistaken, certain types of reverse-engineering may not validlybe excluded by copyright licensing (somewhere in section 50 of the Copyright Designs and Patents Act 1988 (as amended)).
You can redistribute, but you will need a license from linuxant... Posted Oct 12, 2003 10:31 UTC (Sun) by dambacher (subscriber, #1710) [Link] I just installed this driver, since v 1.04 you will need a license number from linuxant to activate the driver.They currently give away a trial-license up to november for free in their web shop(you have to register for it, though) But get me right: I thik it's just fair to ask for some amount, since linuxant has the work, takes the support and takes the NDA's we don't want to take. bye Ulf
You can redistribute, but you will need a license from linuxant... Posted Oct 13, 2003 9:14 UTC (Mon) by mmarq (guest, #2332) [Link] "It seems like they will charge a fee for it like they do with the connexant modem drivers"The big question here, is not what kind of reverse engineering applyed but transform what was always a pure "freeware"(the only one), as in "free beer", of the M$ windows world into a payed model in the Linux world. IMO THIS IS NOT EVOLUTION, BUT INVOLUTION... IN A STANDARD FORM, ITS LIKE M$ NGSCB/Paladium,..., YOU END UP TO HAVE TO PAY HEAVLY, WITH LOTS OF ACTIVATION KEYS, JUST TO USE YOUR MACHINE. Wouldn't it be better to the Linux community to team up with the "Hardware Industry", and define interfaces for "NATIVE" Linux drivers ??? There would be a hole new era of "payed oportunitys" for hackers to develop drivers for Linux, proprietary and Open-source. With defined interfaces in place, i doubt there would be a single Vendor/Manufactor that wouldn't release their older and out of production "Hardware Driver" in a Open-Source form,..., patents allowing and M$ bulling against, even so it would be far better than today model where in 2.4.23 "we" are still fixing drivers, by presentiment and reverse engineering, mostly for 1 year old hardware.
Linuxant releases DriverLoader Posted Oct 12, 2003 10:53 UTC (Sun) by standby (guest, #9796) [Link] Under fair use , it is likely permitted by law for reverse engineering...Sony v. Connectix, 203 F.3d 596 Sony v. Bleem *(U.S. District Court of the Northern District of California)
Linuxant releases DriverLoader Posted Oct 12, 2003 10:53 UTC (Sun) by s52d (guest, #2199) [Link] Does it work for ANY device? Video, disks, serial etc?On reverse engineering: It has been done by mplayer loading windows.dll codecs. No, I can not do it, I do not have any information on MS side. BR
Linuxant releases DriverLoader Posted Oct 12, 2003 11:23 UTC (Sun) by dr_lha (guest, #86) [Link] Nice idea, but it doesn't get rid of the need for an opensource version of the Broadcomm driver (for example). In my case for instance I have a linux laptop with a Broadcomm WiFi card for which this won't work: An Apple Powerbook. So unless we want to live in a world that Linux only supports x86 I'll be staying clear for now.
Linuxant releases DriverLoader Posted Oct 13, 2003 2:42 UTC (Mon) by d726f6e (guest, #150) [Link] Yes, I agree. As Powerbook owner, I would say that we should NOT support this particular innovation from Linuxant.
Linuxant releases DriverLoader Posted Oct 12, 2003 13:16 UTC (Sun) by jeroen (subscriber, #12372) [Link] ... By using the same driver on both platforms, significant resources can be saved.Enjoying the freedom? But if you use those drivers, that part of that operating system doesn't give you any freedom anymore. That's why we should request and/or write free drivers, because an operating system with non-free drivers is still a partly free operating system.
Linuxant releases DriverLoader Posted Oct 12, 2003 13:17 UTC (Sun) by rknop (guest, #66) [Link] I have mixed feelings about this sort of thing.On the one hand, it makes devices available to people who otherwise couldn't use them. On the other hand, it makes it that much easier for a company not to release programming specs for its devices, saying, oh, well, you can use the windows driver with this adapter widget. The result: as more new hardware comes out, we're going to be less and less likely to get open source drivers, and the community is going to be less and less likely to get the information it needs to make open source drivers. In the long run, this is not a good thing. On the gripping hand, if for philosophical or practical reasons you don't want to use binary-only drivers (or driver adapters) (and, yes, there are practical reasons too), it will get harder and harder to stick by your guns. E.g., where I am, in a scientific department, right now I try to convince people to buy hardware supported by default open source drivers. As others become aware of binary-only solutions, though, they'll consider them solutions, and will consider me some sort of fuddy-duddy idiot for trying to avoid them. I already get into debates I'd just rather not get into about whether or not one ought to buy nVidia video cards. (I'm not the "not" side of that discusson, because I don't want to be trapped into needing binary-only drivers forevermore for the hardware that I buy.) -Rob
Linuxant releases DriverLoader Posted Oct 12, 2003 17:20 UTC (Sun) by dr_lha (guest, #86) [Link] On the gripping hand, if for philosophical or practical reasons you don't want to use binary-only drivers (or driver adapters) (and, yes, there are practical reasons too)Too right, the main practicle reason being that this driver won't work on a non-x86 architecture machine, Apple laptops for example.
