On Thu, Mar 31, 2022 at 09:37:42AM +0100, Daniel P. Berrangé wrote:
On Wed, Mar 30, 2022 at 10:39:09AM -0700, Andrea Bolognani wrote:
> On Wed, Mar 30, 2022 at 06:13:58PM +0100, Daniel P. Berrangé wrote:
> > On Wed, Mar 30, 2022 at 09:59:11AM -0700, Andrea Bolognani wrote:
> > > + if (virQEMUCapsGuestIsNative(hostarch, guestarch)) {
> > > + if ((ret = virFindFileInPath("/usr/libexec/qemu-kvm"))
!= NULL) {
> > > + return ret;
> > > + }
> > > + }
> > > +
> > > return ret;
> > > }
> >
> > That overwrites the existing 'ret' value that we want to keep
> > when qemu-kvm isn't present. Needs
> >
> > if (!ret && virQEMUCapsGuestIsNative(hostarch, guestarch)) {
> > ...
>
> Does it? All previous attempts immediately return ret if it's not
> NULL, so if we get to this point we know that there's no existing
> value to preserve.
Honestly I didn't look at the earlier context, just assumed that
'return ret' was meaningful. If 'ret' is always NULL, then we
should be writing 'return NULL' rather than 'return ret' to
make it clear we've not expected any value here.
Totally agree, but it should be a separate patch. Let me just prepare
and post it right now :)
--
Andrea Bolognani / Red Hat / Virtualization