Re: UBSAN: Undefined behaviour in mm/page_alloc.c

[Michal Hocko]

On Fri 09-11-18 19:07:49, Tetsuo Handa wrote:
> On 2018/11/09 18:56, Michal Hocko wrote:
> > Does this following look better?
> 
> Yes.
> 
> >> Also, why not to add BUG_ON(gfp_mask & __GFP_NOFAIL); here?
> > 
> > Because we do not want to blow up the kernel just because of a stupid
> > usage of the allocator. Can you think of an example where it would
> > actually make any sense?
> > 
> > I would argue that such a theoretical abuse would blow up on an
> > unchecked NULL ptr access. Isn't that enough?
> 
> We after all can't avoid blowing up the kernel even if we don't add BUG_ON().
> Stopping with BUG_ON() is saner than NULL pointer dereference messages.

I disagree (strongly to be more explicit). You never know the context
the allocator is called from. We do not want to oops with a random state
(locks heled etc). If the access blows up in the user then be it, the
bug will be clear and to be fixed but BUG_ON on an invalid core kernel
function is just a bad idea. I believe Linus was quite explicit about it
and I fully agree with him.

Besides that this is really off-topic to the issue at hands. Don't you
think?
-- 
Michal Hocko
SUSE Labs