Re: [Minios-devel] [UNIKRAFT PATCH] lib/ukalloc: fix wrong return value in uk_malloc_ifpages

[Florian Schmidt <Florian.Schmidt@xxxxxxxxx>]

Hi Simon,

On 08/24/2018 03:55 PM, Simon Kuenzer wrote:
> Hey Florian,
>
> On 24.08.2018 13:23, Florian Schmidt wrote:
> > uk_malloc_ifpages returned the wrong pointer, because it added
> > sizeof(size_t) to a pointer of type size_t. Hence, the return value
> > wasn't offset from intptr by size_t bytes, but by size_t*size_t bytes.
> >
> > This patch makes intptr a void* in places in which its property as
> > size_t* isn't used.
> >
> > Signed-off-by: Florian Schmidt <florian.schmidt@xxxxxxxxx>
> > ---
> >   lib/ukalloc/alloc.c | 8 ++++----
> >   1 file changed, 4 insertions(+), 4 deletions(-)
> >
> > diff --git a/lib/ukalloc/alloc.c b/lib/ukalloc/alloc.c
> > index 52e9a77..2f5e364 100644
> > --- a/lib/ukalloc/alloc.c
> > +++ b/lib/ukalloc/alloc.c
> > @@ -167,7 +167,7 @@ static inline size_t uk_alloc_size_to_order(size_t 
> > size)
> >   void *uk_malloc_ifpages(struct uk_alloc *a, size_t size)
> >   {
> > -    size_t *intptr;
> > +    void *intptr;
> >       size_t order;
> >       size_t realsize = sizeof(order) + size;
> > @@ -181,7 +181,7 @@ void *uk_malloc_ifpages(struct uk_alloc *a, size_t 
> > size)
> >       if (!intptr)
> >           return NULL;
> > -    *intptr = order;
> > +    *(size_t *)intptr = order;
> >       return intptr + sizeof(order);
> Hum... pointer arithmetics on your calculation of your returned value 
> was causing the problem. I somehow do not like that the solution is 
> placed somewhere else in the function to make the return statement correct.
> What if you would have done this on the return?:
>
> return (void *)((uintptr_t)intptr + sizeof(order))
>
> Or you could declare intptr as uintptr_t. I think this is clearer to 
> understand what it going on.
Hmm... I actually want to get away from making this pointer a size_t*, 
because it's really pointing to much more than that. The size_t is just 
the internal information, the whole memory chunk that is passed to the 
caller follows. Hence I would not go back to that. If you feel more 
comfortable making this a uintptr_t than a void* though, I'm fine with that.
I'll send a v2 in a few.

Cheers,
Florian

--
Dr. Florian Schmidt
フローリアン・シュミット
Research Scientist,
Systems and Machine Learning Group
NEC Laboratories Europe
Kurfürsten-Anlage 36, D-69115 Heidelberg
Tel.     +49 (0)6221 4342-265
Fax:     +49 (0)6221 4342-155
e-mail:  florian.schmidt@xxxxxxxxx
============================================================
Registered at Amtsgericht Mannheim, Germany, HRB728558

_______________________________________________
Minios-devel mailing list
Minios-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/minios-devel