Re: BUG: mmapfile/writev spurious zero bytes (x86_64/not i386, bisected, reproducable)

[Andi Kleen <andi@firstfloor.org>]

Linus Torvalds wrote:
> 
> On Tue, 17 Jun 2008, Andi Kleen wrote:
>> If that fixes anything:
>> - The caller is broken because it shouldn't pass a faulting source to copy_to_user()
>> - And you broken copy_from_user error reporting which shares the same code
> 
> Andi, I'm sorry I cc'd you. You are the author of that crap, but the bug 
> seems to be that you never even understood what copy_from_user() is 
> supposed to do.
> 
> The whole *and*only* reason for copy_to/from_user() existing AT ALL is 
> exactly the fact that the source or destination access can fault. 

yes, but only one of them (destination for copy_to_user and source for
copy_from_user)

Or are you're describing copy_in_user()?

> I don't really see why you continually start arguing about things that are 
> OBVIOUSLY BUGGY, as if they weren't buggy. Once somebody has debugged a 
> buggy routine, you shouldn't argue against it. 
> 
> So here's a hint: next time I claim some code of yours is buggy, either 
> just acknowledge the bug, or stay silent. You'll look smarter that way.

Ok if I'm really wrong on this (but frankly I don't see the mistake, sorry)
for my person edification: what's a legitimate case for copy_to_user()
where the source can fault?

-Andi