Linuxant releases DriverLoader Posted Oct 12, 2003 20:55 UTC (Sun) by rknop (guest, #66) [Link] Too right, the main practicle reason being that this driver won't work on a non-x86 architecture machine, Apple laptops for example. Even if you have the supported architecture, there are practical reasons. For instance, if the company stops supporting their driver, you may be stuck forever more with a given version of the Kernel or of glibc; it will be impossible to upgrade and keep up with distributions. Not a problem for a year or so, but eventually a very nasty problem. (I've even had this problem with drivers where the source was available; the RaidZone devices, for example, never bothered to get their drivers into the main kernel tree, so I was using 2.2.x for a long time after 2.4.x came out. There, at least in principle I or somebody else *could* have updated the patches myself, though, so it wasn't a worst-case scenario. If it were a binary-only driver, I would simply have had no choice at all in the matter.) -Rob
Linuxant releases DriverLoader Posted Oct 13, 2003 10:20 UTC (Mon) by iabervon (subscriber, #722) [Link] On the other hand, this is a good argument for having an independant shim between a binary driver and the kernel. If you're using a Win32 driver with a shim to connect to Linux 2.4, and the driver maker goes out of business, you can use a shim to connect to Linux 2.6. I've actually thought for a while that nVidia (or someone) should write and release a GPL shim like the thing Linuxant has written (possibly only sufficient for the hardware they want to support in the version they release). Then it is perfectly clear that the binary driver isn't violating the GPL, because it's a Windows driver which happens to be usable in a patched version of Linux.
Linuxant releases DriverLoader Posted Oct 14, 2003 5:59 UTC (Tue) by Wol (guest, #4433) [Link] ESPECIALLY if that shim gets round the module restrictions in the kernel GPL by loading the module into user-space.People might not like that "evasion of the spirit of the GPL", but it means (a) that those of us who want a Free kernel don't have it tainted by closed binaries, and (b) it provides a very strong marketing advantage for any company that IS prepared to release a Free driver. We all know how the user-kernel context switch can hammer performance. Not too bad if you've only got one or two such devices, but it's not going to be popular with the punters. And then one company bites the bullet, and the rest are forced to follow suit :-) Cheers,
Linuxant releases DriverLoader Posted Oct 13, 2003 7:16 UTC (Mon) by beejaybee (guest, #1581) [Link] I'm ambiguous, too.Another problem is that having this non-transparent conversion layer in place does two things which may not be immediately apparent: (1) overheads may make the load on linux systems excessive; (2) if there are _any_ flaws (like buffer overflows) in the conversion layer then these could be used to attack linux systems as a class. There is a particular problem here if a driver software provider wants to ensure that their driver only works properly under Windows. As others have pointed out, so long as it's a realistic option for people who don't need drivers for "state of the art" peripherals to not run this module, I'm reasonably happy. However, I'd be _much_ happier with an open-source alternative - so long as it was built from first principles, not ripped off by reverse engineering Linuxant's work.
Linuxant releases DriverLoader Posted Oct 13, 2003 9:53 UTC (Mon) by mmarq (guest, #2332) [Link] "if there are _any_ flaws (like buffer overflows) in the conversion layer then these could be used to attack linux systems as a class"One more reason to consider, that a well defined interface for "Native" Linux drivers (not this translation layer), would contemplate a "hardening mechanism", that not only would kill a faulty driver from running, but also would stop the LinkSys sindrome of kernel abuse... There was a "hardening patch" circulating by last year october, near the 2.5 feature freeze, and all the others interface(s) would not be much different from what is today for the USB layer... With the progress in the udev, kobjects, sysfs, dev_t, structured driver classes, only misses not much than a in kernel D-BUS, for we to have well defined interfaces for all classes of hardware out there.
Linuxant releases DriverLoader Posted Oct 13, 2003 14:52 UTC (Mon) by pflugstad (subscriber, #224) [Link] > On the gripping hand,Been reading Niven/Pournelle lately :-)
DriverLoader GPL Posted Oct 12, 2003 13:21 UTC (Sun) by johanrd (guest, #10010) [Link] Can it be a more direct and clear breach of the GPL?
DriverLoader GPL Posted Oct 12, 2003 19:48 UTC (Sun) by roystgnr (guest, #14278) [Link] It's not very clear to me. For one thing, Linux is licensed with an exception to the GPL to allow binary modules; NVidia's been doing it for a while. For another, to break the GPL you actually have to redistribute GPL'ed code. It's possible for driver authors to create a Linux driver using just the public interfaces (rewriting header files if necessary) and then distribute that driver in binary form without redistributing any derivatives of Linux code with it. Acts like that are allowed by copyright law, and the GPL can't take that permission away.
