Slashdot Mirror

← Back to Stories (view on slashdot.org)

The 2.7 Kernel: Back To The Future For Linux

Posted by timothy on from the but-what-does-linus-want-in-it dept.

Anonymous Coward writes "Now that the Linux 2.6 kernel has been released and is being worked into distributions, many in the open-source community are turning their attention to the next development and test kernel, known as the 2.7 tree. To get an early glimpse at some of the thinking going into the next kernel, key vendors that aid in shaping the Linux kernel helped eWEEK last week put together a long-range wish list for 2.7."

9 of 437 comments (clear)

  1. move along by Anonymous Coward · · Score: 5, Insightful

    There is nothing specific about anything. What a useless article. You can say you want a milkshake with your 2.7 kernel and it be just as valid as the things mentioned.

  2. What would be a great "desktop focus" by BizDiz · · Score: 5, Insightful

    Is just great driver compatability. That seems like the primary hurdle that can really keep people out, as well as a large area that is easily neglected in a more server-oriented mindset (especially in terms of user peripherals).

    1. Re:What would be a great "desktop focus" by Cthefuture · · Score: 5, Insightful

      I want to second this opinion. One of my major problem areas with Linux has been the drivers or lack of.

      I know the die-hards will nay-say this, but being able to use native Windows drivers would be absolutely great. Now, maybe you don't use MPlayer (and the other "native" driver apps) but there are a hell of a lot of us that do and love it. The same thing should be done for all drivers. Video, USB, firewire, PCI, whatever... Make it so we can use Windows drivers in Linux because there are way too many half-assed reverse engineered Linux drivers that just don't work right. I mean, when in the hell will my Wacom Intuos2 tablet finally work correctly?! (I this is not just a kernel problem but XFree too) Yes, yes, I know about those patches here and there, but try to get them to work with XFree 4.3 and kernel 2.6... Ain't gonna happen. Just let me use the Windows drivers please.

      I don't give a crap about some utopian vison of Linux greatness because all manufacturers support Linux. It isn't happening any time soon and I have real work to do.

      With that said, my #1 greatist wish for 2.7/8 would be to get the damn SBP2 Firewire drivers working correctly. Dammit, that thing has been broken since it was introduced. Nearly every time I boot my system I have to plug and unplug the firewire cable (sometimes several times) to get the devices reset and loaded properly so I can access them (I'm using kernel 2.6, but has always been broken like this). The read/write/timeout errors have gotten better but they still occur with large drives. I'm absolutely terrified that one day I'll have to fschk my 90 GB partition on my firewire drive again. The last couple times I had to do that it toasted the partition every time (I/O errors and timeouts).

      --
      The ratio of people to cake is too big
    2. Re:What would be a great "desktop focus" by Cthefuture · · Score: 5, Insightful

      For one, I believe that much of the Windows driver API is an industry secret.

      Nah, get a MSDN subscription (or just the DDK). Not everyone that writes drivers works for Microsoft. The device developers kit contains everything you need to know.

      Second, people won't write Linux drivers anymore, and requests for native drivers will be responded with "You can just use the Windows drivers!"

      Frankly, I don't see a problem with that. If it works, it works. You're still running Linux. When enough people start using Linux then they can create real native Linux drivers. It will happen, given enough time. You can't just expect a company to drop everything to support some niche market where they won't make money or will lose money. We need a bridge, even if only temporarily.

      Third, Linux and Windows have very different driver models, meaning incompatibility and having to code it in bug-for-bug.

      Bah, there are so many crappy, incomplete, or just plain missing Linux drivers. Something is better than nothing. I have not noticed any superiority of any Linux driver over its Windows conterpart. The nVidia drivers are sometimes a bit faster in Linux but guess what? Those are made by the manufacturer not some wannabe college student Linux programmer. Not all the time, but very often the best software is a result of someone getting paid to write it. It works because they simply must finish it or they will get fired (or not payed). Plus generally the management listens to the customers (the users), and stuff gets done (enhanced or fixed) because money is at stake.

