From mboxrd@z Thu Jan  1 00:00:00 1970
From: keith.busch@intel.com (Keith Busch)
Date: Tue, 4 Dec 2018 15:37:10 -0700
Subject: [PATCH] nvme-core: small cleanups
In-Reply-To: <20181204222602.GA12719@infradead.org>
References: <20181129171804.5241-1-chaitanya.kulkarni@wdc.com>
 <1543876976.185366.175.camel@acm.org>
 <20181204222602.GA12719@infradead.org>
Message-ID: <20181204223709.GA17048@localhost.localdomain>

On Tue, Dec 04, 2018@02:26:02PM -0800, Christoph Hellwig wrote:
> On Mon, Dec 03, 2018@02:42:56PM -0800, Bart Van Assche wrote:
> > On Thu, 2018-11-29@09:18 -0800, Chaitanya Kulkarni wrote:
> > >  static int nvme_identify_ctrl(struct nvme_ctrl *dev, struct nvme_id_ctrl **id)
> > >  {
> > > -	struct nvme_command c = { };
> > > +	struct nvme_command c;
> > >  	int error;
> > >  
> > > +	memset(&c, 0, sizeof(c));
> > 
> > Please drop this and all similar changes. I think the preferred style in the Linux
> > kernel for structure initialization is to use "= { }" instead of memset().
> 
> I defintively prefer it.  But I don't think changing things either
> way is really worth the effort anyway unless you change the surrounding
> code to start with.

The empty braces initialization is a gcc extension. For standard C
portability, you may use "= { 0 }" ... just in case anyone cares. :)