Re: [PATCH for-next 2/7] RDMA/bnxt_re: Replace chip context structure with pointer

[Devesh Sharma <devesh.sharma@broadcom.com>]

On Mon, Jan 27, 2020 at 1:34 PM Leon Romanovsky <leon@kernel.org> wrote:
>
> On Mon, Jan 27, 2020 at 01:09:46PM +0530, Devesh Sharma wrote:
> > On Sat, Jan 25, 2020 at 11:33 PM Jason Gunthorpe <jgg@mellanox.com> wrote:
> > >
> > > On Fri, Jan 24, 2020 at 12:52:40AM -0500, Devesh Sharma wrote:
> > > >  static void bnxt_re_destroy_chip_ctx(struct bnxt_re_dev *rdev)
> > > >  {
> > > > +     struct bnxt_qplib_chip_ctx *chip_ctx;
> > > > +
> > > > +     if (!rdev->chip_ctx)
> > > > +             return;
> > > > +     chip_ctx = rdev->chip_ctx;
> > > > +     rdev->chip_ctx = NULL;
> > > >       rdev->rcfw.res = NULL;
> > > >       rdev->qplib_res.cctx = NULL;
> > > > +     kfree(chip_ctx);
> > > >  }
> > >
> > > Are you sure this kfree is late enough? I couldn't deduce if it was
> > > really safe to NULL chip_ctx here.
> > With the current design its okay to free this here because
> > bnxt_re_destroy_chip_ctx is indeed the last deallocation performed
> > before ib_device_dealloc() in any exit path. Further, the call to
> > bnxt_re_destroy_chip_ctx is protected by rtnl.
>
> ??? Why does device destroy path need global lock to already existing
> protection from ib_core?
Oh! probably I confused you, What I meant to say was exit path is
executed under netdev-notifier context which already holds the lock.
RoCE driver is not taking that lock explicitly.
>
> Thanks