On Mon, 9 Jul 2007, Fabrice Bellard wrote:

> No. Ideally you should use the same conventions as the Linux kernel and 
> assume that you cannot access the user data directly.

That's what I had already started doing today.

> For the time being, I would suggest to minimize the number of changes and 
> just extend lock_user()/unlock_user() as you began to do to handle -EFAULT. 
> The rest is mostly a question of cosmetics.

The attached patch is my in-progress work of the complete overhaul to use
the kernel conventions. It needs some more work to finish the
conversion, but enough should be done to see how it is going to turn
out. Overall, I think the converted code is easier to read, especially
if you are familiar with reading kernel code. I also think it will end
up being more correct both becasue of the additional time now spent on
reviewing each syscall, as well as the kernel conventions tend to make
you be more thorough and explicite.


                                 Stuart

Stuart R. Anderson                               anderson@netsweng.com
Network & Software Engineering                   http://www.netsweng.com/
1024D/37A79149:                                  0791 D3B8 9A4C 2CDC A31F
                                                  BD03 0A62 E534 37A7 9149