You're right that the ability to peer-review driver code makes for better drivers. I totally agree. I also think that designing better OSes is even more effective. I would rather design these kinds of problems out of a system rather than rely on human intervention to catch every single bug. Sort of like how compilers help us find type errors before our code even runs...
This is not a Mac/Windows/Linux/whatever issue. It is an OS architecture issue.
This exploit is yet another reason why drivers should be run in user space. I can't think of a popular OS that does this universally... Linux has nooks, which is not the same thing, and Vista is going to run some, but not all drivers as services instead of in the kernel. Network drivers have traditionally been run in kernel mode for the sake of performance... When is security going to trump performance as a design goal in the major OSes? Enough is enough I say...
Slashdot Mirror
You're right that the ability to peer-review driver code makes for better drivers. I totally agree. I also think that designing better OSes is even more effective. I would rather design these kinds of problems out of a system rather than rely on human intervention to catch every single bug. Sort of like how compilers help us find type errors before our code even runs...
This is not a Mac/Windows/Linux/whatever issue. It is an OS architecture issue.
This exploit is yet another reason why drivers should be run in user space. I can't think of a popular OS that does this universally... Linux has nooks, which is not the same thing, and Vista is going to run some, but not all drivers as services instead of in the kernel. Network drivers have traditionally been run in kernel mode for the sake of performance... When is security going to trump performance as a design goal in the major OSes? Enough is enough I say...