DriverLoader GPL Posted Oct 12, 2003 20:14 UTC (Sun) by torsten (guest, #4137) [Link] "Linux is licensed with an exception to the GPL to allow binary modules;" If intent matters to copyright licenses, then Linus intended the exception to apply only to binary user-space programs. kernel-space drivers are still of questionable legality.
Linus & binary only modules -> pragmatic Posted Oct 13, 2003 6:20 UTC (Mon) by Duncan (guest, #6647) [Link] > Linus intended the exception to apply only to binary user-space programs.> kernel-space drivers are still of questionable legality. Check again. Linus personally has proven quite pragmatic in his approach to closed source modules, as he has in other areas. Isn't it he that in the context of BitKeeper has said only an idiot would choose idealism over functionality, or something equally discomfitting to the idealists (including me, to some extent, tho I do unfortunately still run an NVidia driver, but my excuse there is that I got the card b4 I actually switched from MSWormOS, and b4 I realized there was a difference between Linux support and LIBRE Linux support, and intend to purchase a more compatible solution next time)? Yes, the 2.5/2.6 kernel *DOES* have a number of changes that make it stricter in this regard, but from all I've read, it's other kernel hackers driving that, and Linus going along with it, tho he's specifically stated that binary modules do NOT violate the Linux kernel license as far as he's concerned, as long as they don't change the preexisting code already there and try to keep THAT closed source. In fact, that has, again from my reading, along with the BitKeeper thing, been one of the reasons some of the more idealistic folks including RMS to some extent have been rather uncomfortable with Linus and the Linux kernel, and possibly one of the reasons keeping the HURD project going. That said, as mentioned, 2.6 is more strict with closed source module requirements than ever before, and there's speculation that such modules may be eventually disallowed altogether, but such will be kernel 2.7 at the very EARLIEST.
Linus & binary only modules -> pragmatic Posted Oct 13, 2003 10:34 UTC (Mon) by mmarq (guest, #2332) [Link] "he that in the context of BitKeeper has said only an idiot would choose idealism over functionality"But somehow Linus is right!... the only problem here is that there is a "idealogic" confusing "allowing binary only" pragmatism, with the functionality of the implementation of that pragmatism(worst than Babel tower and crapy)... IMO, its crystal clear like mountain water, that well defined and consistent "interfaces"( always say it in plural for not confusing with the M$ general HAL style) are absolutly necessary to bridge "binary only" with "Open-Source". If you lose some flexibility of what you can do with the kernel, but with that you raise sky high their (kernel) usefulness, than you are raising functionality, even considering that are less functions than in the wildest form.
Linus & binary only modules -> pragmatic Posted Oct 13, 2003 12:04 UTC (Mon) by Zakaelri (guest, #15087) [Link] That said, as mentioned, 2.6 is more strict with closed source module requirements than ever before, and there's speculation that such modules may be eventually disallowed altogether, but such will be kernel 2.7 at the very EARLIEST.I think that in the case where proprietary drivers are disallowed entirely, that there should be a major version number change... because the way that you program the kernel changes. It's not an interface change from the machine-side of the fence, but the machine-side is not the only side that you have to worry about. You have to think of the human interface as well. As far as I am concerned, I dislike the fact that Linuxant has closed-sourced their driver, but I support them for doing so... Predominantly because that shields the kernel from supposed `copyright infringement' (or a reasonable facsimile) from Microsoft (should they be able to find a way to do so). As for the purpose of the driver, I again disagree on moral grounds, but one could recall Wine's Argument... in all practicality, it (making windows drivers work in linux) is the best way to get users to bridge the windows-linux gap. After we get more people on `our side of the fence', then it would make more sense for the vendors to provide drivers... all we would have to do is convince them to open-source. (or if the kernel requires it, they would really have no choice. Personally, I would rather have them come to terms with the morality and practicality of open source, rather than the terms of a lisence.)
Linus & binary only modules -> pragmatic Posted Oct 13, 2003 12:15 UTC (Mon) by mmarq (guest, #2332) [Link] "As for the purpose of the driver, I again disagree on moral grounds, but one could recall Wine's Argument... in all practicality"But that is the real trick,... WINE could had been a trap also,... if you had "drink" it to much to the point of making it the general standard than there wouldn't be a native linux software thriving,... on the other end it can prove to be so usefull in small doses.
