View Single Post
Old 02-23-03, 11:42 AM   #15
Registered User
Join Date: Aug 2002
Location: NC, US
Posts: 55

from the Software License:

2.1.3 Limitations.

No Reverse Engineering. Customer may not reverse engineer, decompile, or disassemble the SOFTWARE, nor attempt in any other manner to obtain the source code.

No Separation of Components. The SOFTWARE is licensed as a single product. Its component parts may not be separated for use on more than one computer, nor otherwise used separately from the other parts.
And yet you can clearly download the src rpm.. so thats lame. screw nVidia, steal their code and FIX it so it WORKS.
The provided source code just provides the "glue" between the nVidia driver and the kernel. The guts of the driver are in a pre-compiled object that gets linked to the portion for which the source is provided. In short, nVidia does NOT provide the source for the portions they want to protect.

I am not a programmer and know little about this, but has anyone tried installing the nForce drivers with an nVidia card in the slot and then swapping the card? From my limited knowledge, an AGP driver should follow AGP protocol and be compatible with ANY AGP CARD.
Probably not. The nVidia driver code--AGP and all--is in a single module. It gets loaded and unloaded as a unit. There is a strong probability that the nVidia driver would prevent another from loading, and vice versa. Even if this were not the case, the nVidia driver is unlikely to do anything if it can't find a video card to work with. Last, the AGP portions of the driver may well have some differences from AGPGART, which could cause problems for code written for AGPGART.

From my perspective, it appears that the primary purpose of nVidia's own AGP implementation is to provide an optional way to support problematic or poorly supported hardware. It is probably written strictly to address compatibility issues between motherboards and nVidia cards, and isn't terribly generic (from a video card perspective). Such an effort is commendable and, in general, nVidia would be perfectly justified in keeping this work fairly closed so that it can derive the benefits of its compatibility work.

The nForce is a special case, however. If the motherboard chipset weren't designed by nVidia, I would not blame nVidia AT ALL for not releasing its custom AGP work. However, since the chipset was made by nVidia, it doesn't slide off so easily in this case. It really needs to provide developers with the necessary information to allow for AGPGART support.

In theory, a ccustom, losed-source module also sounds like a solution, but in practice, it would do littel but create headaches for other vendors. Why should video card vendors have to support a different AGP interface from each manufacturer? It is far better to have a set software interface between the motherboard and the video card than to require video card manufacturers to support multiple interfaces.

This makes it sound like nVidia is opposed to releasing this information, but I don't really know whether this is the case or not. I'm not terribly up on active kernel development, but I have not been able to find anything on this one way or the other. My suspicion is that nVidia has refused to provide the information, but I'd really prefer to know rather than suspect.
etr is offline   Reply With Quote