[Libvir] Remote & the problem with qemu deadlock

[Daniel Veillard]

On Wed, Apr 18, 2007 at 04:33:58PM +0100, Richard W.M. Jones wrote:
> Daniel Veillard wrote:
> >  Can we recognize that the connection is on localhost and just serve
> >the request directly and synchronously instead ?
> 
> Yes you can.  In real terms what this would mean is that all the 
> functions in qemu_internal.c are changed so that they become:
> 
> static int qemuDestroyDomain(virDomainPtr domain) {
>     qemud_packet_client req;
>     qemud_packet_server reply;
>     qemuPrivatePtr priv = (qemuPrivatePtr) domain->conn->privateData;
> 
>     req.data.type = QEMUD_CLIENT_PKT_DOMAIN_DESTROY;
>     req.data.qemud_packet_client_data_u.domainDestroyRequest.id = 
> domain->id;
> 
> #if QEMU_INTERNAL_IS_REALLY_LINKED_TO_THE_SERVER
>     /* some code to get the fake server & client, and then: */
>     qemudDispatchDomainDestroy (server, client, req, rep);
>     /* followed by some code to decode the fake "reply packet" */
> #else /* existing code: */
>     if (qemuProcessRequest(domain->conn, priv->qemud_fd, NULL, &req, 
> &reply) < 0) {
>         return -1;
>     }
> 
>     return 0;
> #endif
> }
> 
> So it can be done but qemu_internal is sure going to look ugly afterwards.

  Maybe you can centralize the dispatcher one level up and keep it like
a driver, this could avoid the issue, no ?

Daniel

-- 
Red Hat Virtualization group http://redhat.com/virtualization/
Daniel Veillard      | virtualization library  http://libvirt.org/
veillard at redhat.com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine  http://rpmfind.net/