Re: [PATCH V7 03/16] rust: cpu: Add from_cpu()

[Viresh Kumar]

On 15-01-25, 08:54, Greg KH wrote:
> Ah, but that's not really something that SAFETY should override, right?
> 
> Yes, you know your implementation of this will stop using the pointer in
> the hotplug callback before it goes away but that's not documented here.
> And having the device "fail" afterward isn't really ok either as you are
> relying on the driver core to always check for this and I'm not so sure
> that it always does on all codepaths.
> 
> But, I'm ok with this for now, as you are just copying the bad C model
> at the moment, but it really feels like a huge foot-gun waiting to go
> off.  Any way to put some more documentation here as in "use this at
> your own risk!"?

What about marking it unsafe ? That would require callers to document
why it is safe to call this. And yes add more documentation here too.

-- 
viresh