Linus & binary only modules -> pragmatic Posted Oct 13, 2003 14:03 UTC (Mon) by coriordan (guest, #7544) [Link] > pragmatic in his approach to closed source modulesPragmatism is in the eye of the beholder. I see Linus's stance as short-sighted. If he had insisted on "Free Software"-only from the start, we would have a lot more Free drivers today. Ciaran O'Riordan
Linus & binary only modules -> pragmatic Posted Oct 13, 2003 14:51 UTC (Mon) by mmarq (guest, #2332) [Link] "If he had insisted on "Free Software"-only from the start, we would have a lot more Free drivers today"That's the intolerant, more ideologic or more pragmatic, point of view... Can you also consider the possibility , in the ever more complex hardware path, of a counter-reaction that could leave, that "Free Software" with no clues what so ever about the hardware that it would suppose to run on ? Reverse Engineering is getting ever more complex and difficult, like the hardware, and is not the cure for the problem, but only a band-aid,..., and what use is a "COMPLETLY FREE SOFTWARE" if it has no place to run ? (M$ answer to that is the forward coming NGSCB/Paladium=kill free software)
Linus & binary only modules -> problematic Posted Oct 13, 2003 15:43 UTC (Mon) by coriordan (guest, #7544) [Link] A lot of companies stand to increase their profits if they can convince us that we won't win without their proprietary software. I won't believe this. They told us we'd never have a Free gui browser, desktop, office suite, ...As the market-share for GNU/Linux is ever increasing, eventually companies would realise they are harming themselves. Free Software has fended off too many giants to give up at this point. We should stand firm. How long will they hold out? depends on how much they have to lose (not much really). We stand to lose a lot more. I'm not saying no one should ever run these binary drivers, or anyone that does is evil. Life is too complex for simplifications like that. I am saying we should aim (do our best) to reject proprietary add-ons such as drivers, think of the long term problems, and let others know that these problems exist.
Linus & binary only modules -> problematic Posted Oct 13, 2003 17:35 UTC (Mon) by mmarq (guest, #2332) [Link] But if you make "binary only" or "source" drivers interface the same way, because everybody can make a binary driver of the source, and is of covenience of the free software, and not of the proprietary parasites, to have drivers splited from the rest of kernel,..., then free software has all to win.If not, can you tell me why should developers rewite endlessly "good and stable" drivers, for obsolete hardware, every time a kernel feature change ? If not, and Linux happen to get the same hardware support than Windows, why should developers end up wasting perhaps 4, 5 or 6 years, fixing drivers that no one or very few people use by that time, to get to a stable serie, only because they refuse, in fear of giving in, any other way ? IMO the "proprietary binary vs source" driver, is a false question!
Linus & binary only modules -> problematic Posted Oct 13, 2003 18:38 UTC (Mon) by coriordan (guest, #7544) [Link] Drivers do not have to be rewritten endlessly, no one wastes time supporting unused drivers, and a stable release is never held up for a driver. I don't know where you got these ideas.It is very rare for a change in Linux to require a change in the code of a driver. But the driver interface changes frequently. This is not a problem for Free drivers because when you recompile Linux, the driver is automatically updated (compiled against the new header files). This is convenient. These interface changes do break binary modules though. So people using proprietary modules can only update Linux to a version that is supported by their proprietary driver vendor. This is inconvenient. For some this is a small inconvenience, but it's completely unnecessary. By hiding the source, they're crippling it. If they GPL'd it, people could tweak the performance and use the driver with any kernel version etc.
Linus & binary only modules -> problematic Posted Oct 13, 2003 19:28 UTC (Mon) by mmarq (guest, #2332) [Link] Sure;" a stable release is never held up for a driver" But its a obvious idea that its functionality is. So it makes full circle.
Not that very rare, as consequence of what you point next "But the driver interface changes frequently"
Isnt that a terrible time consuming, if you consider thounsands and thounsands of in kernel drivers (near the support of windows), and are a "very rare" hundreds and hundreds of drivers that need real fixing ?? And what about if you consider that "even more rare" hundreds of drivers that only need new header files are still very much in use by Linux users, but very clearly marked as absolet and out of production by the Hardware Manufactores. Dosent that gets worst every year that pass ??
Linus & binary only modules -> problematic Posted Oct 13, 2003 20:15 UTC (Mon) by coriordan (guest, #7544) [Link] me>"It is very rare for a change in Linux to requireme>a change in the code of a driver" mm>Not that very rare, as consequence of what you point next you misunderstand. The interface is defined by header files. When a person is compiling a driver, the compiler learns the interface from the header files and creates a compatible binary. With proprietary drivers, only the vendor can compile them, so you can only use the binaries they give you. If they don't give you a binary that works with 2.4.22, you can't use your video card with 2.4.22. With a Free driver, anyone can compile it. So drivers will usually be available for a larger number of kernels, and if no compatible driver is available, you can compile it yourself (or ask a friend to do so). > Isnt that a terrible time consuming ... a "very rare" No. I've never seen a Linux change that has required all drivers to be altered/fixed. Driver changes can often be handled by generic wrapper code, or the changing of a library common to all drivers of a given type or brand. These changes are simple when you have the code, impossible when you don't. If you are using a cool video card with a proprietary driver and the vendor loses interest in GNU/Linux, how will you use your card under Linux 2.6 or 3.0 or 3.2? You won't be able to. Compiling a driver isn't a hassle. Everyone doesn't have to compile their own drivers. Only one person (anyone) has to compile it an make it available to everyone. With proprietary drivers, only one person has the ability to compile the driver. If they don't feel like compiling a driver that's compatible with the kernel you want to use, you can't do anything about it.
