Re: [libvirt] [PATCH] conf: Actually make virDomainChrSourceDef an object
On 04/12/2018 04:44 AM, Erik Skultety wrote:
On Thu, Apr 12, 2018 at 09:14:50AM +0200, Michal Privoznik wrote:
> In 2ada9ef1465f we've tried to turn virDomainChrSourceDef into
> virObject. Well, this requires 'virObject' member to be stored on
> the first position of the struct. This adjustment is missing in
> the original commit leading to all sorts of funny memleaks and
> data corruptions.
>
> Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com>
> ---
> src/conf/domain_conf.h | 2 ++
> 1 file changed, 2 insertions(+)
>
Doh! Thanks.... I guess not only do we need a way to detect that we're
using VIR_ALLOC on an object (as Eric pointed out in my original
patches), but we really should detect when we use virObjectNew without
the virObject parent. For some reason, I have a recollection of altering
changes during my meanderings through virObject code in order to point
out more directly some misuses, but it was rejected. Although I cannot
recall if this particular instance of not having virObject parent was
addressed...
Tks -
John
> diff --git a/src/conf/domain_conf.h b/src/conf/domain_conf.h
> index 89a7131fdb..1426f115ed 100644
> --- a/src/conf/domain_conf.h
> +++ b/src/conf/domain_conf.h
> @@ -57,6 +57,7 @@
> # include "virtypedparam.h"
> # include "virsavecookie.h"
> # include "virresctrl.h"
> +# include "virobject.h"
syntax-check fails ^here because the header is already included on line 49
With that fixed:
Reviewed-by: Erik Skultety <eskultet(a)redhat.com>
--
libvir-list mailing list
libvir-list(a)redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list