Re: [dpdk-stable] [dpdk-dev] [PATCH] malloc: fix memory element size in case of padding

[Thomas Monjalon <thomas@monjalon.net>]

26/11/2019 14:45, Thomas Monjalon:
> 26/11/2019 14:39, Xueming(Steven) Li:
> > 
> > > -----Original Message-----
> > > From: Thomas Monjalon <thomas@monjalon.net>
> > > Sent: Tuesday, November 26, 2019 9:30 PM
> > > To: Burakov, Anatoly <anatoly.burakov@intel.com>
> > > Cc: Xueming(Steven) Li <xuemingl@mellanox.com>; dev@dpdk.org; Asaf
> > > Penso <asafp@mellanox.com>; stable@dpdk.org;
> > > david.marchand@redhat.com
> > > Subject: Re: [dpdk-dev] [PATCH] malloc: fix memory element size in case of
> > > padding
> > > 
> > > 26/11/2019 13:57, Burakov, Anatoly:
> > > > On 25-Nov-19 11:24 PM, Thomas Monjalon wrote:
> > > > > 21/11/2019 16:14, Burakov, Anatoly:
> > > > >> On 21-Nov-19 2:25 PM, Xueming Li wrote:
> > > > >>> This patch fixes wrong inner memory element size when joining two
> > > > >>> elements.
> > > > >>>
> > > > >>> Fixes: af75078fece3 ("first public release")
> > > > >>> Cc: stable@dpdk.org
> > > > >>>
> > > > >>> Signed-off-by: Xueming Li <xuemingl@mellanox.com>
> > > > >>> ---
> > > > >>> --- a/lib/librte_eal/common/malloc_elem.c
> > > > >>> +++ b/lib/librte_eal/common/malloc_elem.c
> > > > >>> @@ -487,6 +487,10 @@ join_elem(struct malloc_elem *elem1, struct
> > > malloc_elem *elem2)
> > > > >>>    	else
> > > > >>>    		elem1->heap->last = elem1;
> > > > >>>    	elem1->next = next;
> > > > >>> +	if (elem1->pad) {
> > > > >>> +		struct malloc_elem *inner = RTE_PTR_ADD(elem1, elem1-
> > > >pad);
> > > > >>> +		inner->size = elem1->size - elem1->pad;
> > > > >>> +	}
> > > > >>>    }
> > > > >>
> > > > >> Reviewed-by: Anatoly Burakov <anatoly.burakov@intel.com>
> > > > >
> > > > > I don't understand this patch.
> > > > > The variable inner is never used.
> > > > > What am I missing?
> > > > >
> > > >
> > > > For padded elements, malloc element has two headers - the "outer"
> > > > header with empty space after it, and the "inner" header, after which
> > > > the user memory actually starts. This makes it so that, when joining
> > > > elements, if the outer element had a pad, we also update the inner
> > > element size to match.
> > > 
> > > Where the variable "inner" is used in this function?
> > > 
> > Rte_realloc, inner size is used to copy data.
> 
> I still don't get it. Am I missing half of the patch?
> Please give explicit line number.

OK after a quick chat, I understood my miss:
inner is a pointer helping to reach the size field.
Sorry for the noise.