Driver flag days Posted Oct 13, 2003 20:33 UTC (Mon) by corbet (editor, #1) [Link] "I've never seen a Linux change that has required all drivers to be altered/fixed." 2.5 had a few flag days. See the interrupt handler change, for example. The early 2.5 kdev_t change was another. It happens reasonably often. This isn't necessarily a bad thing; I believe it's vastly preferable to accumulating compatibility cruft over the years. But it does most certainly happen.
Driver flag days Posted Oct 13, 2003 22:54 UTC (Mon) by coriordan (guest, #7544) [Link] It didn't seem like a big deal but a quick grep says about 600 files were altered for the interupt handler change. Maybe 600 three-line changes isn't a big deal to the horde of Linux hackers.Still, it's a good example of the proprietary driver problem. The fix was dead simple, but people using proprietary drivers can't fix it. The vendor may or may not be interested in providing a new Linux driver. They probably will, but maybe next time they won't. etc.
Linus & binary only modules -> problematic Posted Oct 14, 2003 17:22 UTC (Tue) by mmarq (guest, #2332) [Link] I'm not an Expert on the matter, But,...I've compiled plenty of Linux kernels and other binarys,..., to have some clue! The core of diference of opinion here is that you insist in seeing the problem has Binary vs Source, while i insist in seeing it as a matter of better driver model for kernel developers;... And yes i belive that a SPLIT driver model, that has its lowest driver part source out of the main kernel source and interface to it by the form of a API/ABI (something like DKMS !?), its the best approach. If more strong opined kernel developers tend to tell me to go way, i guess they will get to see my point, if all the hardware industry starts to give what they demand, and happens truth the possibility of the main source of the kernel to have all the drivers in the planet and its source size reach near 1 Gigabytes. If 150Mb of 2.5.* takes near to 2.5 years(?) to get to a decent stable 2.6.0, how many years will take about 500 Mb(example) ??? (I never mentioned not even once "proprietary" and or "facilitate them")
DriverLoader GPL Posted Oct 13, 2003 1:45 UTC (Mon) by johanrd (guest, #10010) [Link] No, there is no allowing of binary modules, this is a misconception, check Alan Cox's many posts on the subject.The program in question is clearly a program derived from Linux, therefore its creators have to comply with the GPL. Among other things, it have to be distributed with source.
Linuxant releases DriverLoader Posted Oct 12, 2003 18:01 UTC (Sun) by ccchips (guest, #3222) [Link] What if Linus and his friends decide to modify the open-source part of the kernel so that these binary-only driver loaders don't work, and LinuxAnt goes out of business before then?I will never undarstand how people fall for this crap, and then come complaining to administrators and programmers like me about orphaned products, and how can you fix it for us, CCC? If you money-oriented Linux users are smart, you'll reverse-engineer this thing, develop an open-source alternative, and test the license in court. There are other ways to make money besides making a fast buck convincing manufacturers they no longer have to support Linux or distribute their hardware specs. Very, very bad idea. Don't fall for it.
Linuxant releases DriverLoader Posted Oct 12, 2003 19:05 UTC (Sun) by erat (guest, #21) [Link] What if Linus and his friends decide to modify the open-source part of the kernel so that these binary-only driver loaders don't work, and LinuxAnt goes out of business before then? Then LinuxAnt goes out of business... ...and in the process, lots of folks who were considering installing Linux on their machines now that it indirectly supports driver X will decide "screw it" and re-install Windows. If you money-oriented Linux users are smart, you'll reverse-engineer this thing, develop an open-source alternative, and test the license in court. Reverse engineer the loader, or reverse engineer the hardware whose drivers are loaded by the loader?
