Re: [PULL 22/27] hw/xen: Add emulated implementation of XenStore operations

[Peter Maydell]

On Wed, 12 Apr 2023 at 19:22, David Woodhouse <dwmw2@infradead.org> wrote:
>
> On Tue, 2023-04-11 at 19:07 +0100, Peter Maydell wrote:
> >
> >
> > > +static void xs_be_unwatch(struct qemu_xs_handle *h, struct
> > > qemu_xs_watch *w)
> > > +{
> > > +    xs_impl_unwatch(h->impl, DOMID_QEMU, w->path, NULL,
> > > xs_be_watch_cb, w);
> >
> > Coverity points out that this is the only call to xs_impl_unwatch()
> > where we don't check the return value. Is there some useful way
> > we can report the error, or is it a "we're closing everything down
> > anyway, no way to report anything" situation? (This particular
> > Coverity heuristic is quite prone to false positives, so if that's
> > the way it is I'll just mark it as a f-p in the coverity UI.)
>
> This is because the Xen libxenstore API doesn't return an error, and
> this is the ops function which emulates that same API. I suppose we
> could explicitly cast to void with a comment to that effect, to avoid
> having it linger in Coverity? I think that's sufficient to make
> Coverity shut up, isn't it?

I've just marked it a false-positive in the UI. Coverity's generally
good at not resurfacing old false-positives, so don't bother
changing the code unless you think it would improve clarity for
a human reader.

thanks
-- PMM