Re: [PATCH -mm] swsusp: userland interface (rev 2)

["Rafael J. Wysocki" <rjw@sisk.pl>]

On Wednesday, 25 January 2006 03:46, Benjamin LaHaise wrote:
> On Wed, Jan 25, 2006 at 12:35:38AM +0100, Rafael J. Wysocki wrote:
> > > > +		if (!access_ok(VERIFY_WRITE, (unsigned long __user *)arg, _IOC_SIZE(cmd))) {
> > > > +			error = -EINVAL;
> > > > +			break;
> > > > +		}
> > > 
> > > Why do we need an access_ok() here?
> > 
> > Because we use __put_user() down the road?
> > 
> > The problem is if the address is wrong we should not try to call
> > alloc_swap_page() at all.  If we did, we wouldn't be able to return the result
> > and we would leak a swap page.
> 
> Then access_ok() is not the droid you are looking for... since it won't 
> catch several cases (out of memory being the most obvious).

Thanks, I haven't thought about it.

> Doing an early put_user() wouldn't hurt and reduces the chance of later failure 
> even further.  __put_user() should never be used outside of a select few 
> performance critical code paths.

Do you mean to use a fake put_user() instead of access_ok()?  And then
put_user() once again or is it reasonable to call __put_user() with the same
arg?

Rafael