On Tue, Oct 7, 2014 at 4:35 PM, Nithin Raju <nit...@vmware.com> wrote:
> The init function must be dummy for the unit tests to pass. The
> unit tests don't require the kernel datpaath to be loaded, and
> without the kernel datapath, any calls netlink calls will fail.
>
> Signed-off-by: Nithin Raju <nit...@vmware.com>
> Reported-by: Gurucharan Shetty <gshe...@nicira.com>
Wouldn't it be better to simply not set .init for NETDEV_WINDOWS_CLASS
in this case?
Couple of trivial comments below.
> ---
> lib/netdev-windows.c | 13 ++++++++++++-
> 1 files changed, 12 insertions(+), 1 deletions(-)
>
> diff --git a/lib/netdev-windows.c b/lib/netdev-windows.c
> index 0518156..56a26ae 100644
> --- a/lib/netdev-windows.c
> +++ b/lib/netdev-windows.c
> @@ -78,6 +78,7 @@ static int query_netdev(const char *devname,
> struct netdev_windows_netdev_info *reply,
> struct ofpbuf **bufp);
> static int netdev_windows_init(void);
> +static int netdev_windows_init_(void);
There is a comment below which needs correction.
>
> /* Generic Netlink family numbers for OVS.
> *
> @@ -102,6 +103,16 @@ netdev_windows_cast(const struct netdev *netdev_)
> static int
> netdev_windows_init(void)
> {
> + /* This must be dummy for the unit tests to pass. The unit tests don't
> + * require the kernel datpaath to be loaded, and without the kernel
> + * datapath, any calls netlink calls will fail.
s/datpaath/datapath
s/any calls/any/
> + */
> + return 0;
> +}
> +
> +static
> +netdev_windows_init_(void)
> +{
> int error = 0;
> static struct ovsthread_once once = OVSTHREAD_ONCE_INITIALIZER;
>
> @@ -250,7 +261,7 @@ query_netdev(const char *devname,
> ovs_assert(info != NULL);
> netdev_windows_info_init(info);
>
> - error = netdev_windows_init();
> + error = netdev_windows_init_();
> if (error) {
> if (info) {
> *bufp = NULL;
> --
> 1.7.4.1
>
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
