Re: [PATCH 16/16] mm/sl[au]b: check if large object is valid in __ksize()

[Marco Elver <elver@google.com>]

On Wed, 13 Jul 2022 at 12:07, Christoph Lameter <cl@gentwo.de> wrote:
>
> On Wed, 13 Jul 2022, Hyeonggon Yoo wrote:
>
> > > Why return 0 if there is an error and why bother the callers with these
> > > checks. BUG()?
> >
> > I thought BUG should be used when there is no other solution.
>
> Spurios returns of 0 that the caller has to check for is a solution?

It's never really been about the caller checking for 0, see below.

> > > I guess this is an error since the order-0 page cannot come from slab
> > > allocations.
> >
> > comment in ksize() says:
> >       "The caller must guarantee that objp points to a valid object
> >       previously allocated with either kmalloc() or kmem_cache_alloc()."
> >
> > It should not be used on order-0 page that is not allocated from slab. No?
>
> I guess we would need to check. Code could exist that does this.
>
> Getting a 0 size would be surprising too here. BUG()? Or WARN() and return
> PAGE_SIZE.

We shouldn't crash, so it should be WARN(), but also returning
PAGE_SIZE is bad. The intuition behind returning 0 is to try and make
the buggy code cause less harm to the rest of the kernel.

From [1]:

> Similarly, if you are able to tell if the passed pointer is not a
> valid object some other way, you can do something better - namely,
> return 0. The intuition here is that the caller has a pointer to an
> invalid object, and wants to use ksize() to determine its size, and
> most likely access all those bytes. Arguably, at that point the kernel
> is already in a degrading state. But we can try to not let things get
> worse by having ksize() return 0, in the hopes that it will stop
> corrupting more memory. It won't work in all cases, but should avoid
> things like "s = ksize(obj); touch_all_bytes(obj, s)" where the size
> bounds the memory accessed corrupting random memory.

[1] https://lore.kernel.org/all/CANpmjNNYt9AG8RrGF0pq2dPbFc=vw2kaOnL2k5+8kfJeEMGuwg@mail.gmail.com/

Thanks,
-- Marco