[mythtv-users] Is NVIDIA worth the bother?

[Bill Williamson]

On Thu, Oct 1, 2009 at 7:30 AM, Eric Sharkey <eric at lisaneric.org> wrote:
> On Mon, Sep 28, 2009 at 9:36 PM, Brad Templeton
> <brad+myth at templetons.com> wrote:
>> It gets stranger when you bring firmware into the mix.   Perhaps I am mistaken,
>> but I have not heard of anybody doing code that will not allow it to run
>> on a system that has "soft" propietary firmware loaded into a card or
>> device.  Ie. your system loads the proprietary firmware into the device and
>> then the driver talks to this device.
>>
>> Now I don't see any difference between that, and a proprietary module that
>> looks like a black box and is talked to by an open source driver in the
>> kernel.
>
> I think the distinction is how the talking to is done.
>
> If the binary module is linked in to the kernel, and running in the
> same memory space as the kernel, and can tweak random bits in memory
> because, hey, it's part of the kernel now, that's a lot different in
> my mind than something that communicates via a socket or some other
> message passing interface to an otherwise unprivileged piece of code.
>
> The question is, is there any possible way for the closed code to
> cause a kernel panic or security breach, assuming that the open code
> is properly written?  If the answer is yes, this makes a lot of people
> nervous.
>


A video card driver writes directly to the memory and registers of the
video card.  There is no way anything with that sort of access is not
exploitable in some way no matter the interface it's given.