Re: [PATCH v3 07/11] tmem: Make the uint64_t oid[3] a proper structure: tmem_oid

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Tue, Sep 01, 2015 at 09:37:48AM -0600, Jan Beulich wrote:
> >>> On 01.09.15 at 17:18, <konrad.wilk@oracle.com> wrote:
> > On Mon, Aug 31, 2015 at 12:14:16PM -0400, Konrad Rzeszutek Wilk wrote:
> >> > >And the 'container_of' macro looks to require only one level of
> >> > >nesting.
> >> > 
> >> > I'm pretty sure the macro can deal with both.
> >> 
> >> OK, let me experiement with it as at the first blush it does not work for 
> > me.
> > 
> > I can't get it to work.
> > 
> > My understanding (and the test case - see attached) seem to agree that
> > the 'container_of' is to be used when you have the pointer to the nested
> > member (oid) and want the pointer to the structure in which it is
> > embedded in (op). We have the opposite case - we have the pointer to the
> > structure in which it was embedded (op):
> > 
> >  struct tmem_op *op;
> >   
> >  oidp = (struct tmem_oid *)&op.u.gen.oid[0];
> > 
> > Hence using the container_of macro won't work.
> 
> All of these work and behave the same for me:
> 
> struct tmem_oid *test1(struct tmem_op *op) {
>  return (struct tmem_oid *)&op->u.gen.oid[0];
> }
> 
> struct tmem_oid *test2(struct tmem_op *op) {
>  return container_of(&op->u.gen.oid[0], struct tmem_oid, oid[0]);
> }
> 
> struct tmem_oid *test3(struct tmem_op *op) {
>  return container_of(op->u.gen.oid, struct tmem_oid, oid[0]);
> }

Woot! That will definitly work. Thank you!

> 
> Jan
>