Linuxant releases DriverLoader Posted Oct 13, 2003 9:28 UTC (Mon) by BrucePerens (subscriber, #2510) [Link] and in the process, lots of folks who were considering installing Linux on their machines now that it indirectly supports driver X will decide "screw it" and re-install Windows.That's why we'd rather they not use Linux in the first place. We know that we can't support them long-term. We can't provide them the quality of experience that is available with an Open Source driver. If the company that makes this "shim" goes out of business, their software is out of business too. Putting Linux in an application where it is likely to fail does not promote Linux. So, I think really the best thing we could do with these people is explain why they still shouldn't use Linux now, if they are absolutely bound to Broadcom devices. Bruce
Linuxant releases DriverLoader Posted Oct 13, 2003 11:21 UTC (Mon) by mmarq (guest, #2332) [Link] "That's why we'd rather they not use Linux in the first place. We know that we can't support them long-term"But isnt about time to bring the proper Vendor/Manufactor to the responsability table ? A Linux Vendor/Distro cant give endless excuses to customers because "Hardware Makers" decide not to release specs, or Linux drivers, simply because is not he's fault. If you dismiss a linux user because there is no way to make a driver for a particular piece of hardware or because the "guessed" implementation is not up to the job, than, sooner or later, you have to dismiss the majority of them. Isn't there something that can be done to invert the situation, and bring the "Hardware Makers" to take the majority of responsability for proper driver support under Linux!?... team up with them and define interfaces, perhaps!? As to this DriverLoader, its obviously a trap, that will keep the hardware industry even farthest away. Its motivation is an obvious business opportunity(that shouldn't be there already)... but i wouldn't be surprised if M$ "facilitated" somehow. Mario Marques
Linuxant releases DriverLoader Posted Oct 13, 2003 14:19 UTC (Mon) by BrucePerens (subscriber, #2510) [Link] Remember that the Linux distributions, for the most part, do not own the software copyrights and are not the responsible parties. The distributions don't really have the right to change the licensing terms on the kernel, so they can't promise to fix this or "take responsibility". The responsible parties are the kernel developers.The kernel developers are not making excuses for the hardware manfacturers. The hardware manufacturers know exactly what interface the kernel developers want to write to. They invented it, and they have already documented it. It's the bus-level interface of their device. All they have to do put that document where we can see it. We don't want their microcode, their VLSI design, or any of their property below the bus level. They have ample room to conceal their design from their competitors even while providing an Open Source driver. But if you want the kernel developers to make life easier for closed driver providers, you won't hear any excuses. What you will hear is "no, thanks", or if you are persistent about it: "go away". Bruce
Linuxant releases DriverLoader Posted Oct 13, 2003 15:59 UTC (Mon) by mmarq (guest, #2332) [Link] "The responsible parties are the kernel developers." But that's my all point,... they shouldn't be! "We don't want their microcode, their VLSI design, or any of their property below the bus level. They have ample room to conceal their design from their competitors even while providing an Open Source driver." So, why arent they doing that in mass ? Couldn't they be invited(next Ottawa symposium), not for their ruling and ease, but to have a word into the process somehow ? "But if you want the kernel developers to make life easier for closed driver providers, you won't hear any excuses. What you will hear is "no, thanks", or if you are persistent about it: "go away"." But i belive that at this point Open Source should make life harder for driver providers, making the interfaces more complex, robust and consistent, not on a wimp but to copy with the more complex hardware and buses,... and that while it has the power to define interfaces at will (before NGCSB/Paladium);... With open source first in mind , but allowing binding "EXTERNAL" binary modules in kernel space also,..., USB is almost there (couldnt it use DKMS ? for in kernel all the way) and ALSA or OSS are half the way, of a "SPLIT DRIVER MODEL", that has not the purpose of facilitate closed driver provider, but mainly facilitate open source developers that should not rewrite, sometimes from sratch, every driver they write most of times for years old and out of production hardware, simply because some feature changes in the kernel. Mario Marques
Linuxant releases DriverLoader Posted Oct 13, 2003 22:19 UTC (Mon) by BrucePerens (subscriber, #2510) [Link] Well, think of all the work that has gone into optimizing the network in 2.4 and 2.6 . A lot of this was concerned with zero-copy I/O where DMA happens almost directly from the disk controller to the network controller, optimizing the interrupt path, detaching entire busses while keeping the kernel running, and so on. Those are all very close to the hardware. Last year, we wouldn't have known how we were going to code some of this. We could not have forecast all of the things we'd need in such a driver API, and it would not stay the same from month to month if we wanted to continue development.Why do you think Linux performs 2.5 times as well as Windows Server 2003 while serving Microsoft's own file and printer-sharing protocol? Because we have control over things that Microsoft doesn't. What is the number one reason for Windows crashes? Badly-coded third-party device drivers. Much of Linux performance and reliability comes from the fact that we do control our device drivers. If we settle for less, we will have taken a big step toward becoming "just another run-of-the-mill operating system". Bruce
Linuxant releases DriverLoader Posted Oct 14, 2003 17:56 UTC (Tue) by mmarq (guest, #2332) [Link] One thing that amazed me in Linux from the start was his versability!So, there isnt nothing that prevents a stable and more static API for a split driver model where is wise to do it, and the actual paradigma, at the same time in other parts, where its better. I belive one thing is certain, there isnt a clear remedy for all deseases, and even with a large adoptance of a split model, the best argument stays with the "feature maintainer" and his experience and know-how. I belive something has to change, because as i state above that 150Mb of 2.5.* is going to take near to 2.5 years(?) to get to a decent stable 2.6.0., then, if the flush of documentation happen truth and the main kernel source gets "every driver on the planet", how many years will take about 500 Mb(example) ? Mario Marques
Linuxant releases DriverLoader Posted Oct 13, 2003 14:15 UTC (Mon) by coriordan (guest, #7544) [Link] > Putting Linux in an application where it is> likely to fail does not promote Linux "promoting linux" is a job for the marketing department of a MegaCorp. These people make proprietary software. They are a parasite to Free Software. They benefit from all the code we share, but not only do they not give back, they encourage people to give up on freedom. If they fail, it will be another example of a company that "didn't get" Free Software.
