From mboxrd@z Thu Jan  1 00:00:00 1970
From: bvanassche@acm.org (Bart Van Assche)
Date: Mon, 08 Oct 2018 15:26:42 -0700
Subject: [PATCH 04/16] nvme-core: Complain if nvme_init_identify() fails
In-Reply-To: <20181008221650.GA5979@localhost.localdomain>
References: <20181008212854.68310-1-bvanassche@acm.org>
 <20181008212854.68310-5-bvanassche@acm.org>
 <20181008221650.GA5979@localhost.localdomain>
Message-ID: <1539037602.64374.42.camel@acm.org>

On Mon, 2018-10-08@16:16 -0600, Keith Busch wrote:
> On Mon, Oct 08, 2018@02:28:42PM -0700, Bart Van Assche wrote:
> > This patch avoids that Coverity complains that some but not all callers of
> > nvme_init_identify() check the return value of that function. See also
> > Coverity ID 1423964.
> 
> That seems like an odd thing to complain about. If we required the
> return value be checked, we have the "__must_check" attribute. In this
> particular path, we don't care if nvme_init_identify returns an error,
> and the failure is logged in kernel messages within the function already.
> 
> Is there another way to suppress the complaint?

One way is to change nvme_init_identify(ctrl) into the following:

if (nvme_init_identify(ctrl) < 0) {
}

However, I'm not sure we should do that. Another possibility is to mark the
complaint in the Coverity database as "intended". That means that the return
value is ignored on purpose.

Bart.