Re: [RFC][PATCH] mm: Remove ZERO_SIZE_PTR.

[Pavel Machek <pavel@ucw.cz>]

Hi!

> > > Everybody should check for ptr != NULL, and most callers are actually checking
> > > for ptr != NULL. But nobody is checking for ptr != ZERO_SIZE_PTR.
> > 
> > That is so intentionally because some kernel subsystem can do a zero size
> > allocation.
> > 
> So, not only users *can* do zero size allocation,
> but also there *are* users who are intentionally doing zero size allocation.
> Then, we can't remove ZERO_SIZE_PTR.
> 
> > > Yes, this is the fault of caller. But ZERO_SIZE_PTR is too small value to
> > > distinguish "NULL pointer dereference" and "ZERO_SIZE_PTR dereference" because
> > > address printed by oops message can easily exceed ZERO_SIZE_PTR when
> > > "struct foo" is large.
> > 
> > Correct.
> 
> Maybe PAGE_SIZE / 2 is better than 16?

Actually maybe 1 is better than 16?

It is 'nicer' number, and <= ZERO_SIZE_PTR will actually be correct.

Accessing field it struct will also be easy to differentiate from NULL
pointer -- as most members are word aligned...

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html