Linuxant releases DriverLoader Posted Oct 13, 2003 13:43 UTC (Mon) by ccchips (guest, #3222) [Link] I was thinking about reverse-engineering the loader, but what the hey? Why not both?The thing is, I can't picture how any kind of kluge is going to last very long; Microsoft and its bed-partners aren't going to wait around for anyone; they'll symbiotically (or parasitically??) modify their code to meet whatever criteria they choose, without regard to how it will affect Linux users. I was driving at this: these reverse-engineering laws need to be tested and rescinded. They are wrong, because when I buy a piece of equipment, it belongs to me. Period. If I choose to figure out how it works, then explain my discovery to whomever I want, and some hardware maker's buddy loses money on some proprietary software because of it, "that's the way the ball bounces, G!"
Linuxant releases DriverLoader Posted Oct 12, 2003 20:58 UTC (Sun) by proski (subscriber, #104) [Link] There are ways to isolate the binary module from the kernel by an open source wrapper that needs to be recompiled for the given kernel. It's quite easy to do. Lucent wavelan driver used it. OSS (sound driver) uses it. No user intervention is required if gcc is installed and the kernel is within the supported version range.If compatibility is broken delibetately, that would be a political act, which is very unusual for Linus. Even if Linus does it, there will be a fork or a patch to reverse the change, and most distributions will use it. The value of a distribution only increases if it allows certain things that other distributions don't.
802.11g supported by other drivers... Posted Oct 12, 2003 20:23 UTC (Sun) by goonie (subscriber, #4252) [Link] Whilst I wish the best of luck to these people, from a practical and ideological perspective if you can choose an 802.11g compatible card to work under Linux, there are better options than trying to use proprietary drivers with a binary kernel module that might work now but could stop working with any kernel upgrade.The Intersil Prism Duette chipset is supported by this project. Alternatively, there is the madwifi project which supports the 802.11a/b/g tri-mode cards using the Atheros chipset.
802.11g supported by other drivers... Posted Oct 13, 2003 0:50 UTC (Mon) by dambacher (subscriber, #1710) [Link] OK then, where can I bye a Mini-PCI Card fitting in my dell inspiron 5100?THat's the real problem: The broadcom Chips are sold packaged with the computer, you have it but you can't use it. And then you will use the linuxant driver, open source or not, just because it's availiable! Greetings
802.11g supported by other drivers... Posted Oct 13, 2003 5:05 UTC (Mon) by mathieu_lacage (subscriber, #3967) [Link] If you want to avoid binary-only drivers, you are out of luck with the madwifi driver. It contains a binary x86-compiled core which is named the HAL and is clearly proprietary...
Linuxant releases DriverLoader Posted Oct 12, 2003 22:37 UTC (Sun) by pgarland (guest, #270) [Link] Here's something I was wondering recently:Is there actually a documented case where a company has lost money as a because they either released a driver as Free Software or provided the full specs to hackers so that a Free driver could be written for their hardware? The possibility of a competitor gaining an advange over a company that provides Free code or specs by incorporating ideas for its own products from the released code or specs is often given as a reason companies are reluctant to release them, but I've never heard of a company that suffered financial loss from helping out with kernel or XFree development. Surely if this fear is a legitimate one there must have been a case in which it's happened. ~Phillip
Linuxant releases DriverLoader Posted Oct 13, 2003 11:54 UTC (Mon) by mmarq (guest, #2332) [Link] "The possibility of a competitor gaining an advange over a company that provides Free code or specs by incorporating ideas for its own products from the released code or specs is often given as a reason companies are reluctant to release them"Try patents, confidance over brand and merchantibility, and the cost of maintaining closed binary drivers for ever changing interfaces, also!... But most of all, the natural and legitimate fear of losing control over their own stuff.
