Re: PCI driver module unload race?

[Roman Zippel]

Hi,

On Tue, 11 Mar 2003, Patrick Mochel wrote:

> > > Greg, and Rusty, are right. Dealing with this is a PITA, and I think will 
> > > always be. I'm willing to take the Nancy Reagan platform, too. 
> > 
> > Right with what? 
> 
> With the idea that unloading modules is a bad idea. 

With the current module refcount model I can only agree.
OTOH I need a sufficiently complex example, which gets the module locking 
right (file systems are just too simple), then I can actually produce a 
patch, which shows the advantages of a different model and the driver 
model/sysfs looks like an interesting victim. :)

> > What is the "Nancy Reagan platform"?
> 
> "Just say no". It was a big anti-drug campaign in the US targeted at
> schoolchildren, spearheaded in the mid-80's by Nancy Reagan.

Well, this always look simple, but it hardly ever solves the real problem.

bye, Roman