Re: [PATCH 2/5] virnetlink: Provide stub for virNetlinkBridgeVlanFilterSet()
On 5/13/25 09:53, Peter Krempa wrote:
On Mon, May 12, 2025 at 15:37:13 +0200, Michal Privoznik via Devel
wrote:
> From: Michal Privoznik <mprivozn(a)redhat.com>
>
> In virnetlink.c there are two sections: the first one when
> building WITH_LIBNL support, the other that provides stubs for
> functions declared in the corresponding header file when building
> without netlink support. But the stub implementation for
> virNetlinkBridgeVlanFilterSet() was missing.
>
> Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com>
> ---
> src/util/virnetlink.c | 11 +++++++++++
> 1 file changed, 11 insertions(+)
>
> diff --git a/src/util/virnetlink.c b/src/util/virnetlink.c
> index 206646d9d7..2555457cd2 100644
> --- a/src/util/virnetlink.c
> +++ b/src/util/virnetlink.c
> @@ -1344,6 +1344,17 @@ virNetlinkNewLink(const char *ifname G_GNUC_UNUSED,
> }
>
>
> +int
> +virNetlinkBridgeVlanFilterSet(const char *ifname G_GNUC_UNUSED,
> + int cmd G_GNUC_UNUSED,
> + const unsigned short fflags G_GNUC_UNUSED,
s/fflags/flags/
No, this needs to be anything else but 'flags' otherwise our
sc_flags_usage syntax-check rule will complain. unusedflags perhaps?
> + const short vid G_GNUC_UNUSED,
> + int *error G_GNUC_UNUSED)
> +{
Based on the usage I thin this function needs to still set error to 0.
All callers for now do that before calling but still the documentation
states that non-zero code means a netlink error, which is not the case
here.
Good point.
> + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _(unsupported));
I wanted to complain that INTERNAL_ERROR is not appropriate; but it
seems that all the stubs in virnetlink.c have terrible errors.
Indeed, I went with consistency over correctness.
> + return -1;
> +}
> +
> int
> virNetlinkGetNeighbor(void **nlData G_GNUC_UNUSED,
> uint32_t src_pid G_GNUC_UNUSED,
> --
> 2.49.0
>
Reviewed-by: Peter Krempa <pkrempa(a)redhat.com>
Michal