      Plus, how can you ensure that the Windows drivers won't trample over the Linux ones, or are you going to isolate them, reducing their effectiveness?

      It's just code. It's not that terribly complicated except for the undocumented Microsoft crap, but believe it or not, drivers have well defined interfaces. That's how all those 3rd parties create Windows drivers in the first place.

      However, an operating system kernel is not the place for untrusted code that depends on a lot of stuff working at the right time.

      Um... whatever, you want to eliminate all binary drivers? Sorry, but that isn't going to happen. Linux will always be behind if it doesn't allow binary drivers. Companies have to make a living.

      --
      The ratio of people to cake is too big
    3. Re:What would be a great "desktop focus" by John+Hurliman · · Score: 4, Insightful

      To provide an example of a superior driver in Linux, my D-Link DWL-650 (Intersil firmware). In Windows, with both the Microsoft and all versions of the D-Link driver I get frequent disconnects, I have to unplug and plug the card in sometimes several times to get a connection in the first place, CPU usage spikes severely when the connection is being made. In Linux I pop the card in, it blinks three times and I have a signal that doesn't drop until I move out of range. How about USB mice? In Windows I patiently wait while the hard drive grinds away and the system tray informs me I've plugged a mouse in, eventually letting me use it. In X11 it works no more than three seconds after plugging it in with no CPU spike. What about those horrible HP all-in-one drivers in Windows that are half driver and half system tray program, and a couple more processes in the system tray that are anyone's guess as to what their purpose is. The drivers have been released and re-released for months and months, every time fixing a nice bug like "prevents USB from randomly disconnecting" or "no longer floods the network with traffic" but yet it never gets to a point that doesn't make you want to throw it out the window. Unless you're running Linux, where CUPS handles it exactly like a printer should be handled, and the scanning is quick and efficient with SANE.

  3. Pointless article by Theovon · · Score: 4, Insightful

    That article was amazingly content-free.

  4. Re:One has to wonder by adrianbaugh · · Score: 4, Insightful

    I don't see why the two are necessarily contradictory. After all, the bits to support enterprise class hardware can easily be omitted from compiling an embedded or desktop platform: if they can make a kernel with modular scheduler and tunable latency (which was the way it seemed to be heading with Con Kolivas' patch set) then the enterprise boys can increase the latency for minimum kernel CPU usage, the desktop people can knock it down for good responsiveness and the embedded folks can plug in an alternative scheduler to suit their own particular needs.

    --
    "'I pass the test,' she said. 'I will diminish, and go into the West, and remain Galadriel.'"
    - JRR Tolkien.
  5. Re:just some SATA support by ender81b · · Score: 4, Insightful

    Which brings up a good point for the 2.7 kernel. You might have better SATA support if they would actually freeze a kernel driver api.

    How about we stop politicizing the kernel and actualy make a stable Driver API? One that doesn't change with every point release of the kernel?

    I know that people want open source drivers but it's extremely hypocritical to complain about companies lack of support for linux then do absoultey *nothing* to help them out by changing the api every point release. Listen, besides some fanatics nobody cares about open source drivers. People would rather their stuff just work.

    I understand that, fundamentally, open source drivers are technically a better solution but there is no chance in hell of convincing Nvidia or any other company that has substantial IP and reserach in their drivers of publishing them open source. Same thing with Intel's Centrino drivers.

    Make a stable api darnit! :)

  6. Re:What I would like to see by Abcd1234 · · Score: 5, Insightful

    Installing and un-installing device drivers would become much easier for users.

    Is insmod so difficult?

    Manufacturers would like this too because then there would be less concern about GPL and device drivers. It would be easier to release binary-only drivers.

    Since when did we care? Linus has flat out said he doesn't like binary drivers, for pretty good reasons, I think (harder to debug being the main one). Why encourage this?

    So, any other good reasons why you'd want userland drivers? Are those reasons good enough to offset the additional overhead that this would incur (additional context switching,etc)? The new layers of indirection that would have to be added?

    Frankly, I think you might have been bitten by the microkernel bug. But, sorry, Linux ain't no microkernel. And, so far, it hasn't needed to be. So, why start now?