Linuxant releases DriverLoader Posted Oct 14, 2003 6:17 UTC (Tue) by Wol (guest, #4433) [Link] It wouldn't surprise me if companies HAVE lost money - but that would be the M$ effect ...I think as soon as M$ is out of the picture, hardware vendors will be falling over themselves to release open-source drivers. But don't forget the suituation HP found themselves in - where they discovered they didn't own the specs to their own printers and COULDN'T release either linux drivers, OR the information to write said drivers ... Reading this thread, I must admit I've been well cheesed off with the idealists who want to drive linux into a Free ghetto. It's all very well *wanting* open drivers, but to all those foaming fanatics, can I point out the consequences? We'll end up with hardware support that's worse than SCO's (apart from all the drivers they appear to have "borrowed" from linux...). At which point, nobody in their right mind would bother with linux, and everybody will be FORCED to go back to Windows in frustration, because you just won't be able to GET any hardware that's compatible with linux, except if it's 5-yr-old stuff on the second hand market. Basically, the problem at the moment is we live in a world where the market is rigged, and the manufacturers have their arms twisted behind their backs to make our life difficult. LIVE WITH IT !!! Once the linux market is big enough for even ONE manufacturer to think that they don't need the Windows market (in other words, we need just ONE minnow to think they stand a chance of taking on the big boys :-) we will have a hardware manufacturer releasing hardware, with open drivers, that are rock solid and optimised to fly. If that shows any real sign of damaging the "big boys", you can bet sparks will fly. What's wrong with reverse-engineered drivers and closed-source drivers? Yes, I know they're crap. I damn well hope they get forced into user space in the next kernel rev. But at the moment, M$ is in a position where they can dictate terms to the hardware guys. As soon as we can break THAT, the dam will collapse :-) Cheers,
Legality of this product is questionable Posted Oct 13, 2003 9:23 UTC (Mon) by BrucePerens (subscriber, #2510) [Link] The legality of this product is questionable.Linus' own word on binary-only device drivers is that you should consult your lawyer and only go ahead only if you believe they are legal. And the fact is that your lawyer can't give you a conclusive answer. There is no direct exception for binary device drivers added to the GPL for Linux - the exception is for user-mode programs. Its application to binary device drivers is ambiguous. Linus discusses it in the old Open Sources book as if it would allow binary device drivers, but has since contradicted himself on that. The only case in the U.S. that reads on whether these drivers might be derivative of the Linux kernel is Nintendo v. Goloob Games, which doesn't directly apply and has no precedential value because it stopped in the Superior court. That means that a subsequent case may decide differently. And of course a subsequent Linux kernel may simply not support this product. Later kernels have been explicitly witholding their internal information from drivers that aren't marked as being under a free software license. IMO the best solution is to refrain from using Broadcom products, and larger products that integrate them, until they can join the rest of the world of component providers and publicly document their products. If you just buy products that can't be supported with Open Source drivers, do not expect to use them with a Linux kernel. Not only don't you have a warranty of further support, the Linux developers will laugh in your face if you ask for support and can't provide documentation on how the device works. Bruce
Is it Broadcom, or is it the FCC? Posted Oct 13, 2003 15:00 UTC (Mon) by pflugstad (subscriber, #224) [Link] I was under the belief that it's not so much Broadcom not wanting to open source their wireless drivers (although it's certainly possible that Broadcom is being obstinate). Rather, it's the FCC that is blocking "programmable" wireless drivers, as that may allow people to reprogram them to interfere with military frequencies. This prevents Broadcom and other wireless vendors from publishing the APIs.Or maybe I'm confused and this was a UK problem?
Is it Broadcom, or is it the FCC? Posted Oct 13, 2003 22:30 UTC (Mon) by BrucePerens (subscriber, #2510) [Link] I'm sorry, but Broadcomm doesn't even document their MIPS-clone CPU. It doesn't have an RF interface.Closing source doesn't really help, though - what is to prevent someone from disassembling it? Better to do the band-limiting in the chip. On the other hand, I am a radio amateur and have the right to work on some frequencies that you aren't allowed upon. I should be able to change that frequency. Bruce
Linuxant releases DriverLoader Posted Oct 13, 2003 12:45 UTC (Mon) by dambacher (subscriber, #1710) [Link] Reading the last Comment, I think, the question is:is there a web portal Do we have this? Do'nt we need this? bye
Linuxant releases DriverLoader Posted Oct 13, 2003 14:21 UTC (Mon) by BrucePerens (subscriber, #2510) [Link] I did this once. It was called openhardware.org . I got too busy and turned it over to someone else. It didn't quite fly. Now they seem to have lost the domain and it's selling watches. But I think perhaps we could start this up again and be more successful this time.Bruce
Linuxant releases DriverLoader Posted Oct 13, 2003 14:29 UTC (Mon) by coriordan (guest, #7544) [Link] Here's a URL to the article where Georg talks about a similar project:http://www.gnu.org/brave-gnu-world/issue-51.en.html It may be useful to work together(?) or at least share ideas.
Linuxant releases DriverLoader Posted Oct 13, 2003 22:31 UTC (Mon) by BrucePerens (subscriber, #2510) [Link] See also open-hardware.org
Linuxant releases DriverLoader Posted Oct 13, 2003 23:22 UTC (Mon) by coriordan (guest, #7544) [Link] hmm, I read all their pages, but I still don't know what they are going to certify :) hardware for which the producer (or third party hackers?) provides: specs? drivers? Free Software drivers? They claim backing from a lot of big hardware manufacturers though, so I hope it goes somewhere good.As a note, the idea Georg discusses sounds impractical to me. Maintaining such a database would be a full time job. But the direction is interesting. A chart where people could check "is card X supported by Linux and XFree86" would be great and should be maintainable. (ps/OT glad to hear you're doing something on the GFDL/FSF/Debian issue, hope it goes well.)
Linuxant releases DriverLoader Posted Oct 13, 2003 14:22 UTC (Mon) by coriordan (guest, #7544) [Link] Amazingly, it doesn't exist.Georg Greve of fsfeurope is trying to get such a portal going.
|
|||||||||||||