Re: [PATCH for-next 4/7] RDMA/bnxt_re: Refactor net ring allocation function

From: Jason Gunthorpe <jgg@mellanox.com>
To: Devesh Sharma <devesh.sharma@broadcom.com>
Cc: Leon Romanovsky <leon@kernel.org>,
	linux-rdma <linux-rdma@vger.kernel.org>,
	Doug Ledford <dledford@redhat.com>
Subject: Re: [PATCH for-next 4/7] RDMA/bnxt_re: Refactor net ring allocation function
Date: Tue, 28 Jan 2020 00:35:41 +0000	[thread overview]
Message-ID: <20200128003537.GD21192@mellanox.com> (raw)
In-Reply-To: <CANjDDBhxVC0ps8ee5NTW3QrN9bFNVdEcwxS2=Kfn1uOfDR2v_A@mail.gmail.com>

On Mon, Jan 27, 2020 at 01:10:10PM +0530, Devesh Sharma wrote:
> > >  static int bnxt_re_alloc_res(struct bnxt_re_dev *rdev)
> > >  {
> > > +     struct bnxt_qplib_ctx *qplib_ctx;
> > > +     struct bnxt_re_ring_attr rattr;
> > >       int num_vec_created = 0;
> > > -     dma_addr_t *pg_map;
> > >       int rc = 0, i;
> > > -     int pages;
> > >       u8 type;
> > >
> > > +     memset(&rattr, 0, sizeof(rattr));
> >
> > Initialize rattr to zero from the beginning and save call to memset.
> I moved from static initialization to memset due to some sparse/smatch
> warnings, rattr has a "pointer member".

That is why you need to use = {} not the weird '= {0}' version

0 initializes the first member to zero and default initializes the rest
which doesn't work properly if the first member is not an integral
value.

Jason