Why NDISwrapper special case?

Why NDISwrapper special case?

Posted Mar 7, 2008 13:24 UTC (Fri) by ldo (subscriber, #40946)
Parent article: NDISwrapper dodges another bullet

It seems to me that, regardless of the legalities or moralities of the case, having the mainline kernel include a special check for a particular out-of-mainline module by name is totally the wrong thing to do. Instead, whatever that check is doing, make it an API routine that a module can call, and persuade the NDISwrapper developers to voluntarily put in a call to it.

Because it seems from some of the other comments here that NDISwrapper is hardly an isolated case. And having to add checks to the kernel for all these cases is clearly an activity that 1) doesn't scale well, and 2) is subject to breakage, either inadvertently or deliberately, as a result of setting off an "arms race". I don't think the kernel developers want to paint themselves into that corner.

---

(Log in to post comments)

The NDISwrapper developers did exactly that.  The module *does* set the taint flag as soon as
it loads an NDIS driver.

As I recall it some whiny RedHat guy pushed the special check in because he was getting bug
reports from kernels using *old* versions of ndiswrapper and he figured it was easier to
change the kernel than to get users to update their perfectly working ndiswrapper versions.