Linux kernel APIs

Discussion in 'NZ Computing' started by Lawrence D'Oliveiro, Jul 28, 2007.

  1. In message <f8egke$p9r$>, RL wrote:

    > The attitude that closed-source
    > drivers are evil, and nothing is to be fully supported unless it is also
    > GPL'd and in the main source tree is the problem. If companies could
    > produce binary-only drivers, that worked reliably from one kernel
    > release to the next, then support would be greatly improved IMO.


    That would simply put you back in the Windows situation, with chronic driver
    instability being responsible for so many crashes.

    This sort of thing has been discussed before. Greg Kroah-Hartman likes to
    mention the example of USB: the Windows USB driver APIs were reworked about
    3 times (with Vista probably counting as a fourth). The ones in Linux were
    also reworked 3 times. The difference is that Windows still has to carry
    around the obsolete backward-compatible versions and support them forever,
    because there are old, obsolete, binary-only drivers that continue to rely
    on them. Linux doesn't need to carry around obsolete baggage in the kernel
    this way: when the APIs were reworked, all the drivers could also be
    reworked at the same time to get rid of the old calls.

    So you see, the Linux kernel is actually less bloated than the Windows one.
     
    Lawrence D'Oliveiro, Jul 28, 2007
    #1
    1. Advertising

  2. Lawrence D'Oliveiro

    peterwn Guest

    Lawrence D'Oliveiro wrote:
    > In message <f8egke$p9r$>, RL wrote:
    >
    >> The attitude that closed-source
    >> drivers are evil, and nothing is to be fully supported unless it is also
    >> GPL'd and in the main source tree is the problem. If companies could
    >> produce binary-only drivers, that worked reliably from one kernel
    >> release to the next, then support would be greatly improved IMO.

    >
    > That would simply put you back in the Windows situation, with chronic driver
    > instability being responsible for so many crashes.
    >


    There is another aspect - security. Paranoid users (eg SIS, GCSB, etc)
    would not want binary 'blobs' in kernels or associated modules intended
    for top secret applications.

    By the way I have no idea what these outfits actually use - this would
    be classified information in itself.
     
    peterwn, Jul 28, 2007
    #2
    1. Advertising

  3. In message <46aaf38a$>, peterwn wrote:

    > Lawrence D'Oliveiro wrote:
    >
    >> In message <f8egke$p9r$>, RL wrote:
    >>
    >>> The attitude that closed-source
    >>> drivers are evil, and nothing is to be fully supported unless it is also
    >>> GPL'd and in the main source tree is the problem. If companies could
    >>> produce binary-only drivers, that worked reliably from one kernel
    >>> release to the next, then support would be greatly improved IMO.

    >>
    >> That would simply put you back in the Windows situation, with chronic
    >> driver instability being responsible for so many crashes.

    >
    > There is another aspect - security. Paranoid users (eg SIS, GCSB, etc)
    > would not want binary 'blobs' in kernels or associated modules intended
    > for top secret applications.


    Unfortunately, that kind of thing is already present, e.g. in the madwifi
    drivers.
     
    Lawrence D'Oliveiro, Jul 29, 2007
    #3
  4. In message <f8ehih$qqj$>, I wrote:

    > In message <f8egke$p9r$>, RL wrote:
    >
    >> The attitude that closed-source
    >> drivers are evil, and nothing is to be fully supported unless it is also
    >> GPL'd and in the main source tree is the problem. If companies could
    >> produce binary-only drivers, that worked reliably from one kernel
    >> release to the next, then support would be greatly improved IMO.

    >
    > That would simply put you back in the Windows situation, with chronic
    > driver instability being responsible for so many crashes.


    And also, for example, these sorts of issues
    <http://theinquirer.net/default.aspx?article=41440>.

    The only way to root out crap drivers is to expose the source code to
    ruthless peer-review scrutiny.
     
    Lawrence D'Oliveiro, Aug 4, 2007
    #4
    1. Advertising

Want to reply to this thread or ask your own question?

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. Replies:
    0
    Views:
    502
  2. Ron Martell

    Linux Kernel Fragmented IPv6 Packet Filtering Bypass

    Ron Martell, Nov 7, 2006, in forum: Computer Support
    Replies:
    18
    Views:
    702
    The Ghost In The Machine
    Nov 9, 2006
  3. Adam Warner

    2.4+ Linux Kernel Exploit

    Adam Warner, Dec 1, 2003, in forum: NZ Computing
    Replies:
    10
    Views:
    625
    Adam Warner
    Dec 2, 2003
  4. SW

    Looking for EAPOL APIs

    SW, May 6, 2008, in forum: Wireless Networking
    Replies:
    1
    Views:
    472
    James McIllece [MS]
    May 8, 2008
  5. amsc

    Bluetooth HFP and Windows APIs

    amsc, Mar 10, 2010, in forum: Software
    Replies:
    0
    Views:
    1,798
Loading...

Share This Page