On 11/30/06, M Harris <harrismh777@earthlink.net> wrote:
On Thursday 30 November 2006 13:07, Andre Truter wrote:
I don't follow. What hooks might MS have in the kernel and how would that get in the kernel in the first place? Please be patient with me on this because I'm making it up as I go along... but how about this hypothetical scenario:
Suse gets (permission) to install proprietary drivers (using existing kernel hooks) which provide genuine *interoperability* at the cost of degraded performance. In other words compatible kernel modules could conceivably be introduced (not Linus sanctioned) into the distro (I can change the kernel if I want to, for crying out loud) that may or may not provide *interoperable* functionality, but would provide deliberate performance degradation. [ this might have been done on exiting platforms in the past in order to slow down say, Netscape ]
You are talkin here about a 3rd party kernel module, like for instance the NVidia or ATI drivers. I can write a kernel module that does nothing productive, all it does is it writes a 1 into a file every time the system clock reaches a primary number. But my module (or NVidia's or ATI's) does nothing to the kernel itself, it is not part of the kernel and will never become part of the kernel, so it cannot affect the kernel at all. I cannot gain any form of control over the kernel or it's design or it's future by distributing my module, because it is not part of the kernel. Even if the kernel developers think that my module is very cool and they accept it as an official kernel module that will be distributed with the kernel, then I still have not control over the kernel or any other module. I will have to GPL my module and it will be under the scrutiny of the other kernel developers and if I suddenly decide to change my module to write "MS Rulez" in the file instead of the 1, then my module will most probably be kicked out of the list of supported kernel modules and I would still not have gained anything.
Yes, it would be detected, and some folks might even be sharp enough to patch it and publish their findings, but the damage would be done. If the damage were subtle enough, it might not even be suspected immediately. But even if it is suspected, it would take time and money to search, remove, and/or remedy the damage.
You would have to write a REALLY, REALLY good module that will affect kernel performance (or any performance or negative effect) so that it will be noticed by users and not by the kernel testers. If anything has an affect to such an extent that it would do damage that users will notice, then the kernel developers and testers will have to be totally blind to oversee it before it is accepted as a module. This situation is just not very practical. I think we have a better chance of being hit by an egg that was thrown on one of the planets in the Alpha Centauri system that managed to escape the gravity and atmosphere....
I agree that no distro company would be foolish enough to fork their own kernel version... but its not inconceivable. I want a promise from Novell that the kernel will be controlled by the community outside of the Novell corporate environment... and not under the monetary control of M$.
How can it ever be controlled by MS, or even Novell. No corporation controls the kernel, only the kernel team controls it. Novell can go totally nuts and employ Steve Ballmer as thier new head of kernel development and it would still not make a difference, because the only way that those contributions would be accepted into the kernel is if aliens brainwashed Linus Torvalds and took over his body and got bought out my MS. This whole idea is inconceivable... -- Andre Truter | Software Consultant | Registered Linux user #185282 Jabber: andre.truter@gmail.com | http://www.trusoft.co.za ~ A dinosaur is a salamander designed to Mil Spec ~ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org