On Mon, Jul 23, 2012 at 2:55 PM, Laszlo Ersek <ler...@redhat.com> wrote: > On 07/20/12 14:01, Stefan Hajnoczi wrote: > >> diff --git a/net.c b/net.c >> index e7a8d81..f2752c7 100644 >> --- a/net.c >> +++ b/net.c >> @@ -25,6 +25,7 @@ >> >> #include "config-host.h" >> >> +#include "net/hub.h" >> #include "net/tap.h" >> #include "net/socket.h" >> #include "net/dump.h" > > This includes "net/hub.h" for the second time in "net.c" (still true at > the end of the series). Causes no problems, of course.
Thanks, will remove. >> @@ -876,17 +879,17 @@ static int net_client_init1(const void *object, int >> is_netdev, Error **errp) >> } >> >> if (net_client_init_fun[opts->kind]) { >> - VLANState *vlan = NULL; >> + VLANClientState *peer = NULL; >> >> /* Do not add to a vlan if it's a -netdev or a nic with a netdev= >> * parameter. */ >> if (!is_netdev && >> (opts->kind != NET_CLIENT_OPTIONS_KIND_NIC || >> !opts->nic->has_netdev)) { >> - vlan = qemu_find_vlan(u.net->has_vlan ? u.net->vlan : 0, true); >> + peer = net_hub_add_port(u.net->has_vlan ? u.net->vlan : 0, >> NULL); >> } > > "u.net->vlan" is int32_t ('int32' in the schema > <http://lists.nongnu.org/archive/html/qemu-devel/2012-06/msg01675.html>); > net_hub_add_port() > expects unsigned. Will have to decide how to handle the vlan id input. Ideally the option itself would be type-checked in enough detail (uint32_t, int32_t, etc) so that we can just use the value here. I'll figure out which way to resolve this best. >> diff --git a/net/dump.h b/net/dump.h >> index 0fa2dd7..a1beb0d 100644 >> --- a/net/dump.h >> +++ b/net/dump.h >> @@ -28,6 +28,6 @@ >> #include "qapi-types.h" >> >> int net_init_dump(const NetClientOptions *opts, const char *name, >> - VLANState *vlan); >> + VLANClientState *vlan); >> >> #endif /* QEMU_NET_DUMP_H */ > > Should be called "peer", but it will be renamed later on. Same for > "net/slirp.h". I'll push the renames as early as possible in the series. Thanks. Stefan